It’s the end of the season and the World Cup is in full swing, so that makes right now the ideal time to talk about transfers, right?

Like it or not, I’ll be dishing up some servings of piping-hot rankings to supplement your World Cup fix over the next few weeks, and the first step in my rankings has always been midfielders, beginning at the back.

There are some reports that Arsenal could target someone who can play the 6, so this list could prove handy as the summer goes on!

As always, these ranks are purely by the numbers and should not be used as a replacement for good, old-fashioned scouting. They also do not necessarily reflect my own views of these players. Now, let’s get it!

Read more

I am usually starting this process as the final whistle blows in the final game of the season, but this season was different. With the celebrations around winning the Premier League and playing in the Champions League final, it was a nice change of pace to soak up the joy instead of instantly closing the book on a season. I have also been working hard on building out the CannonStats app, which I am incredibly excited about. I have started a beta test, and if you are a premium subscriber, more information on full access will be coming out in the coming weeks.

Now it does truly feel like it is transfer season and I am ready to dive in.

For this I am going to group things by position and give my quick thoughts on the group.

Goalkeeper

David Raya has gone from a player that a large portion of the fan base was skeptical about to being in the conversation for the best keeper in the world. He turned in another big season for Arsenal, making some crucial saves on the way to the title while controlling his box and starting off attacks. I don’t think there is any rush to try and move away from him at all given he is in the heart of his prime.

The players behind him are fine and less of a concern. I don’t know enough about the youth at Arsenal to have strong opinions on any of them. My opinion on Kepa Arrizabalaga is largely unchanged from the start of the season. He is a totally fine backup. You wouldn’t choose to start him in a game that REALLY mattered to Arsenal and he is clearly a downgrade, but given that he is a low cost player he is about as good as you can get for the full package. He could leave for a fee that basically matches what Arsenal got him for and that is fine, but if he stays and backs up Raya again, that is fine with me too.

Full Backs

Arsenal have spent quite a bit on full backs over the years and they have built out a very talented bunch of players who can play in these positions. I think this summer you will see at least one player come in here, and it wouldn’t shock me if it was two (though it could be a nominal center back capable of playing wide).

On the left you have the incredibly talented Riccardo Calafiori and Piero Hincapié. I think the late season switch to midfield for Myles Lewis-Skelly is also a major signal that, outside of an emergency situation, his time playing full back is coming to an end. I would be happy and content with this group going into next season, but given the injury track record of Calafiori, who has missed 238 days and 44 matches to injury with Arsenal, I don’t think needing a third player is crazy here.

On the right hand side Arsenal are on paper in great shape with Jurriën Timber, Ben White, plus emergency coverage from Cristhian Mosquera. In real life, I think this is the first realistic departure and the area where I am pretty highly confident there will be an incoming.

The need for an incoming, and probably the outgoing, comes down to the injury situation and not the talent. Timber ended the season banged up and that has now also taken him out of the World Cup. White dealt with injuries all season and they seemed to hit at just the wrong times, so that it was not practical or ideal to effectively share the load at right back.

This has been the curse of the right back position. Whoever was supposed to provide cover got injured, and the primary player had to push through their own fitness troubles, leading to injury down the road.

Hopefully that can get resolved this summer.

Center Backs

Arsenal have two of the best center backs in the world. They have a number of quite good alternatives that can cover for them, as long as they aren’t covering for an injured full back.

Overall this is a strong position for the quality and depth. The big part here will be getting the full back injury situation under control, and this should be fine.

I don’t anticipate any incoming nor outgoing here in a strict sense, but the way Arsenal favor versatile players capable of covering both central and wide defending means that whatever happens at full back will help here.

Central Midfield

Arsenal’s midfield is a bit fluid right now. The team is moving away from the more strict 4-3-3 shape into something a bit harder to describe, and it is getting more and more position-less with players granted lots of freedom to interchange.

I generally view the situation as one where Arsenal start with two central midfielders and one more attacking midfielder, and this will focus more on the former positions.

The depth here is solid, with Declan Rice, Martín Zubimendi, Myles Lewis-Skelly, and Christian Nørgaard all looking to take two positions. If you want to get a bit broader you could even throw Mikel Merino and Martin Ødegaard into this mix, even if I see them fitting better into the more attacking midfield role.

Lewis-Skelly making a late season switch from full back into a more central role potentially changes the calculus of what is required to bolster this group in the summer. That move may have saved Arsenal from spending upwards of 50-60 million to bring someone new in.

I think you could still see a central midfielder added to the group, especially if Nørgaard wants more playing time as his career winds down. I would love to see another younger talent added, but honestly if this was the group that Arsenal went into next season with it would be quite strong.

Attacking Midfield

Arsenal have added to this group in both of the last summers and it looks quite full. The options here in my depth chart would be Martin Ødegaard, Eberechi Eze, Mikel Merino, Kai Havertz, and Ethan Nwaneri.

You have a number of different ways of playing the position from these players, and depending on what you do with the other midfield position you can play two of them together.

Given the multi-faceted skill set of these players they also provide cover in other positions.

The only realistic potential exit from this group, and it could be another loan, is Nwaneri. If he views his position in the middle of the park long term he needs to play there to develop, and that will be hard to do at Arsenal.

There is an Ødegaard question here given his contract situation, and that is a topic for a longer and specific article. My TLDR is that I don’t think any sale gives you enough in return and you would have to spend a lot to replace him, so stay the course on the status quo.

Wide Attack

I am lumping all of these players together rather than splitting left and right.

The right side has Bukayo Saka, Noni Madueke, and Max Dowman. That is strong, and actually a quite young group that looks set for the next few seasons. I don’t think there is any need to add on here, nor for any of the players to leave.

The left side is much more uncertain, with Leandro Trossard and Gabriel Martinelli still the main guys in this group. You could put Eze here, but he ended up playing less than 20% of his minutes from the left hand side, so it would take a fairly drastic change overall.

Both of the players on the left could end up leaving, but realistically, and maybe for continuity’s sake, it will only be one of them. That would necessitate an addition (or two), and realistically in the squad this is the area where you could see a new player actually breaking into the primary XI rather than more of a squad addition.

If and when I try my hand at the director of football exercise, this is where the bulk of the action will be taking place.

Striker

Arsenal spent pretty big on a striker last summer in a market where it was tough to pick a perfect option. Viktor Gyökeres isn’t everyone’s favorite, and I think he ultimately did a passable job and met my (admittedly muted) expectations for him in his first season.

I still view Kai Havertz as the primary choice here, especially for big games, but he is a player with his own question marks around his injury history.

I don’t anticipate that either of these players leaves this summer, but I can’t say the same for Gabriel Jesus. He is in the final year of his contract, and that might make him more amenable to trying to figure out a place where he can play in the longer term, even if that means taking a pay cut to do it. I don’t have super high hopes on the sort of fee he could bring in, but maybe I’ll end up pleasantly surprised by it.

I don’t see it as the highest priority, but maybe there is room here to add a versatile young attacker to fill the spot left by Jesus, especially if Havertz is viewed as having coverage in a number of positions by Arteta.

The Big Picture

Arsenal just won the Premier League and made it to the Champions League final. They were a consistently great team all season long, with just some minor blips largely driven by key players missing time.

Given that, I think the squad needs tweaking and more surgical changes rather than a major overhaul. I would be shocked if we saw a summer as busy as 2025, where Arsenal added 8 first team players. I think it will be more like 3-4 depending on the players that end up leaving.

I like to play around with depth charts, and this is where I see the current situation before Arsenal start adding or subtracting players this summer.

It should be a fun one, and I can't wait to examine all the new players and rumors as they happen.

The new Similar Player Search in Cannon Stats turns that instinct into something you can actually query: pick a player, and get a ranked list of the players whose statistical profile looks most like theirs.

It sounds simple and you will have seen other ways to go about trying to accomplish this same goal. Underneath the clean list of names sit two genuinely hard questions that I spent a long time trying to get right, and that are worth pulling apart because they shape every result you'll ever see: how do you measure "similar," and which stats, plus how many, do you use to decide? This post walks through both decisions going over the methods and their differences, and the metrics I chose and why.

The core idea: comparing players on the same ruler

You can’t compare a striker’s 0.6 non‑penalty xG per 90 to a center back’s 6.1 clearance and expect to learn anything from the raw numbers; they’re measuring different action with different units on different scales.

So the first thing this model does is standardize everything so that it is all on the same scale. For each stat, we look at the whole population of players in that role and express each player’s stat as a z‑score. A z-score is just a fancy way of saying how many standard deviations above or below the average they are. A z‑score of +2 in shot volume would mean “elite,” a ‑1 means “below average,” and now every stat speaks the same language and gives the same meaning for the number used.

Doing this for all of the stats turns each player into a vector, basically their stats become a string of ~16-18 numbers describing where they sit on every meaningful trait for their position/role.

This turns “Similar” into a geometry problem and the goal is to figure out which other players’ vectors point in the same direction, have the same volume, or sit closest in that space. The different methods are just different definitions of “closest.”

The methods: four ways to define "closest"

Ultimately I landed on giving four different ranking methods, and they help answer subtly different questions about how similar a player is to another.

Cosine compares the shape of two players’ profiles, not the size. This is the default method and after testing it out, it seems to be the nice happy medium position for answering the “which player is similar?” question. It measures the angle between the two vectors and ignores their length.

In football terms: a high‑volume, do‑everything winger and a lower‑volume player with the exact same balance of strengths will register as very similar, because they’re “the same kind of player” even if one simply does more of everything. When we ran it on Alexis Sánchez’s 2015‑16 season, cosine returned Arjen Robben, Neymar, Nolito, Lorenzo Insigne, Franck Ribéry, Lionel Messi, Mohamed Salah and Raheem Sterling, exactly the cohort of elite, high‑impact attacking wingers that would make sense if you’re familiar with Alexis.

PCA Cosine is cosine’s more sophisticated cousin. Before measuring the angle, it runs the profiles through Principal Component Analysis, which compresses the stat set down to a handful of independent “directions of variation” and throws away the noise. The effect is a de-noised ranking, it stops highly correlated metrics from getting double counted. On Sánchez it returned a near‑identical but slightly cleaner list (Neymar, Robben, Mario Götze, Érik Lamela, Salah, Ribéry, Gareth Bale), swapping a couple of borderline names for better ones. As we will go over in a bit, talking about the stats chosen, the list for each role has already been curated to avoid the kitchen sink method and the cleanup is mild but it could be useful in certain situations.

Euclidean distance is the most intuitive and the least forgiving method. It measures the absolute straight‑line distance between the two points in that space, which means it cares about magnitude as well as shape. Two players have to be similar in both what they do and how much, this is the tool if you want to find players at a genuinely comparable level and volume, not just a comparable style. It’s useful when you want “same player, same tier,” but it’s more sensitive to one stat being far off, and it degrades faster as you add more metrics (more on that below).

Category profile zooms out from individual stats to the seven Cannon Stats skill categories: shooting, creating, passing, carrying, receiving, defending, aerial. Instead of matching on the individual stats it matches on the shape across those seven buckets: a player who is 90th‑percentile shooting / 40th‑percentile passing / 20th‑percentile defending will be matched to others with that same shape. It’s a higher‑level, more forgiving view — good for “who has the same broad strengths and weaknesses,” less precise about the details.

We also built and tested Mahalanobis distance, which is the theoretically “correct” way to handle the fact that football stats are heavily correlated. On paper it should be the best of the methods but in practice, on our data, it ended up just being a bit too noisy. For Sánchez it surfaced James Milner, Gonzalo Castro and Markel Susaeta, workmanlike players who could share similarity if you squint but weren’t really what we were after.

