Should Eddie have passed? The judge is in session

This has become my "thing" so let's look at the Nketiah opportunity against Brentford

I have sort of inadvertently made a name for myself as the guy who comes in and second-guesses players on the pass vs shot decision. I didn’t mean for this to happen but it is a burden that I will carry and the judge is in session ready to make the call on the latest decision.

The play here is Eddie Nketiah getting a chance in the 85th minute against Brentford.

From the screenshot above you can see why this is one that is having the question asked. As he goes to shoot here, he has both Bukayo Saka and Martin Odegaard in a very good spot to take his chance and turn it into a very good chance.

Snapshots of a moment in time are hard to judge and can make things look easier than they actually are so here is the full sequence and even a nice second-angle replay from the match footage.

Watching the clip there is certainly enough time for him to get his head up and see where his teammates are, he receives the ball 26 yards from goal, takes three touches to carry it to 19 yards out, and from there he takes the shot.

So let’s set the baseline here for what actually happened, this is by no means a bad shot, my model rates it as an 11% chance of being scored while Opta rates it as a 4% chance. I think my rating might overrate it some, while Opta’s probably underrates it some so let’s split the difference and call it a 7% chance which I think is roughly about right for a shot from that location but following an error with a defense that is scrambling to get back into shape.

Ultimately he doesn’t do a lot with it, he does put it on target but it is weak and very savable I rate this as a 92% saveable shot with Opta putting at 95%. This is kind of irrelevant to the decision-making analysis but still important information.

So baseline for this shot is that taking the shot turns this into a 7% chance of scoring, on my goal probability added model the starting chance of scoring from that spot is 2.9% so his choice to shoot is a positive outcome but let’s try out some hypothetical options and see what could have been.

Pass to Saka or Odegaard

Making this pass on the model actually rates as one that is not overly difficult, with an expected completion percentage of 72%. I think that is right, he’s not under immense pressure and has a pretty decent window to pass to Saka, and he is on his stronger right foot.

This pass on my goal probability added also improves the chances of scoring quite a bit going from 2.9% to 6.3% just by moving the ball central.

Based on that it is still a slightly less optimal choice because his shot took it from 2.9% to 7% but the odds of getting a shot off by Saka there are probably pretty high.

This is now getting into a little more difficult estimation but let’s assume Saka is able to control and get a shot off 50% of the time from that spot. I am also assuming the shot happens basically right away after receiving the pass and is from blue area that I have highlighted below:

Plugging that into my xG model gives us an xG estimate of 30-50%. That is a pretty big increase in the chance quality.

Turning these estimates into an expected value gives us, roughly 11% to 18% for the chance of the choice of trying to pass to Saka turning into a goal with the above assumptions. Even taking a bit off the pass completion probability with Nketiah being a below-average passer that puts the higher error bars of this in the 8% to 20% range of this turning into a shot.

Now let’s turn to the Odegaard pass opportunity.

This pass is even clearer and one that rates as a little easier to complete, this is expected to be completed nearly 80% of the time.

This pass on my goal probability added also improves the chances of scoring quite a bit but not nearly as much as the Saka pass going from 2.9% to 4.2%. This is probably more likely to turn into a shot with Odegaard able to time things and probably take a first-time left-footed finish that we have seen so many times. Let’s say this turns into a shot 90% of the time here.

It is further out so the xG on this is more 20-30% when I plug it into the model.

Turning these estimates into an expected value gives us, roughly 14% to 22% for the chance of the choice of trying to pass to Odegaard turning into a goal with the above assumptions. Doing the same discount for the passing skill of Nketiah and putting higher error bars of this choice it comes out in the 11% to 25% range of this turning into a shot.

Judge Rules

Taking the shot is not a bad choice here. Nketiah does add value here from the situation by taking the shot.

He also doesn’t make the best choice in this situation.

How much we think this is not optimal depends on the starting assumptions but I think I made some fairly conservative choices here and doing either pass is a choice that increases Arsenal’s chances of scoring, with the best choice here going to Odegaard (but that is highly dependent on how much you think Saka is able to turn his chance into a shot is, and I think that’s probably a bit harder).

He does turn a decent opportunity into a pretty good shot. He doesn’t put a ton of power on it or finish it right but we shouldn’t get too outcome-driven here.

What he gives up is probably a bit better with at worse a roughly equal probability of it turning into a goal and at best it being about 3 times as good as taking a shot.

Not a bad choice, but not the best choice.

Jude rules: Should have passed, but not a mistake

This is and in my mind is always a fun exercise to do. What do you think was passing the right choice here?