Can a Current Chelsea Striker Score 20 League Goals Next Season?
No.
But let’s do a fun exercise in math to see hypothetical situations under which they could. Please note, none of this is mathematically rigorous, this is all just speculation and fun.
Goals scored can be calculated using the following formula
where npxg per shot is the player’s non-penalty xG per shot and npxgplus per shot is the player’s non penalty xG overperformance per shot, ie (non penalty goals-npxG)/shots. This isn’t some fancy unproven probabilistic model, this is pure primary school arithmetic. As it should be pretty obvious, in order to score a lot of goals, a player should take a lot of shots, those shots should be of relatively good quality (as indicated by high npxg per shot) and the player should be a pretty good finisher (as indicated by high npxgplus per shot). Below are all the players who have scored at least 20 league in one season in one of the Top 5 European leagues.
And here is the same data in a table format.
The minimum number of shots taken by a 20 goal scorer was 87, and that was done twice, once by Saido Mane in 2018 and once by Danny Ings in 2019. The quality of shots varies greatly, from Lionel Messi, who made a career taking a lot of difficult shots (aggregate npxg per shot of 0.11 over the last 4 seasons), to Erling Haaland this year who did a remarkable poaching job with 0.23 npxg per shot. However, one thing that seems almost invariant is that you have to be a clinical finisher to get 20 goals. Only one such player had a negative npxgplus per shot over the last 4 seasons, Robert Lewandowski in 2017 who had a negligible negative npxgplus per shot of -0.01.
With that out of the way, lets look at how the three Chelsea players who are possible candidates to play striker next season did. There is an obvious caveat here that they did not always play as strikers and that Chelsea changed coaches half way through the season.
Well, this is not great. Only Werner had shot volume approaching anything you’d expect from a 20 goal scorer, though in the case of Abraham and Havertz, its understandable as they played a lot less minutes, in the case of the former, and often not as a striker in the case of the latter. All three underperformed their xG, in the case of Werner and Havertz, significantly so. But, at least they got high quality shots, so there’s something to build on.
Let’s see if we can use this baseline to build something resembling a 20 goal per season striker out of these three. For starters, the average 20 goal striker on the list above played 2,712.5 minutes per season, so let’s normalize everyone’s minutes to that first. Maintaining the same shots per minute ratio, we get the following.
Clearly, Abraham is the primary beneficiary of increased minutes here. With Average 20 Goal Striker Minutes, he’s already up to 15.6 goals. Considering that last season he scored 15 in 2,215 minutes, this is not all that surprising. But we are not there yet.
Both Werner and Havertz had epically bad seasons in terms of finishing. Out of 880 players who took at least 50 shots over the last 4 seasons, Werner’s npxgplus per shot this season ranks 865th. Havertz, had he taken 50 shots, would have ranked 859th. It’s safe to say that neither is actually that bad. Prior to this season, they’ve recorded the following npxgplus per shot in their previous three seasons
Considering they’re young and this represent their developmental seasons (especially so in the case of Havertz), let’s take a weighted average of these, applying greater weight to more recent seasons (I told you this will get wild). This gives us npxgplus per shot of 0.034 for Kai Havertz and 0.024 for Werner. This would place them at the bottom of the range for the 20 goal scorers above, but not outside of it, like their current numbers are. Using these npxgplus per shot numbers gives us the following results
Things are starting to look more functional, but we’re still not there yet. We’re still not taking enough shots. We need to take more shots.
We could assume one of them would play more minutes, but historically we probably shouldn’t. Our last 4 20 league goal scorers (Didier Drogba twice and Diego Costa twice) have played 2,944, 2,770, 2,068 and 3,082 league minutes for an average of 2,716, so our estimate of 2,712.5 is right on the money. Instead we will need to take more shots per minute. More shots per minute isn’t free, though. Shooting more has an inverse relationship to shot quality. Instinctively this makes sense — if you’re taking more shots, it’s because you’re now taking the kind of shots you’d previously pass up, not tap ins 3 feet in front of goal. How much does increasing the number of shots reduces shot quality? By performing a linear regression (a topic for another time) on the 20 goal scorer dataset, I approximate that each additional shot you take reduces npxg per shot by -0.000773. Let’s have each of our footballers take 100 shots and reduce npxg per shot appropriately.
We are almost there! Let’s have them take 120 shots instead. For what its worth, 120 shots would put each of them 14th in shots taken out of the 29 players who have scored 20 or more league goals in the last 4 seasons.
And there we have it. Our first 20 league goal scorer since Diego Costa in 2017, taking 120 shots and scoring 20 goals, Kai Havertz! Incidentally, in 2017, Diego Costa scored 20 goals by taking 111 shots.
So there you have it. We, too, can have a 20 goal scorer. All it takes is Kai Havertz playing a lot more minutes, taking a lot more shots, reverting to finishing numbers that are closer to his historical average and not having our shot quality degrade more than expected due to increased shooting volume. Simple, right?