The harder question: which stats get a vote?

In my database, I have well over 300 different stats that I could use to try and build out a similarity model and that makes getting this part especially hard. My initial attempts at this threw in a ton of stats following the intuition that more stats is more better.

After running some tests, and a lot of weird results with similar players coming back looking very funky, I learned that the opposite is true and that there is an optimal level.

Two problems sink the "throw everything in" approach. The first is the curse of dimensionality: as you pile on metrics, the distances between players flatten out — everyone drifts toward being "moderately similar to everyone," and the top of your list gets noisy and unstable.

The second, and bigger, is multicollinearity: football stats are massively correlated with each other. When I measured it for wingers, deep touches and box touches correlated at 0.97, aerial duels won and aerial duels attempted at 0.94, carry threat (xT) and box‑entry carries at 0.91, expected assists and deep completions at 0.90. When you start feeding this into the model, you haven't added information, instead all you've done is told it that "being involved in attacking areas" matters four or five times as much as everything else, just because four or five stats all measure roughly the same thing in a slightly different way.

I started with a list of roughly 50 stats that I had narrowed down from the initial 300 and picked out the key ones from different families to try still capture the skill but to keep things balanced. In the end I was able to create role specific lists of around 18 stats, with no pairs of stats at more than a 0.8 correlation.

How did we know roughly 16 was the right number and not 10 or 40? Two checks. First, we ran PCA on the full set and looked at how many independent dimensions actually exist in winger play: 8 components explained 80% of the variation, 13 explained 90%, at 17 we hit 95%. In other words there are really only ~10–15 distinct "things" a winger can be good or bad at, beyond that and there is a lot of repetition.

Second, we ran a stability test: take the target player's top‑10 most‑similar list using 10, 20 and all metrics and measure how much it changes. Going from 10 to 20 metrics shifted about a third of the names, suggesting that at 10 we are under‑specified. But going from 20 to the full set barely moved it (82% overlap), so we didn’t get much change for the added under of metrics.

This pointed to a nice target around 15 metrics and that is where I landed. The lists are also position/role specific. The metrics for a striker are purposefully different than a center back. I have made it so that you could look at a player on different profiles and see how that changes the players come back as similar.

Using it

For this I have made this as flexible as possible:

• you can set the position/role

• you can set the season(s) and competition(s) for the player you want to find the similar player to

• you can set the season(s) for where you want to search against

• you can limit by age for who you want to compare against

• you can limit by position for the players to compare against

For example, here are the U23 players most similar to Bukayo Saka last season on the winger profile.

No similarity model is the truth. It’s a lens, and it inherits the limits of its inputs.

A model only knows what’s in the stat set, so it can’t (and will never) see things those stats don’t capture (a player’s temperament, their fit in a specific system, the quality of their league, etc).

But used the way it’s meant to be this should be a fun and useful too. Who doesn’t want a fast, rigorous starting point for answering the question “who plays like this, and where do I find more of them?” I am looking forward to playing around with it more and once the web app has passed a bit more validation hopefully you all will as well.

It’s been more than a week since we last laid our eyes upon top-tier, competitive football, and boy does it feel like that’s just way too long to go without it.

Thankfully, the World Cup kicks off this Thursday, less than two weeks since PSG lifted the Champions League trophy, giving us more than 105 total matches to soothe our need for the beautiful game! Who needs hamstrings, anyway?

Last World Cup feels like a lifetime ago, but it gifted us names like Enzo Fernández, the rising Argentine starlet who would move to Chelsea for €121 million a couple of months later. Fellow Benfica legend Gonçalo Ramos also showed out in Qatar and earned himself a €65 million move to PSG in the next summer, and yes that’s worked out just fine, no further questions.

Similarly, many Arsenal fans watched in awe as Nico Williams dazzled in Spain’s uptempo, attacking setup, while others admired Azzedine Ounahi’s ballcarrying prowess for Morocco.

As you can see, World Cup starlets can be a little hit or miss, as international success doesn’t always translate to the club game. But as long ago as Pele in the 1958 World Cup, this tournament has been flagging young players worth keeping an eye on. With that in mind, here are a sampling of the young names I’ll be paying close attention to in the USA/Mexico/Canada World Cup of 2026:

Read more

After Arsenal lost back to back matches against Manchester City and Bournemouth, when Manchester City briefly had the title in their full control, it was a dark time. I will admit even though I have been extremely confident about Arsenal this season I wrestled with the doubts.

I ultimately still felt that Arsenal were best positioned to win the title. Even though the odds had slipped considerably; this team, this organization, this coaching staff, still had it in them to win out and end the long wait for a Premier League. That is ultimately how things played out and Arsenal ended the title drought. I am incredibly thankful for that and it has brought me, and the whole Arsenal fan base immense joy.

One of the big questions that we don’t have to contemplate is what would we have done had Arsenal not won the title.

Because I am myself, I still want to think about that and what exactly the process would have been had the happy path not occurred.

This is tough given that we know the good thing happened and Arsenal won the title. I know for certain that the buildup to the Champions League match would not have felt as easy going and the match would have taken on existential stakes. The shootout would have somehow been even more unbearable and the pain even more acute.

I have tried to be fair and consistent along the way here. If Arsenal didn’t win anything I think that it would have been fair to ask questions about the manager and his position. That is a major reason that I want to go through this now, I want to try and ask myself those questions and try and think through what would have been the best path for this club.

I have long said that Trophies are the goal of the club and what we want as fans but they are very poor measures of how well the team is performing. I stand by that and I think that if the criteria for keeping or firing a coach relies solely on if a team wins silverware or not you are going to end up making more mistakes than good decisions. There is so much randomness and tiny margins that can decide things that are outside of the coaches control.

I know that is the way of the world, but that doesn’t mean that it has to be the way that I analyze things.

I think the biggest things that I am looking for to make these calls are as follows:

1. Is the team performing below their talent level?

2. Have the players lost belief and they are not giving their all?

3. Is the manager losing his ability to adapt and coming up with new ideas?

4. Has the coach been able to get the most out of the players that he has at his disposal?

With regards to Mikel Arteta I think that for the most part, even if Arsenal had failed to get over the line the answers to these would have still pointed towards keeping him. There may have been a growing negative swell inside some of the fan base but they would have been wrong and the data would have backed that up.

Read more

If you are a premium subscriber, you can listen to an ad-free version through Substack or you can add to your preferred podcast app through the personal RSS feed.

If you are a regular subscriber the episode is available to be added by the RSS Feed Here (copy the link) to the podcast player of your choice or click here to listen to the episode directly in your browser, through apple, or below through Spotify:

The format is simple: 20 teams, bottom to top, against the points totals from the preseason betting markets.

The headline result is that both hosts finished ahead of the coin flip, Adam at 13 of 20 and Scott at 12, which feels respectable until you sit with the ones that got away. Here is how the table looked for the over/unders, our predictions, plus the actual points won for each team.

With this we will start to turn our attention away from 2025/26 and into the summer transfer season.

Read more

It’s the go-to stat when people want to talk about finishing quality. Player X scored 15 goals from 10 xG, so they must be a great finisher. Player Y scored 8 from 12 xG, so they’re clearly wasteful. Case closed.

Except it’s not.

G-xG is one of the noisiest stats in football. A player can over-perform xG by 5 goals one season and under-perform by 3 the next, with no change in actual ability. Small samples, random variance, and the binary nature of goals (you either score or you don’t) make raw single season G-xG deeply unreliable as a measure of finishing skill.

Just to illustrate this a bit more, here is how season to season the values correlate with each other for the Premier League 2017/18 through 2025/26.

If this was a good measure we would expect to see the numbers be bunched around the 45 degree line that runs through the chart, instead the red regression line is nearly flat and shows that the season to season relationship is extremely weak.

Here is what the correlation looks like for all of the big five leagues.

The trend holds and at the high level, just using goal - expected goals to try and identify finishing skill is tough.

A smarter approach: Bayesian finishing skill

The Cannon Stats finishing model uses Bayesian hierarchical logistic regression to estimate each player’s true finishing ability. This approach was guided by previous work down by Martin Eastwood on his blog pena.lt/y and Marek Kwiatkowski with his work at Statsbomb.

Bayesian hierarchical logistic regression is a mouthful, so here’s what it actually means in plain terms.

The model starts from the assumption that a player is an average finisher and as evidence (shots/goals) starts coming in, the prior belief about the finishing skill is updated up or down.

The key difference from simple G-xG is that the model accounts for sample size and uses this to help us get an idea of how certain we should be about what has happened.

A player with 30 shots gets a wide range of uncertainty, because the model knows it doesn’t have enough evidence yet. The model learns about the overall population variance and uses that on the player level pulling players with smaller sample back towards average. A player with 4 goals from those 30 shots could be elite, but the model stays skeptical and keeps their estimated finishing skill around 0. It takes sustained over or under performance to move the estimates away from average.

The final output is a finishing skill value. Positive means the player converts chances at a higher rate than xG predicts. Negative means lower. Zero is perfectly average. Alongside the estimate, the model produces a 90% confidence interval: the range where the player’s true finishing ability most likely falls and helping give us a ballpark of just how confident we should be in that we are seeing true skill vs luck.

What it looks like in practice

Here is where the rubber meets the road. To do a sanity check here are the top 15 players on finishing skill.

It is always reassuring when you see Lionel Messi top an attacking metric and this is no different. Overall the list is filled with players that I think would be thought of as good ball strikers and finishers.

I have taken an extra step to build out a bit of a dashboard as well to be able to present the data here.

The graphic has three key parts.

1. At the top, a gauge shows where the player sits on the spectrum from poor to elite. For this I have made average cover the area from +/- 1 standard deviation, Above average is from 1 to 2 standard deviations, and Elite is 2 or more standard deviations above the mean. The reverse is true on the negative side

2. Below on the left, a population distribution shows how they rank against all 5,000+ players in the database with 30 or more shots. It is a very nice normal distribution type shape.

3. On the right, two charts break down their finishing with feet versus their head. The bars on the body part charts represent the confidence interval for where the “true” talent lies. A tight bar means the model is confident in its estimate. A wide bar means more data is needed before drawing conclusions.

To interpret this the finishing skill estimate can be thought of as the relative change for that player from the baseline xG. A player that is an elite finisher will be +0.20 and that would turn a 10% chance into a 12%, that’s not a ginormous absolute change but that translates into roughly 2 goals over 100 shots (roughly a season’s worth for a striker).

To give a bit more Arsenal flavor, here are the three main center forwards for Arsenal and they give us a nice overall view of what good, average, and poor finishing skills would look like.

Why this matters

Finishing skill is one of the hardest things to measure in football. Goals are rare events, xG is imperfect, and randomness plays a huge role.

A model that accounts for uncertainty and sample size gives you a much better picture than raw G-xG ever can.

For scouting and transfer analysis, the confidence interval is the key output. A target with a strong finishing skill and a tight interval is a reliable bet. A target with a strong GxG but a wide interval might just be on a hot streak. The model separates the two, and helps us identify the signal in a nosy random world, giving us a much stronger view of what is actually player skill.

If you are a premium subscriber, you can listen to an ad-free version through Substack or you can add to your preferred podcast app through the personal RSS feed.

If you are a regular subscriber the episode is available to be added by the RSS Feed Here (copy the link) to the podcast player of your choice or click here to listen to the episode directly in your browser, through apple, or below through Spotify:

The episode opens 24 hours on, with both hosts landing somewhere closer to proud than devastated. Adam keeps coming back to Gabriel, who skied the decisive penalty and deserved that ending less than almost anyone, and to the sense that a shootout loss tells you very little about a team. Scott frames it against the alternative: this was not Inter getting blown out in last year’s final. Arsenal were in this one. It played out almost exactly as he predicted, a tight game decided by a handful of moments, right on the knife edge, and this time the edge fell PSG’s way.

The match breakdown is the heart of the episode, and both hosts land on the same read. Arsenal made a conscious choice to cede possession, sit in a medium-low block, and use PSG’s strengths against them. It worked. Scott notes that outside the penalty, PSG managed only four chances above ten percent xG, and through 90 minutes the non-penalty xG was roughly 0.8 to 0.6 in Arsenal’s favor, with the two biggest chances of the game both falling to Havertz. The argument both hosts make is that the 72-28 possession split and lopsided shot count are not evidence of a problem, they are evidence of a plan. PSG had no central access, no zone 14 touches, and visibly ran out of ideas against a team built to deny exactly that. As Adam puts it, there is a difference between not threatening because you cannot and not threatening because you are not trying to, and that distinction is the whole game. Doing this to one of the two or three best attacks in the world is an accomplishment, not a failure.

From there the conversation turns to the officiating, which both hosts felt was unusually prominent. The referee twice blew his whistle before Arsenal could take a corner, a pointed and demonstrative call you almost never see outside a blowout, and the run of ticky-tack fouls and Saka’s yellow consistently punished whichever side was out of possession, which by design was Arsenal. On the two major decisions, they agree the Mosquera penalty was correct if soft, and that the Madueke no-call on Nuno Mendes in extra time was a far bigger miss. Scott makes the point he keeps returning to: a no-call is a call, it shaped the game in PSG’s favor just as surely as a whistle would have shaped it in Arsenal’s, and being afraid to make the call is the real failing.

The substitutions get a full airing. The Timber-for-Mosquera change at the hour was the easy one to back. Bringing Odegaard off for Gyokeres made sense for transitional speed and tired legs, but Adam thinks pulling that much technical quality out of midfield hurt Arsenal’s ability to build, a pattern they have seen before this season. The bigger sticking point is Saka. Both hosts understand the fitness logic given his calf, but Saka is the moments player, the one you want available for the four chances a game like this offers and for the shootout itself. The counterpoint is that Enrique pulled Dembele, Kvaratskhelia and Vitinha too, so both teams reached penalties without their best takers.

On the shootout, the hosts dig into the margins, from Safonov reading the scouting report to sit middle for Rice, to Eze’s hesitation-filled, twisted run-up that produced a poor strike, to the general debate over the stutter-step approach and whether it is worth the loss of balance unless you take penalties constantly like Lewandowski or Jorginho. PSG won both coin tosses, and in the end they got just enough bounces to make the game their won.

They close on the long view. It is almost 20 years to the day since the 2006 final loss, and the parallels are there, but the key difference is that 2006 marked the end of a chapter while this team is squarely in its prime. Over the next three to five years, this is a side that should be among the best in Europe with more cracks at the trophy to come. Lewis-Skelly, not yet 20, and a 24-year-old Saka anchor a group with time on its side. PSG will likely be back too, and neither host would mind another go at them.

Read more

There are really two seasons in top-level pro football — the season where you actually play the games, and then transfer season. And every transfer season, rival fanbases flex their respective muscles, convinced that they’ve got the league’s next great new player, while the guy their rival club signed is totally mid.

So for which signings, to borrow a phrase, did it all work out? I booted up ADAM 2.0 to check, let’s go through them below and take stock one year/four months later!

Read more

The 2025-2026 season is over (save for one game…I forget who’s playing in that one) and the stats are final! That of course means it’s everyone’s favorite time of year: Time to look back and say who did the best.

With the ADAM formula in its 2.0 era and working more seamlessly than ever, I decided to race to my desk and spit out the ratings for each and every player across my usual nine domestic leagues and Europe’s top two continental competitions all in one big swoop. This gave me a team of the season for each league, picked purely by robots and with zero editorialization by yours truly.

Now, ADAM is just a formula. ADAM does what the numbers tell it to. So, do I agree with all of the below? Absolutely not! But it’s at least fun to see who gets in and who doesn’t.

Before we do: I don’t have any sort of advanced data or advanced knowledge of the goalkeeper position, so just mentally put David Raya on all of these. OK? OK.

Oh wait - we’re also gonna pick a 4-3-3 in every league. Don’t like it? Tough.

EFL Championship

In England’s second tier, a clear trend emerged very early — Middlesbrough had a lot of guys with great stats. That makes the elephant in the room … their inability to win promotion, especially over a poor side like Hull City, all the more awkward.

• RB: Callum Brittain, Middlesbrough

• LB: Matt Targett, Middlesbrough

• CB: Luke Ayling, Middlesbrough

• CB: Tristan Crama, Millwall

• DM: Aidan Morris, Middlesbrough

• CM: Hayden Hackney, Middlesbrough

• AM: Finn Azaz, Southampton

• LW: Femi Azeez, Millwall

• RW: Ephron Mason-Clark, Coventry City

• ST: Tommy Conway, Middlesbrough

I love that we’ve got an Azeez and an Azaz, and also that Azaz began the season at … Middlesbrough.

Snubs? Probably Matt Grimes, CM extraordinaire for the champs, Coventry. Taylor Harwood-Bellis and Leo Scienza of Soton. Zan Vipotnik scored 23 goals for Swansea but finished a few spots from the top, partly because five of his goals were from the spot. Gustavo Hamer racked up 12 assists for Sheffield United but couldn’t top Azaz or Scienza on all-around score.

Belgian Pro League

Club Brugge ended up winning the league, and you can see why in the team of the season:

• RB: Anan Khalaili, Union St. Gilloise

• LB: Joaquin Seys, Club Brugge

• CB: Kevin Mac Allister, USG

• CB: Brandon Mechele, Club Brugge

• DM: Bryan Heynen, Genk

• CM: Hans Vanaken, Club Brugge

• AM: Thorgan Hazard, Anderlecht

• LW: Christos Tzolis, Club Brugge

• RW: Carlos Forbs, Club Brugge

• ST: Joseph Opoku, Zulte Waregem

I went Opoku - a 20-year-old Ghanaian who scored 10 goals and added six assists as part of a two-man front - over other more traditional 9s like Nicolo Tresoldi or Wilfried Kanga. Why? Because I can. Tresoldi in particular was only just off the pace for the number 9 spot.

Snubs? I feel pretty good here, but USG’s Christian Burgess would be a good pick at CB. Ryotaro Ito from Sint-Truidense VV was close in midfield, and Nilson Angulo from Anderlecht nearly made it despite joining Sunderland mid-season.

Eredivisie

• RB: Bart van Rooij, FC Twente

• LB: Mats Rots, FC Twente

• CB: Jerdy Schouten, PSV

• CB: Tsuyoshi Watanabe, Feyenoord

• DM: Kodai Sano, NEC Nijmegen

• CM: Joey Veerman, PSV

• AM: Ismael Saibari, PSV

• LW: Mika Godts, Ajax

• RW: Jacob Trenskow, SC Heerenveen

• ST: Troy Parrott, AZ Alkmaar

Snubs? The biggest one has to be Ayase Ueda from Feyenoord, who amassed 25 goals but couldn’t edge Troy Parrott, who created more xG per touch, shot a little more effectively and was also a bit more creative. Anis Hadj Moussa from Feyenoord was the most creative player in the league and was the first man to miss out on a winger spot. FC Utrecht’s Souffian El Karouani was also very creative from the left back position and was a first man out.

Primeira Liga

• RB: Alberto Costa, FC Porto

• LB: Maximiliano Araújo, Sporting

• CB: Gonçalo Inácio, Sporting

• CB: Jakub Kiwior, FC Porto

• DM: Morten Hjulmand, Sporting

• CM: Victor Froholdt, FC Porto

• AM: Trincão, Sporting

• LW: Andreas Schjelderup, Benfica

• RW: Sorriso, Famalicão

• ST: Luis Javier Suárez, Sporting

No surprise to see Porto and Sporting - first and second, respectively - so well represented here, including our good friend Maxi Araújo from Sporting. Victor Froholdt, just 20, is a fun shout here, as is Sorriso, an unknown from fifth-place Famalicão who finished eighth in expected assists across the league.

Snubs? Benfica’s Vangelis Pavlidis was great and just missed out on the striker spot, while a host of defenders at the big three — Jan Bednarek, Nicolas Otamendi, Antonio Silva — also missed the cut. But look at our boy Jakub! You go, Jakub!

Europa League

• RB: Ainsley Maitland-Niles, Lyon

• LB: Maximilian Mittelstädt, Stuttgart

• CB: Gustaf Lagerbielke, Braga

• CB: Matthias Ginter, Freiburg

• DM: Bryan Heynen, Genk

• CM: Elliot Anderson, Nottingham Forest

• AM: Petar Stanic, Ludogorets

• LW: Williot Swedberg, Celta Vigo

• RW: Antony, Real Betis

• ST: Deniz Undav, Stuttgart

Interesting to see the champions not represented at all here, but with such few games to analyze that kind of thing can happen. Arsenal legend Ainsley Maitland-Niles is there, though!

Snubs? Stuttgart play a kind of striker/second-striker system where Undav is the latter, but he was great in Europe, can play 9, and finished high, so I put him in that role over Igor Jesus, who also ranked high. Villa’s biggest snubs are probably Ezri Konsa, Emi Buendia and Morgan Rogers. Other guys who missed the cut include Braga’s Ricardo Horta, Betis’s Adbe Ezzalzouli and Vincenzo Grifo from Freiburg, all close. Freiburg’s Johan Manzambi, a great prospect I’m high on, was also right there for Anderson’s spot.

Serie A

• RB: Alexis Saelemaekers, AC Milan

• LB: Federico Dimarco, Inter Milan

• CB: Alessandro Bastoni, Inter

• CB: Pierre Kalulu, Juventus

• DM: Manuel Locatelli, Juventus

• CM: Nicolò Barella, Inter

• AM: Charles De Ketelaere, Atalanta

• LW: Kenan Yildiz, Juventus

• RW: Armand Laurienté, Sassuolo

• ST: Lautaro Martínez, Inter

ADAM clearly has no issue with Lautaro being named striker of the season, nor with Dimarco being league MVP. Yildiz could arguably go into a couple of roles, but I went with left wing here. Also, no Saelemaekers isn’t a right back, but when guys play in a back five, I have to get creative!

Snubs? I feel like the bases are covered quite well here, but my guys I wish I had a bit more room for would be the Real Madrid-bound Nico Paz (Como), Donyell Malen (Roma), ageless wonder Luka Modric (AC Milan) and Nicolo Fagioli, who was a bright spot in a bad, bad Fiorentina side.

Ligue 1

• RB: Aron Dønnum, Toulouse

• LB: Matthieu Udol, Lens

• CB: Lucas Beraldo, PSG

• CB: Malang Sarr, Lens

• DM: Vitinha, PSG

• CM: Adrien Thomasson, Lens

• AM: Florian Thauvin, Lens

• LW: Matias Fernandez-Pardo, Lille

• RW: Bradley Barcola, PSG

• ST: Estéban Lepaul, Rennes

That’s right, only three from PSG, which happens when your stars barely play league games! I made good money this year betting on Lepaul to score, which he did 21 times.

Snubs? An unnamed former Man United player did qualify for a winger spot, but who cares? Joaquin Panichelli was great for Strasbourg and barely missed the 9 role. Ilan Kebbal deserves some special mention for Paris FC, as does (my opinion) Lille’s Aissa Mandi in the CB spots.

La Liga

• RB: Andrei Ratiu, Rayo Vallecano

• LB: Carlos Romero, Espanyol

• CB: Eric Garcia, Barcelona

• CB: Dean Huijsen, Real Madrid

• DM: Aurelien Tchouameni, Real Madrid

• CM: Pedri, Barcelona

• AM: Pablo Fornals, Real Betis

• LW: Vinícius Júnior, Real Madrid

• RW: Lamine Yamal, Barcelona

• ST: Kylian Mbappé, Real Madrid

Snubs? Would you believe me if I told you Nicolas Pepe is the first winger to miss the cut, followed by Antony? Fede Valverde may be a so-so captain, but he was right there for this one, as were Fermín López and Arda Güler.

Bundesliga

• RB: Konrad Laimer, Bayern

• LB: Maximilian Mittelstädt, Stuttgart

• CB: Nico Schlotterbeck, Dortmund

• CB: Edmond Tapsoba, Bayer Leverkusen

• DM: Angelo Stiller, Stuttgart

• CM: Joshua Kimmich, Bayern

• AM: Romano Schmid, Werder Bremen

• LW: Luis Diaz, Bayern

• RW: Michael Olise, Bayern

• ST: Harry Kane, Bayern

Snubs? The electric Yan Diomande is right in the running for a winger spot here, but just missed. Deniz Undav had so many goals in league play (19), but didn’t feel suitable for an AM spot and can’t displace Kane. Aleix Garcia had a strong season, as did David Raum and Alex Grimaldo at left back and Julian Ryerson at right back. Old pal Vladimir Coufal also had a strong season in helping TSG Hoffenheim qualify for Europe!

Champions League

• RB: Achraf Hakimi, PSG

• LB: Nuno Mendes, PSG

• CB: Virgil van Dijk, Liverpool

• CB: Alessandro Bastoni, Inter

• DM: Vitinha, PSG

• CM: Dominik Szoboszlai, Liverpool

• AM: Kenan Yildiz, Juventus

• LW: Vinícius Júnior, Real Madrid

• RW: Luis Díaz, Bayern

• ST: Kylian Mbappé, Real Madrid

This is just what the formula spit out, OK?! Don’t hate me. I’m not concerning myself here with putting players on their correct sides in attack, and Díaz has been great in the CL. Saka has played about 70% of the minutes of the Bayern man, a big disadvantage here. Gabriel Martinelli actually sits seventh among wingers, Arsenal’s best representative there, despite also playing somewhere between half and two-thirds as many minutes as Díaz.

Snubs? Big names at striker who didn’t top Mbappé include Julián Álvarez, Harry Kane, Erling Haaland and … Viktor Gyökeres!? That’s right, the Arsenal man sits fifth on the striker ranking, behind some heavy hitters.

Gabriel ranks fifth at CB, Timber is sixth at FB (MLS and White both in the top 10), Eze is sixth at attacking midfielder, Rice is fifth at CM. Just very good/solid across the board but lots of guys who’ve played less in the comp than their counterparts!

Premier League

The big one…

• RB: Jurrien Timber, Arsenal

• LB: Nico O’Reilly, Man City

• CB: Marc Guehi, Palace/City

• CB: Gabriel, Arsenal

• DM: Casemiro, Man Utd

• CM: Declan Rice, Arsenal

• AM: Bruno Fernandes, Man Utd

• LW: Antoine Semenyo, Man City

• RW: Bukayo Saka, Arsenal

• ST: Erling Haaland, Man City

A lot to react to here. No getting around Bruno as the 10, even if the gap wasn’t as big as you’d expect between him and Cherki, especially.

City defenders and deeper midfielders had an advantage on Arsenal’s in a number of counting stats, and that can be good or bad depending on the tape. But Marc Guehi had stellar progressive numbers especially, whereas Arsenal don’t progress as much via their CBs, which will knock you in any formula setting. Also, kind of ironic to make Casemiro (who scored a bajillion goals) the 6 and Rice the 8, but it is what it is.

Last: How funny to me that, in a season where he was derided for not doing enough (and playing 900 fewer minutes than Semenyo), Saka finishes second in winger rating in the entire damn league. Maybe I was right when I said over and over again that he actually wasn’t playing poorly?

Snubs? Bruno Guimarães quietly had a really good season and was close, as were others like Dominik Szoboszlai (whose position would be really tough to place), Elliot Anderson and Enzo Fernández. Cherki is obviously a big shout, as is Igor Thiago. Ollie Watkins and João Pedro were actually a bit closer to Haaland at striker, though. Arsenal bias aside I’d say Doku, Jarrod Bowen, Jan-Paul van Hecke and Adrien Truffert are other names to think about. We can get into the rest of the Arsenal near-misses in my rankings pieces to come.

I am still finishing up the final collection and cleaning process but I should be fully moved over to all of the new data by the start of transfer season and with the advances in coding assistants we might even be getting some new webapps (currently in a limited beta but planned for a wider release as we get into the summer) that take this advanced data and the different graphics that I have put together over the years.

More information as this develops, in the meantime here are the new items we will have access to.

The “Shooting Zone”

For the attacking on ball events, I have simplified them into 3 different categories, they are pass, shoot, or duel. I then looked at the probability of each of these for the zones across the pitch.

The main purpose here was 1) to satisfy my curiosity, 2) try to refine the area that is especially valuable for shooting.

We track penalty area touches and that is good but not all touches in the penalty area are in a good location to shoot from. That would be the same with my “deep touches” which is a semi-circle 25 yards from the center of the goal. There are still helpful things to understand and show a player in a dangerous location, but it is less tailored towards answering the question of how often does a player or team get touches in good shooting locations.

This shows the probability of an on-ball event being a shot from each zone, with what I am calling the “Shooting Zone” highlighted.

I took a bit of creative license here when defining the boundaries, favoring zones that are more central but overall, I wanted to capture the areas where a shot was very likely to be attempted. I set the threshold at roughly 40% and gave a bit of leeway to the zones just around the “D” because these were close to that level and they were in my central area that I care about.

This is a bit of a throwback reference, but it is basically the inverse of the “Andros Townsend Force Field”.

For this area, I will start looking at pure counting stats for touches, passes completed, and passes received into this zone, plus how well a player/team converts these touches into shots/xG/Goals.

I think that this can give light on how a team plays as well, looking at how they get into these locations, is it pullbacks, through balls, carrying, central penetration/combo play, crosses?

Goal probability added

This isn’t new, but if I am going through and talking about things it is always good to explain this a bit more. Goal probability added is my version of a possession value model, it is in the same family with metrics like VAEP, g+, or on-ball value and should be relatively close to these.

The idea here is looking at the net change from each action on a team’s chances of scoring or conceding a goal in the current sequence plus the next sequence.

Because I am a simple man, I have stuck with the tried-and-true zone-based system. This can have issues with thresholds, for example a pass or carry inside of a zone entirely will show no change but something that moves just across the boundary that is nearly indistinguishable in every other way will be treated differently.

Ultimately, I just decided I am okay with that and that in the aggregate they will likely washout. I have also tried to minimize the occurrence by adding more zones, increasing the resolution of the pitch, for this the field is broken into 400 zones (25 by 16), that translates into roughly 4.2 meters by 4.25 meters rectangles.

This is how the probability of scoring looks across the pitch.

If you have seen the value of the different areas of the field before this shouldn’t surprise you at all. As you get further from your own goal, and as you get more central, the chances of scoring increase.

This is how the probability of conceding looks.

This one is a little more interesting and it is probably something that you would have been less likely to have seen before. It matches intuition in that the closer to the goal you are defending, the more likely you are to concede but it isn’t nearly as centered on the goal. The half spaces are actually a bit more likely to lead to the other team scoring than central locations.

Expected Threat

This is another stat that if you are familiar with more advanced metrics, you will be familiar with. I figured I was updating how I collected and processed my data, it would make a lot of sense to add this as well given the popularity and spread of this metric.

Expected threat shares quite a lot of similarities to goal probability added, it estimates the probability that the team on the ball will score in the next five actions from each zone, but it ignores the oppositions chances of scoring.

This is how the probability of scoring looks across the pitch from an xT perspective.

You may also notice that this uses a slightly larger zone than my GPA metric, diving the pitch into 192 zones in a 16 by 12 layout.

You’ll see this metric coming into my graphics as well.

Splitting out Shot Assisted xG (xAG) and xA

This was something that was done by FBref, that I always liked and I am adopting this to my new way of collecting data.

What this change does is treat and credit the value of a key pass differently.

Shot assisted xG, has always been a better name for what xA measures but my view on names of things don’t always win out. This is the classic way of crediting a player for a key pass, where they make a pass that leads to a shot and it is simply the expected value of that shot.

Often times this works out just fine but there are occasions where the player that receives the pass does a ton of work turning a low probability of scoring chance into a high probability of scoring chance and crediting the player making the pass would give the wrong impression.

Similarly, a player can make a great pass to set up a good shooting opportunity and for whatever reason a shot never happens, and the passer ends up with no credit whatsoever.

One of the remedies for this is to only look at where the pass was completed and credit for the probability of scoring from there. This also removes the dependency on a player needing to convert that pass into a shot.

This will be tracked with the following:

xAG - Value of the shot credited to the passer

xA - Probability of a completed pass being converted into a goal

Shot Creating Actions

This is another metric that was created by FBref that I found was quite valuable and wanted to be able to look at in more detail in my own data.

A shot creating action is one of the two successful attacking actions that lead up to a shot being taken. This is an expansion on the key pass idea and includes the ability to create your own shot.

This would include live-ball passes, set-play passes, successful dribbles, shots which lead to another shot, and winning a free kick (getting fouled) or a corner.

One of the other items that I have expanded on here as well, is including the expected goal value of the shots created. Like shot assisted xG this can give credit where maybe it isn’t deserved fully but it does also differentiate that not all shots are worth the same. We are always making tradeoffs here and I think it is nice to have an idea of the value of the actions.

Here is a visual illustration and a video to drive home where the different ways that actions that lead to a shot get credit.

This is Arsenal’s 2nd goal against Bayern Munich in the 2025-26 Champions League group stage match from November 2025.

Here is how it looks in the event data. It starts with a Declan Rice ball recovery (there will be more on the different ways that recoveries, blocked passes, and interceptions work later), he carries the ball forward before he lays it off to Eberechi Eze. Eze then plays in Riccardo Calafiori making a nice underlapping run. Calafiori plays a nice out swinging left footed cross that finds Noni Madueke at the back post to finish first time to put Arsenal ahead 2-1.

There are a number of ways that this ends up getting valued and credited to the different players let’s break it down.

We are talking about shot creating actions, and for this one we have two. The first is credited to Eze for making the pre-assist, the second goes to Calafiori for making the pass that assisted the shot. Each player would get 1 SCA, they would also have an xGCA (xG Creating Action) of 0.86 because this is a HUGE chance.

Calafiori would also get credit for 0.86, xG Assisted, with a 0.16 xA for completing a cross into this location. He would also get credit for 0.86 on xG Chain but would not get any credit for xG Buildup because he only had a hand in one of the final actions. On GPA, Calafiori gets a bunch of credit with his cross adding +12%.

Eze gets his SCA, but he also gets credit for xG Buildup and xG Chain, plus his pass here does also add in 0.01 xA. On GPA, Eze gets +1% for his pass completion to Calafiori.

Rice gets a bit of the short end of the stick here for credit. His action to intercept the ball probably does the second most to contribute to the goal, and he ends up only getting credit for xG Buildup and xG Chain with no key pass or shot creating action on the scoresheet. His pass technically gets 0.001 xA as well but that’s not going to really move the needle here. Here is where GPA helps him out, his recovery interception is worth +4% and his carry plus pass adds another +0.5%.

Madueke gets credit for the goal (obviously) but would also get the xG Chain added to his tally. On GPA, he gets credit for getting on the end of the pass worth +9% and converting this position on the field into a shot is worth +25%, the execution of the shot here actually brings it down a bit -4% but overall, he gets a ton of credit here at +30%.

Updated xPass Completion Model

My xPass metric has been helpful for me and it helped to differentiate that not all passes have the same difficulty.

This helped me to push out something like pass efficiency, trying to get an idea of a player’s passing skill by comparing the rate at which they completed passes compared to what an average player would have done.

I am not getting rid of this metric, but rather I am updating it by doing a more advanced way of estimating the probability of a pass being completed.

The previous version was zone based, looking at the starting location of the pass and how often passes from that zone to the end zone were completed. This works quite well but I think it is possible to do better than this.

For the new metric I have switched to eXtreme Gradient Boosting model that takes the different features to predict if a pass will be completed or not. It is actually two models and I take the average of the two, because I was a bit worried about over fitting including the distance metric

The features that I have chosen fall into a few different categories.

Descriptions of where on the field that the pass happens and where it is being aimed:

• pass start distance from center

• pass start distance from end line

• pass start distance to goal

• pass angle

The model that includes pass distance also includes these:

• progressive distance (measured from the center of the goal that the team is attacking)

• pass lateral distance

• pass horizontal distance

• pass end distance from center

• pass end distance from end line

The situation for the pass:

• regular play pass

• kick off

• goal kick

• free kick

• corner kick

• direct speed of play in the sequence

• the closest the team has been to goal in the last 5 actions

Descriptions for how the pass was played

• pass from feet on the ground

• pass from feet chipped into the air

• headed pass

• a contested pass from an aerial duel

• a through ball, cross, switch, layoff, pullback, flicked on

Game state for the pass

• Score difference

• Player advantage/disadvantage

• Minute played

Updated xG model

This is one of the items that I was most excited about, and it was something that I had wanted to do for a long time.

My old model has served me well but my skills and the tools that I have available now have opened up the opportunity to improve upon it. One of the big additions that I was able to do was also switch to an event dataset that includes more robust information that I didn’t have previously. The major addition here is the coordinates of the keeper when the shot is taken, plus a measure for pressure that the shooter is under and the overall clarity towards goal.

Knowing when there is an open goal to aim at is a pretty big deal and that is one of the areas where because my model didn’t have that input, it would infer that from other information and that generally worked well in the aggregate, but it could really mess up individual chances.

This is a pretty big one and should help make the overall values more accurate.

I am continuing to add more and more but I wanted to give some updates on where things are going.

It feels like the hype, pressure, and anticipation for this match are much lower than I would have thought going into the biggest single match of Arsenal’s season.

The afterglow of winning the Premier League over the last week has been a massive release and it has rightly gotten the vast majority of the attention in Arsenal focused circles. The World Cup squads are being released and attention for the tournament is ramping up, drawing eyes elsewhere. The transfer window and squad building plans are starting to swing into full gear after months in the background and this is something that always garners attention from everyone.

I am no different either, my attention has been split trying to get the CannonStats web app ready for a limited beta release, to test and work out bugs before opening it up to subscribers here. All of the graphics for this post were generated from this and it is very exciting but it is another thing that has stolen attention from this match.

Today we change that and dig deep into this match.

It is a bit of a free hit (as much as these games can be) with the weight of YOU MUST WIN SOMETHING lifted off Arsenal by the Premier League and with revenge on the team’s mind, I expect Arsenal to come into this game feeling lighter and with something to prove.

Just writing this last sentence has gotten me feeling excited and nervous at the same time, exactly what we want. Let’s get into it.

PSG Performance

It will come as no surprise and it won’t exactly make me a top tier scout to notice this, but PSG are a very good team. They are certainly in the conversation for the best team in the world and they have been that way for the better part of 18-24 months under Luis Enrique.

Their stats for the Ligue 1 and Champions League this season back this up.

They have a ridiculously strong attack and back it up doing a very good job of limiting what teams can generate against them.

In Ligue 1 this season they were worthy winners for the 12th time out of the last 14 seasons.

They got a bit of a push from Lens this season but ended up showing their superior overall talent in the end with the best attacking numbers and the best defensive numbers.

In the Champions League, they didn’t have quite as dominant a time as they did in France. For the second season in a row they ended up outside of the top 8 and had to take the hard road to the final with an extra knock out match and a disadvantage in the seeding.

While they haven’t been the best team in the Champions League, they have still performed at a high level to be in the tier just below and on their way to the final have dispatched both Liverpool and Bayern Munich, who were above them. The last team left in the Champions League with better numbers than them is none other than Arsenal.

That sets things up nicely and whoever emerges here will have a strong case to be the best team from the tournament overall.

PSG Form

This season has gone pretty well for PSG, they have spent most of the current campaign putting up elite (+1 or better) expected goals difference numbers.

However… recently the trend has reversed and they have posted some of their worst numbers in the last several seasons.

Here is the same rolling xG for and against over the last three seasons for Ligue 1 and the Champions League that covers the full Luis Enrique time period. The current run is the second worst it has been only beaten by a brief stretch in March of 2024.

When you look at a more detailed level, you can see that the goals scored look good with 18 scored vs just 9 allowed but the underlying numbers are not nearly as strong. Some of that could be game state and not having the motivation in the League (especially the last match against Paris FC) but it could be a bit of a warning sign for them.

Here is what it looks like in the Champions League only from the Round of 16 knockouts going forward.

Again we could point to game state here as an explanation for some of these but even though they have a +11 goal difference here it has come on a negative expected goals difference. Extremely good attackers like PSG can put on the field matters for beating xG but even the best players don’t and wouldn’t be expected to double their xG in the long run.

The individual matches in the Champions League are fascinating microcosms.

That first match against Chelsea without the gifted goals and naïvety really could have swung the other way with Chelsea getting some very good looks.

The full two-legs against Liverpool is very interesting, they dominated start to finish in the first match and then were able to withstand 2nd half pressure in the return leg while executing the defend and counter tactic to put the game away.

The overall tie against Bayern was just pure fireworks and finishing. That being said, it should have been close to tie over at 5-2 and they nearly threw it away.

In these knockouts they have played well but they have needed some hot finishing to propel themselves to the final.

They also got the positive variance on the defensive side of shooting, allowing fewer goals than expected.

PSG have some of the best attackers in the world so it isn’t exactly unheard of to see them go on a hot finishing run (while still also producing good underlying expected numbers) but the current hot run is certainly VERY hot.

We don’t want to fall victim to the gambler’s fallacy and expect it to suddenly reverse with a cold streak but it sure would be nice if that happened, or if at a minimum we got reversion towards expected finishing in the final.

Style Comparison

This match actually presents an interesting contrast between two teams, especially two teams that are thought of as very good. PSG are much more of a throwback to a different era with the way that they play, while Arsenal represent a bit more of the “new” way that is growing, especially in Premier League football.

In the Champions League, Arsenal have been much more comfortable playing with less possession and being able to play a more direct and quick style. That is not often afforded them in the Premier League. They have fewer overall passes, go long more often, and need crosses as a much larger tool to get the ball into the box.

What really stands out here is that even with those stylistic changes one that doesn’t really budge is where Arsenal matches see the action on the field occur. Even with less on the ball control, they still push the game into the third that they are attacking at a rate that is among the best.

One of the stats that I don’t have captured on the “style” view here is looking at pressing, this is one of the areas where PSG feel like a throwback to that previous generation of elite teams.

In the Champions League no team has put more pressure on the opposition than PSG according to PPDA.

They do so while playing one of the highest lines by measure of where they make defensive actions as well. Arsenal aren’t a passive defense either but PSG are just at a different level here on how much they press. Often the big teams in the Champions League will assume the ball, meaning that they just always play matches where they have the majority of possession and they don’t really have a plan for how to get it back when they come up against a team that want to contest it (like Arsenal will in this match), PSG do not fall into this trap. They have a plan to get it back and they use it as a very potent weapon.

Attacking

PSG are one of the best attacking teams in the world and have been for some time now. They have incredible depth and can throw players like Ousmane Dembele, Khvicha Kvaratskhelia, Desire Doue, Bradley Barcola at you while also having some impressive youngsters in Ibrahim Mbaye and Quentin Ndjantou plus Lee Kang-In and Goncalo Ramos to round things out.

Here is the comparison against Arsenal in the Champions League on the attacking radar. Arsenal aren’t quite as good overall in attack but I think that they surprisingly hold their own here.

Defending

PSG aren’t thought of as a defensive team but given how much of the ball that they have and the overall talent level they still show up quite well in the defensive metrics.

Here is the comparison against Arsenal in the Champions League on the defensive radar. In the Champions League Arsenal have had the edge as the more dominant defensive team this season.

Keys to the game

This isn’t a match where it will be a mystery how the teams play.

I expect that this will be a fairly even match overall but my guess is that it will start off with PSG having more of the ball but with Arsenal able to have their own spells of possession.

I expect that Arsenal will play more of a mid-block game looking to frustrate and limit entry into dangerous locations and picking their moments to try and press PSG. On the other side, I expect PSG will do their best to press hard, force Arsenal into a mistake or to go long.

I think that in the situations where Arsenal win the match, they frustrate PSG’s possession game and keep them in areas where Arsenal want them to play while winning the second balls to start their own attacks without PSG able to apply pressure imminently following the turnovers. They will still need something special from one or more of the attackers but with a much healthier front line than last season that seems like a real possibility this time.

In the situations that go PSG's way, I think Arsenal fail to execute their defensive game plan or succumb to mistakes under pressure. Last season, the pivotal goal in the first game came from Arsenal’s rotations in the press being a touch slower than PSG’s fluid buildup. If the team slips up even a little bit PSG have the quality to make you pay for it and if you gift them a chance (like Chelsea did) that can be fatal.

Last season against PSG we probably saw the worst case scenarios play out over both legs. Arsenal in the first leg had a minor mistake get maximum punishment (but probably were lucky to not concede a goal at the very end) while they had their own very presentable chances squandered.

In the second leg, I saw Arsenal play with a sense of urgency and purpose that was at the highest level and they put PSG under more pressure than just about any team had done. It took Gianluigi Donnarumma making some excellent saves to keep the lead and then another sucker punch goal that all but killed things for Arsenal.

Arsenal fans and the wider soccer watching audience I think came away with slightly different views of the overall match.

Arsenal partisans viewed it much more as one that got away with Arsenal playing well enough that they would often come away the winners and it was the bounces that decided it.

Neutrals or PSG partisans I think generally came away viewing that the scoreline reflected that PSG were the better team.

I think that this divergence of opinion remains, with many non-Arsenal fans underestimating them especially with both of last season’s matches in the semi-final being won by PSG.

Looking at the betting odds right now and the consensus points that way right now, PSG are 1.699 to win on Pinnacle which roughly translates to 57-43% odds of winning for each team. I'm not unbiased, but I can't help feeling this is much closer to a true coin flip. Arsenal aren't being given enough credit, and too much weight is being placed on performances from a period when they were dealing with major injury issues.

Scott and Adam soak in Arsenal’s Premier League title and turn their attention to the Champions League Final against PSG.

If you are a premium subscriber, you can listen to an ad-free version through Substack or you can add to your preferred podcast app through the personal RSS feed.

If you are a regular subscriber the episode is available to be added by the RSS Feed Here (copy the link) to the podcast player of your choice or click here to listen to the episode directly in your browser, through apple, or below through Spotify:

The episode opens with both hosts reflecting on the trophy lift and the emotional weight of finally getting over the line. After years of near-misses, eighth-place finishes, and a fan base conditioned to expect heartbreak, this one hits different. Scott traces his own 15-season journey as an Arsenal supporter and the stretch where a title felt unrealistic given the financial gap with Manchester City. Adam picks up the thread from his own entry point during the Emery transition, walking through the slow build under Arteta from top-four hopefuls to genuine contenders to champions.

The conversation zeroes in on what defined this season. Scott highlights a remarkable consistency stat: Arsenal were the only team in the league to never be outperformed by more than one expected goal in any single match, while doing it to opponents 20 times. Both hosts push back on the narratives that followed Arsenal all year, from questions about Jorginho’s quality to skepticism about the team’s attacking output and Arteta’s in-game management. The verdict: this was a team that won ugly when it had to and never stopped believing, even through the late wobble against Bournemouth and City.

The second half pivots to the Champions League Final against PSG. Scott and Adam break down last season’s semifinal, where Donnarumma’s saves and a couple of fortunate PSG finishes proved the difference. They identify PSG’s pressing intensity and high turnover creation as the primary threat, but also point to vulnerabilities when Hakimi and Nuno Mendes push high and leave space behind. The lineup discussion runs through every position: Calafiori gets the nod at left back for his ability to create moments, Mosquera slots in at right back with Timber’s fitness uncertain, and the midfield debate between Lewis-Skelly, Eze, and Zubimendi alongside Rice and Odegaard takes up the most real estate. Trossard earns the left wing start on big-game pedigree. Both hosts predict a tight, low-scoring affair, with Scott floating a nervy 1-0 or a 1-1 that goes to extra time, and Adam going 2-1 Arsenal. The dream ending: a set-piece goal from a defender.

Read more

26 - Wins for Arsenal, the most in the Premier League this season, next best was Manchester City with 23.

7 - Draws for Arsenal, tied for the fewest in the Premier League this season. This was shared with Fulham and Newcastle.

5 - Losses for Arsenal, the fewest in the Premier League this season, next best was Manchester City with 6.

85 - Points for Arsenal this season. This tally ranks 22nd out of 31 Premier League winners in the 38 matches played era. Interestingly, Arsenal are the only team to have won the League with 85 points. This ends up just below the average tally of 87.5 points for a winner.

Arsenal performed the best overall in the three major results this season in the Premier League. That’s no surprise given that they have won the League but it still reflects well upon the team.

In the end Arsenal ended up winning the Premier League by a fairly comfortable 7 points that was reflective of the team’s consistency and quality from top to bottom this season.

71 - Goals scored from Arsenal, second most this season, 6 behind Man City. (Pretty good for a team that was “boring” and that “couldn’t attack”). In the Premier League history for winner in a 38 match season this ranks 28th out of 31 teams, and ended up 12 goals below the average winners tally of 83.

27 - Goals allowed by Arsenal, by far the fewest in the league this season. Next closest was Man City at 35, with 8 more allowed. In the historical ranks for winners, this puts Arsenal tied for 8th, 5 goals better than the average of 32 allowed.

+44 - Arsenal’s goal difference, the best in the Premier League. It didn’t always look like Arsenal were going to end the season with the best goal difference but the fading of Manchester City ended up giving Arsenal an edge by 2 goals here. In the historical ranks Arsenal end up tied for 20th, 7 goals behind the average of +51.

There will be attempts to dismiss the Arsenal team’s achievements and to downplay them as among the worst ever. Don’t let people fool you into believing that, this team won’t be among the conversation for the best ever but they are worthy Champions and stack up against the historical record quite well.

I took a deeper dive on this in March and the analysis holds up.

Here is the main takeaway from that analysis.

They have put up good but not great attacking numbers and would rank somewhere between 25th and 30th all time. Let’s give the benefit of the doubt for League strength and say closer to 25th.

They have put been a very good defensive team and that would rank somewhere between top 3 and top 10 all time. With the boost from how strong the League is right now let’s say that is a top 5 defense.

Looking at the overall goal difference and that ranks 18th among title winners.

Where exactly you want to rank teams is messy, but I think that this shows pretty clearly that this Arsenal team is shaping up to be far from the worst ever Premier League winner.

Arsenal’s form down the stretch dipped and the team ended up a bit worse on both attack and defense but ultimately they ended up not that far from the averages. Given the strength, depth, and parity of the current Premier League, Arsenal have completed a major accomplishment.

One of the major things that I am going to remember about this team and this season is the consistency of high quality play.

0 - The number of times this season that the opposition beat Arsenal by more than 1 expected goal.

Arsenal went and did that to the opposition 20 times, building out a chance creation lead that should lead to a win more than half of the time.

Not only did Arsenal finish with the best points total, they also had the best expected points total this season.

Attack

1.65 - Arsenal’s non-penalty goals per match this season, this ranked 2nd in the League and 15th out of 96 teams in Europe’s “Big Five” Leagues

1.81 - Arsenal’s non-penalty expected goals per match this season, this ranked 3rd in the League and 13th in the “Big Five” Leagues

0.12 - Arsenal’s xG per shot this season, this ranked 5th in the League and 18th in the “Big Five” Leagues

9.76 - Arsenal’s open play shots per match this season, this ranked 6th in the League and 28th in the “Big Five” Leagues

5.14 - Arsenal’s set play shots per match this season, this ranked 3rd in the League and 11th in the “Big Five” Leagues

56.1% - Arsenal’s possession percentage per match this season, this ranked 4th in the League and 22nd in the “Big Five” Leagues

63.8% - Arsenal’s field tilt percentage per match this season, this ranked 2nd in the League and 9th in the “Big Five” Leagues

21.08 - Arsenal’s passing deep completions (within 25 yards of goal) per match this season, this ranked 3rd in the League and 17th in the “Big Five” Leagues

30.16 yards - Arsenal’s opposition average distance from their own goal for defensive actions (line height more or less) this season, this ranked 20th (lowest avg distance) in the League and 95th in the “Big Five” Leagues

Arsenal’s attack has been a source of discontent all season long. It has been a worry that it wasn’t good enough to produce a title winning season and well it turned out that it was. It wasn’t the best attack in the league and it was far from the elite of Europe.

It was however still quite good, especially in the context of the current season and the across the board decline in attacking numbers.

I don’t think that Mikel Arteta is satisfied with the attack and this was an area where injuries and lack of continuity match to match hurt the team. Being a good attack was enough this season but I don’t think that anyone at this club is happy to rest on good enough.

Defense

0.68 - Arsenal’s non-penalty goals allowed per match this season, this ranked 1st in the League and 5th out of 96 teams in Europe’s “Big Five” Leagues

0.85 - Arsenal’s non-penalty expected goals allowed per match this season, this ranked 1st in the League and 15th in the “Big Five” Leagues

0.10 - Arsenal’s xG allowed per shot this season, this ranked 1st in the League and 37th in the “Big Five” Leagues

6.08 - Arsenal’s open play shots allowed per match this season, this ranked 1st in the League and 2nd in the “Big Five” Leagues

6.08 - Arsenal’s open play shots allowed per match this season, this ranked 1st in the League and 2nd in the “Big Five” Leagues

2.62 - Arsenal’s set play shots allowed per match this season, this ranked 1st in the League and 4th in the “Big Five” Leagues

7.77 - Arsenal’s passes allowed per defensive action this season, this ranked 2nd in the League and 11th in the “Big Five” Leagues

37.86 yards - Arsenal’s average distance from goal for their defensive actions (line height) this season, this ranked 1st in the League and 12th in the “Big Five” Leagues

7.0 - Arsenal’s high possession wins per match this season, this ranked 3rd in the League and 19th in the “Big Five” Leagues

23.35 - Arsenal’s deep touches allowed per match this season, this ranked 1st in the League and 8th in the “Big Five” Leagues

If the attack was just good enough, the defense certainly wasn’t. Arsenal put together one of the strongest overall defensive season’s in the club’s history and would put themselves in the conversation for one of the best defenses in the world this season.

The crazy thing is that I don’t think that we saw the best of this team for the full season either and I think that they could be capable of doing even more. This was a season that still saw the team play stretches without Gabriel, without William Saliba, and without Jurrien Timber. All of these players have a strong argument to be the best defender at their position.

It has long been an open question if it is Defense or Attack that wins you titles, it is often both (Arsenal did that too with good attack) but for this Arsenal team the foundation and the identity of this team was built from the defensive base and that makes everything else possible.

In the incredibly wise words of Myles Lewis Skelly, they called us bottlers but now we’re holding the bottle.

Soak it in and enjoy.

The scenes of the spontaneous celebration at the Emirates Stadium were something to behold. The Club has posted a gallery of the photos and it is just beautiful scenes of ecstasy, release, and celebration.

This Club means so much to so many and every once in a while it is nice to be reminded about that.

It has been a long wait, and there have been so many painful near misses. I don’t have the same longevity as a fan as others do but in my 15 seasons following the club (damn it doesn’t feel like it has been that long and I still feel like I am “new” here) there have been a lot of near misses and you wondered if it was ever going to happen for us.

2011/12 - The 8-2 and the frantic deadline day trolley dash. Per and Mikel ended up being pretty damn important.

2013/14 - Top at Christmas but fading down the stretch due to injuries, plus the pain of the 1,000th Arsene Wenger match and the humiliation by Liverpool.

2015/16 - The high of beating Leicester on Valentines Day to the low of losing to Manchester United and Swansea back to back. I honestly thought that might have been the clubs last shot to win the title and that we’d never have it again. It looked that way for a while.

2016/17 - Getting more points than when we finished second, but still failing to make the Champions League. Still the most points for a 5th place team at 75 and matched the tally that won Manchester United a League in 96/97.

2017/18 - That semi-final against Atletico in the Europa League.

2018/19 - Baku and the fading down the stretch as the team failed to handle fighting on two fronts.

2020/21 - 8th place again, the real nadir of the team. I had serious doubts at this point if Arteta was the right man in that late fall period before it turned around.

2021/22 - Spurs + Newcastle away to throw away the Champions League spot

2022/23 - The winter World Cup, the injuries to Jesus, then Saliba and Tomiyasu

2023/24 - Aston Villa

2024/25 - Those red cards (still never seen any of them repeated to this day) and all of the injuries. Donnarumma standing on his head in Paris.

You could probably write even more about the pain that this fan base has felt over the years. Arsenal’s big and active fan base, combined with all of the near misses has made us the easy target for the media and other fans. Trying to be positive about Arsenal online was so often met with a “Warra trophy” and this gif.

Even though we knew we were a damn good team, there was no great response to that.

Now there is. There will be so many bookmarks emptied in the coming days and it will be pretty glorious to watch. If you thought Arsenal fans were insufferable online before, you haven’t seen anything yet because I think that there will be new levels to this now that we have won the title.

All of that pain along the way has made this moment all the more sweeter and special. I am soaking it all in.

I want to thank everyone who has been on the journey with us, I am so proud of this team and this community.

Arsenal are the 2025-2026 Premier League Champions.

That feels good, doesn’t it?

And yet, despite delivering the best result any Arsenal league campaign has seen in 22 years, many will remember this campaign as being fraught with adversity, close shaves and not-quite-good enough football. But is that really descriptive of what happened? Like, at all?

Listen, there were moments. Dominating Wolves but falling short at the very end tested the mettle of yours truly, my dear reader.

But the persistent thread throughout the season was the expectation — based on previous seasons, with previous versions of Arsenal and previous versions of City and previous versions of “good attacking play” — was always based on something that wasn’t happening here and now.

Did Arsenal blow the doors off everyone? Absolutely not. But we did our best here at Cannon Stats to remind folks of just what was transpiring on the pitch, and just where Arsenal were stacking up in real time. Scott reminded us all just a month ago that no, Arsenal weren’t falling apart:

All Arsenal have done since then: Win the Premier League, and qualify for the Champions League final. And sure, losses to Bournemouth and City kicked off the stretch right after Scott posted the above, but Arsenal haven’t even allowed an open-play goal since Erling Haaland scored his winner in the 65th minute at the Etihad.

I haven’t been at my most prolific down the stretch. Part of that has been pure nerves about what would happen, part is being a Dad and having Dad Stuff to do while also being too busy at work, and part of it was getting sick of making the same point over and over again. Here’s what I wrote in January, after we lost to United and many said the title charge was done and dusted:

That piece concluded with the below:

And I think that’s really the lesson of this season: One day is not the full season. City blowing someone up this week doesn’t mean they’re gonna do it for the rest of the season, particularly when they’ve got large, identifiable holes (Mateo Kovacic started today). Arsenal dropping a turd doesn’t mean they’re going to every week from here out, either.

Lessons aside: Arsenal missed Bukayo Saka for nearly a third of the PL campaign, Martin Ødegaard for about half, and Kai Havertz played only a third of the campaign. Famous(ly poor) results such as Wolves featured Piero Hincapié starting instead of Arsenal’s usual brick walls at the back. And they still won the entire damn thing. No City comeback, no Arsenal bottle, no doubt about it.

We wrap the entire Premier League season on Sunday with Arsenal atop the table. That feels pretty damn incredible. We’ll work to turn out some good content in the immediate wake of that match, including looks ahead to summer, some teams of the season type of stuff I’m planning, final position ranks and more.

Before we do, let me be clearer that ever that we love each and every one of you for supporting our site, and we’ll keep working hard to provide key context and bigger picture analysis and earn your subscription.

UP THE GUNNERS. 2025-26 CHAMPS!

This website, like the standard player team analytics will feature radar charts. They have become an industry norm after being popularized by Ted Knutson formerly of StatsBomb. They are pretty to look at, and they can (and I believe do) give you a good quick look at a player.

They aren’t perfect by any means, no graphic is and this was pointed out in a post from PyMC-Labs called “Radar Plots Must Die.”

I use radars, and I plan on continuing to use them going forward and this made me want to directly address a few of the things brought up and what I do to address them or correct some things aren’t quite right.

Let’s go through this now.

A radar plot takes a vector of percentiles and draws it as a closed polygon in two dimensions. The reader perceives the shape of the polygon; they see area, aspect ratio, pointiness, balance, lopsidedness. Yet, none of those perceptual features correspond to anything real about the player.

This might be a little ticky tacky but right off the bat, there is a generalization here that isn’t universally true and this goes to larger points down the road. It is written here that a radar is based off the percentile value, that is a way that you can present a radar but that is not how ALL radar charts are done.

Statsbomb sets the first ring at the 5th percentile value and the 95th percentile value and then has the chart be linear between those points. For reasons we will go over later, I have also adopted this approach.

The critique on how a reader perceives a graphic isn’t unique to a radar and that would be similar on most charts. It isn’t like bars correspond to anything real a player has done on a pitch than a polygon shape.

Let’s move on to the next critique, Order and adjacency are arbitrary. This is absolutely true and something that can be a problem. Let’s read what was written here in the post:

There is no intrinsic order among a given set of football metrics. xG, tackles, pass completion, aerials won are independent quantitative dimensions. The deeper problem is that a radar is an inherently spatial display, and spatial displays invite spatial inference. When two spokes are drawn next to each other, the brain reads them as neighbours, as if xG and pass completion belonged together because they are physically close on the page (they are even joined together by a line). But that proximity is an analyst's choice, not an inherent property of the data.

This is valid and the stats that you choose, and the order that you place them in matters. There isn’t an intrinsic order but there are stats that do work together and when you are doing a design you can make choices that help preset this information in a coherent fashion. xG and pass completion don’t necassariy belong next to each other but xG and shots, or xG and touches in the box? Those make sense to be near each other.

When I have designed the templates for the radars I use, I try to pick relevant stats to the position things so that they are by category and flow into each other in a way that would make sense.

The author here shows how changing the order can alter the way that you perceive the data.

This is something that is possible and that is why picking the stats and the order that you place them in can’t be done without thought and care. This isn’t a unique issue to radar’s either though. If you did something similar with bar charts (which are viewed as superior here) the same issues can arise.

I mocked up something similar here with bar charts as what was done with the radars.

The same issue can become a problem if you aren’t careful with any visual that you go with. Axis ordering can be arbitrary as pointed out here, but it doesn’t have to do and for a well designed radar it isn’t.

Let’s go to the next issue, radial encoding exaggerates strengths and hides weaknesses.

This is a technically true issue but I am not sure that this is how people that have become familiar with radars as a visual perceive the charts. Here is the critique:

Ordering aside, the radar has a second problem. The radial axis encodes percentile linearly. For example, a p80 spoke is twice as long as a p40 spoke… Every term is a product of adjacent values. Each spoke is one side of two such triangles, one with each neighbour, so its visual weight is proportional to its own percentile times the sum of its neighbours’, not to its percentile alone (and those neighbours, again, are not necessarily related to it). A high spoke flanked by other high spokes is amplified; a low spoke flanked by other low spokes is suppressed beyond what its percentile alone would suggest. The polygon a reader perceives is not the percentile vector; it is a non-linear, ordering-coupled rendering of it.

I can only speak for myself, but I don’t think that I am mentally computing the area but rather viewing the distance from the center. I looked at the example provided in the article and I struggled to even comprehend where the distortion was supposed to be happening here. Maybe I have looked at too many radars and thus the issue is less of a problem for me than it could be with others but ultimately I think that this is overblown.

It's a bit like arguing bar charts are misleading because wider bars look more important than narrow bars of the same height. Technically true about area perception, but with careful design this shouldn’t be a problem.

The careful design and the thoughtful choice and order of stats is still key here. When you intentionally group stats by skill domain, shooting metrics next to each other, creative metrics next to each other, defensive metrics next to each other; the adjacency stops being arbitrary and starts being meaningful. With this, shapes and patterns can become quick to understand things about a player quickly as you recognize the shapes start to tell you important information about a player, if you get a visible "bulge" in the defensive region or a "dip" in the creative region, this shape corresponds to a real profile and skills of a player.

The last critique brought up is another one that is technically true and is valid but again I think asks a radar do something that it is not necessarily the best tool in the toolbox if that is what you are trying to accomplish.

The author makes the point that non-adjacent comparisons are tough:

Each spoke of a radar is its own axis, oriented at its own angle. To compare any two variables, the reader has to mentally rotate one onto the other before judging which value is larger; the further apart the two spokes sit, the larger that rotation and the costlier the comparison.

Yes, if your task is “compare this player’s xG percentile to their tackles percentile with precision,” a radar makes that harder than a bar chart. You’re comparing two spoke lengths at different angles against curved grid lines versus two bars on the same baseline. Humans are better at comparing lengths on a shared axis than lengths at different angles.

Practically, does this matter? When does a reader sit down with a player radar and think “I need to determine whether this player’s progressive carries percentile is higher or lower than their aerial win percentile, and by exactly how much”? Almost never. That’s a table task. If you need precise numerical comparisons between specific metrics, you should be looking at a table, not any visualization at all. Bar charts are also worse than tables for this task.

The actual task radars serve is rapid profile recognition. You glance at it and within a second or two you get an idea of where this player is strong/average/weak and a sense that this is a defensive destroyer, this is a creative playmaker, or this is a well-rounded midfielder.

A radar is a tool, it is not always the best tool for the job. It is important that we recognize this.

This is really the ultimate takeaway form this for me.

It is basically impossible to have one visual that has no tradeoffs and is perfect for every situation.

My own personal preference is to go with a package of visuals. Show different templates, show radars, percentile charts, and bar charts. Here is an example that I created recently looking at Arsenal’ loanee Fabio Vieira.

Any one chart, and even multiple isn’t enough to tell you everything about a player. There is always more that you can try and present and even then, the stats can only go so far.

Radars aren't perfect, no visualization is. The critiques raised in the PyMC Labs piece are technically valid in isolation but they either apply equally to the proposed alternatives, or they assume the radar is being built without care. A well designed radar with thoughtfully selected stats, intentional grouping by skill domain goes a long way to accomplishing the task of communicating information quickly about a player. My own implementation includes a distribution to try and the viewer the best of both worlds without getting overly busy and dense.

The goal shouldn't be to kill radars, it should be to use them alongside other visuals as part of a package, each tool doing what it does best while recognizing and teaching about the limitations.

When the final whistle blew on Arsenal’s 1-0 blotting out of Atletico Madrid Tuesday night, Mikel Arteta and his team of reds let out a collective sigh of relief that crescendoed into unfiltered joy. They’d held on to yet another one-goal win, though truthfully their opponents had once again not truly threatened them for most of the night, picking up just one shot in 90 minutes that xG models gave a better than 10% chance of ending up in the net.

In a season that has been defined by criticism, from both inside and outside the house, and by discussion more about what they cannot do rather than what they can, this Arsenal side confirmed their place in history. They will play for European — and likely global — football’s largest club title on May 30, an honor that even many great players get to fight for maybe once in an entire career, if that.

You read that right — Bukayo Saka, Gabriel, David Raya, William Saliba, Declan Rice, Martin Ødegaard, Kai Havertz, Myles Lewis-Skelly and all the rest will enter that Budapest evening just 90 minutes from being Champions League winners. What has always seemed one round or three games or six weeks away could come their way in just two hours once that game kicks off.

Tuesday’s full-time whistle brought me immense joy, and yet on many levels it still has not fully sunken in just how enormous an achievement this is. Rival fans will cite the path to the final as being too easy, the smothering wins over Leverkusen, Sporting and Atletico as being unimpressive, and they’ll very much pit whichever of PSG or Bayern advances as the “what’s right with football” to the villain that is Arsenal’s boring, ball recycling, defensively robust unit.

Those discussions will take place over the next three weeks and change, during which we’ll also discover the final destiny of Arsenal in this season’s Premier League title race. But for the moment, just consider how significant this step is for Arsenal — not fully done, of course, but on a plane that even Arsenal’s most legendary players rarely have reached.

• The 2026 Champions League final will be only Arsenal’s 2nd in club history, and first since 2006. Max Dowman was still three years from birth when that match was played, and even Christian Nørgaard was still in Danish primary school at the formative age of just 12.

• Not only are Arsenal still alive 14 matches into their Champions League run; they’ve yet to be beaten even once in the campaign. Their 29 goals scored is already the second-highest in a CL season in club history, behind only last season’s 31. Their six goals conceded is half the number that any other knockout round participant allowed, and five less than next-lowest Pafos FC allowed during the league stage alone.

• Despite some of the narrative about their performances this season, Arsenal are on a +23 goal difference in the competition, which matches Bayern Munich’s and bests PSG’s by a goal. Last season’s champion PSG also finished +23, while Pep’s unbeaten 2023 winners finished +27.

Regardless of whatever nits your mate or older relative might want to pick with this side and how they play, Arsenal are in rarefied air here. Mikel Arteta has accomplished what so many fans both within our ranks and not have publicly, boisterously doubted he could by bringing Arsenal to the brink of football’s international promised land. Not only did Arsenal’s players show grit and desire and creativity to make it here - they defended with conviction, without panic and with the kind of fiery confidence that proven winners show, not serial bottlers or paper tigers. This is a far cry from just a handful of years ago, where we were begging Arsenal to show the fight it took to merely qualify for Europe, yet alone run the table to a final.

With that Champions League final ticket officially booked and just eight magical points separating Arteta from a Premier League crown that has eluded them so long, it’s time to set aside all narratives, all trivia and all concerns and embrace this moment and this team for what they are: Special. Though the job is still yet to be finished, this group has delivered something so many have failed to do.

No, it isn’t a trophy, but that’s worthy of celebration.

Happy Monday all, Scott continues to prioritize things like “friends” and “family” and is traveling so you’re stuck with me this Monday.

At times of late, things have just been so busy that there’s almost no time for either of us to get too lost in our thoughts — we’re running family shuttles to and from practices, completing big projects around the house, traveling across the country to attend family events and so on. Heaven knows in my house, we’ve got birthdays for four kids (two still with us, two passed on) in a mad dash from the end of February through the first of May. This week also brings the beginning of another soccer season for my son, which of course means I again don my reflective sunglasses and whistle and do my best to age young men, as it were.

As it happens, I watched the last 30 minutes of Saturday’s Newcastle-Arsenal clash on my phone from a different soccer practice for my son, glancing up and down to watch him make plays while also keeping a close eye on proceedings in London. Far from the first time this season, and I’m sure it won’t be the last over the coming years.

I didn’t really set out to tell you all about my daily schedule here, rather to make the point that things are busy, and while I’ve had so many thoughts about the experience of this season, some of them pretty repetitive, I haven’t had as much time to organize them as I’d like.

Then there’s the tension. As a content creator, for lack of a better term, sometimes it’s easy to dwell in the sanctuary of recurring pieces, as I’ve done with my player rankings that drop weekly or so. But on the topic of reactive or analytical pieces, I admit I’ve been feeling a bit of a paralysis lately. Things have changed pretty rapidly week to week, and we can talk all we want about trends and underlying data, but there’s also the element of…we just don’t know what will happen, and we’re all pretty anxious at this point to reach the end of this and discover our fate.

As we experience this journey, I’ve seen every reaction you could think of, from “this shouldn’t be like this and damn everyone in control for putting us through this” to “actually, it’s pretty impressive we’re still here, all things considered.” Lately I’ve found myself thinking about that contrast of opinions specifically, and while I won’t set out to write in depth on any of this today, it’s left me with a few thoughts I think are worth digging into a little below:

The whole injury thing

It feels like it’s been a long time since we’ve been able to say Arsenal weren’t dealing with a weird number of injuries. Even Saturday, we lost Eze and Havertz and tails immediately tucked between legs, assuming that Kai was done for the season, taking with him any hope Arsenal have of winning anything.

But even that small thought in and of itself introduced a kind of cognitive dissonance we’ve been living with as a fanbase for the better part of two seasons. I posted a bit about it over the weekend, but we got Bukayo Saka back on Saturday after about a month on the shelf. We’re still waiting for Calafiori and Timber and Merino, but I’ll set them aside for now.

Saka is, to me, unquestionably the best player Arsenal have. That’s backed by the years-long panic that his current contract in any moment will be his last at Arsenal, lest we prove to him we can win things. And this weekend’s appearance was his first in the Premier League since March 14, his first in any Arsenal match since the EFL Cup final five matches ago.

Unbelievably, the player many once feared would be ground into dust under the thumb of Mikel Arteta has made just 22 Premier League starts this season, 32 across all competitions. He’s seventh among outfield players in minutes played, behind even Viktor Gyökeres. Even if Saka starts all of the maximum seven matches remaining this season, he would end this season on 69 all competitions starts (nice). For contrast, Federico Valverde has started 106 matches since the beginning of last season. Erling Haaland has started 89. Bruno Fernandes, whose club aren’t even in Europe this season, has started 86.

Of course, Arsenal have had other big injury issues. Martin Ødegaard has made 12 fewer starts than Saka this season and Kai Havertz has started 22 fewer — and to me, those three represent Arsenal’s first and second best attackers and their best creator, for what that’s worth. But just forgetting that for a second, whenever Arsenal come under the microscope for their performances of late, I find myself coming back to this Saka question. Using the above Haaland and Bruno as examples, what would we expect of City and United’s performances if they missed 20 more games than they had? Would we expect those sides to perform as they did with those players, in City’s case remaining in the Champions League and staying atop their domestic league? Would we expect United to remain favorites for third? And if not, why has the expectation for a Saka-less (and Ødegaard-less, for many of the same games) Arsenal to not only stay first, and stay in the Champions League, but to show out and offer their best form of the season? Food for thought.

Needless to say, keeping Saka on the pitch could be a big deal for what remains of the season. As we saw immediately upon his introduction Saturday, he makes a difference.

Reverting to how I began this section, it feels like Arsenal have dealt with a great deal of injuries over the past two seasons, right? I’ve even seen it suggested that this has gone on longer than that. To be clear, Arsenal have absolutely been injured a lot this season and last. Per Premier Injuries, the club suffered the fourth-most time-loss injuries last season (data not yet available for this campaign), but that’s where the outsized injury issues do end. In 23-24, Arsenal were second from bottom in time-loss injuries. In 22-23, third from bottom. As in, having the fewest.

But why does this keep happening? It’s not as simple as 1+1=2 (if it were, building such a deep squad would’ve fixed it), but one domino that has repeatedly set things into motion that needs to be acknowledged, as unsatisfying as it is — shit luck.

When Kai Havertz was clattered into in August, breaking bones in his knee that would keep him out until December or January, that wasn’t an overuse issue. In fact, he’d just come off the bench. Same for Ødegaard getting his ankle rolled in a friendly against Slovenia last season, or getting into a vicious knee-to-knee with Crysencio Summerville this season. And you don’t even need to mention how uncontrollable Mikel Merino’s collarbone or foot breaks in training or Gabriel Jesus’ and Jurrien Timber’s ACL tears were.

But those injuries, crappy luck as they were, came with knock-ons. Ødegaard missed a busy stretch last season at the same time as Merino with his collarbone. That thinned Arsenal’s midfield options and created a lot of work for Declan Rice and Kai Havertz, while also thinning Arsenal’s options at 9. When Timber tore his ACL, that created a body of work that likely contributed to Ben White’s thigh injury, which cost him time last season, further catapulting Timber into a big workload last season and this (and now he’s dealing with a muscle issue, naturally). Havertz’s contact injury this summer essentially forced Gyökeres and Jesus minutes that may not have even happened this season had things gone to plan, let alone driving reduced attacking output in the side.

I’ve certainly seen it suggested before that this could be addressed through squad build, adding more pieces, but it’s hard for me to get on board with the idea that Arsenal’s three left backs, three strikers, 8 or 9 midfielders, two very good right backs, etc., are a thing group. And while there’s been some sentiment that Kai Havertz is “glass” or Ødegaard can’t be relied upon, what’s happened these past two seasons is an outlier for both.

In fact, you may remember Havertz being praised by Mikel Arteta for almost never being injured just last season. And the 63 games he’s missed over the past two seasons are a shift from his career trends and then some.

Ditto Ødegaard, whose ankle roll cost him 14 games last season, as many or more games than he’d missed in every other season to date than one. By the way, as he’s gained a rep these days for being inconsistent…he also only missed one more game last season after returning in December.

But that’s then, this is now, right? Can these two be counted on? That’s the question Arsenal’s physios and front office will have to do their best detective work on to answer. Are Havertz’s knee injury — which accounts for nearly half of his time missed of late — and Ødegaard’s rolled ankle and clashed knee injury — more than half his time missed right there isolated incidents, or have they set off a chain of injury problems that do happen to some athletes as they age? Honestly, it’s impossible to say.

Ekeing it out

This is another one I touched on via social media earlier this weekend, but the concept of barely hanging on to wins has followed me throughout this season. The narrative even goes back to the good times, such as a 1-0 October win over Palace where Arsenal conceded 7 shots and less than half an expected goal - about half of that coming from one shot.

Or there’s a December win at Everton where Arsenal conceded five shots, the last coming at 78 minutes. A week later, Arsenal beat Brighton 2-1 in a match where the Seagulls took four shots after a very lucky bounce let them halve the lead in the 64th minute. These wins were largely dominant and involved Arsenal snuffing out chances on the other side but were not considered by many Arsenal fans and talking heads to be “Good Wins.”

We’ve seen this rise to prominence again lately, be it Arsenal’s controlled two-leg win over Sporting, the draw away to Leverkusen, against Chelsea in the Carabao, Saturday’s result or even the return leg with Brighton, where Arsenal defended a 1-0 lead for 81 minutes but Hurzeler’s men didn’t even get a shot off after the 69th minute.

It’s so, so easy to understand why it’s more fun to win by two goals or three (hell, four) rather than one. One is what an American football pundit might call a “one-possession game.” But digesting all of this information led me to post the below this past weekend:

As I said, like everyone else I will never prefer a one-goal win, particularly in the moment. But another thing worth thinking about - particularly as we enter an age of Premier League talent compression and the potential further reduction of overall goal production, especially from open play - is what under siege really looks like, versus closing out a mostly uneventful one-goal win.

(Note: I have thoughts about how this is impacting attackers’ output and whether players are actually “off” versus just facing more obstacles, but that’s for another day.)

Of course, all it takes is a moment, one mistake. But those come more frequently when you’re not snuffing out the majority of chances before they really get started.

What the heck is gonna happen?

Man, I wish I knew. But Saturday, whether you hated the performance or not, was a huge three points. 20% of what was available from there out. Newcastle are 14th but they’re good, and just got Bruno G back.

By the time the match with Fulham ends next Saturday, Arsenal will have played two matches more than City, just three to go. Potential is there to be six ahead when City-Everton kicks off on the 4th of May, but then City also play Brentford before Arsenal kick another PL ball. So you could go from tied, to six ahead, to tied again, in pretty short order.

It’s going to be a wild ride, with odds fluctuating too much for comfort, and too many scenarios and permutations to count. The biggest thing at this point is to just win…no style points up for grabs here.

Arsenal finally took the big swing on an attacker this past offseason, signing their first number 9 since Kai Haver — er, Gabriel Jesus in 2022. And while seemingly all anyone ever wants is “big guy, kick ball hard,” some of the more refined elements of a striker’s game have been largely missing this season - the ability to consistently hold up play without losing the ball, getting on the end of crosses in the air and making deft movements in the box, to name a few.

Not all of that stuff is easy to measure through a formula, which is why I put a ton of weight on actual film study as well. Last summer that led me in a different direction than what Arsenal did, though the old version of the ADAM formula made the same recommendation.

So, how about going forward? It’s possible Arsenal could move for a new attacker this offseason, and if they decide they’re not done at the number 9 spot, some of the names below will certainly come up.

As per usual, things to keep in mind as you read: This is based on stats, and stats aren’t always perfect. I have always encouraged watching the games for yourself rather than creating an entire opinion based off only the numbers.

The final rank is based on full two-season score, with guys in the top 25 fitting your standard profile for “transfer-friendly age,” which is usually slightly pre-prime. Let’s get into it!

Read more