Somewhere between a fifth and a third of it. We measured it in 14 leagues and got much the same answer in every one.
Corrected on 25 September 2026. Figures updated after a data fix. The range of the real share moved up a little (19% to 34%, from 18% to 32%), so the headline now says a fifth to a third. Details at the end. The calls and the register are unchanged.
- Roughly one part skill to two to four parts sample size. Depending on the league, between 19% and 34% of the difference in finishing between strikers is real. Never below 19%, and never more than about a third.
- So a hot season mostly doesn't travel. And the noisier the league, the more one big season flatters the player.
- Our ranges mean what they say. When we gave an 80% range, the outcome landed inside it 82% of the time, across 277 seasons.
- 61 calls on this season, frozen today. They're all below. They settle in May, including the ones that go against us.
"Finishing" here means one number: league goals divided by shots on target. Cups and internationals are excluded. "Striker" means everyone the data feed labels an attacker or a forward, so centre-forwards and wide forwards both.
How you separate skill from luck
Take every striker season with at least 15 shots on target. Their conversion rates spread out, and most land somewhere between 20% and 50%.
Two things cause that spread. Strikers really do differ, and a season is only a few dozen shots, so every rate has noise in it.
The second part can be estimated directly from the number of shots. Take it away and what's left is the real difference between players.
| League | Real share |
|---|---|
| Championship | 34% |
| Serie B | 34% |
| La Liga | 33% |
| Premier League | 32% |
| Bundesliga | 32% |
| Belgian Pro League | 32% |
| Ligue 2 | 31% |
| Serie A | 29% |
| Ligue 1 | 28% |
| Primeira Liga | 26% |
| 2. Bundesliga | 26% |
| Scottish Premiership | 24% |
| Segunda División | 21% |
| Eredivisie | 19% |
Fourteen leagues, four seasons each, and the answer barely moves. In any one season, most of the difference in finishing between two strikers comes down to sample size.
It also lines up with our earlier persistence work, which got at the same thing a different way in the Premier League.
We ran the same split on midfielders, and there it depends heavily on the league: close to nothing in the Premier League and Serie A, over 40% in La Liga. The data provider's position labels vary between leagues as well, so we're not making any general claim about midfielders.
What that means when you buy one
The obvious guess is that a striker will repeat last season. We compared that guess with ours on every season-to-season pair we have.
- In the Premier League, repeating last season is as good as anything we do. Over 45 pairs it's a tie: we come out a fraction behind, by an amount chance alone easily explains.
- Across the other 13 leagues taken together we get closer, and the gap grows with how much a league's numbers bounce around from season to season. The correlation between the two is 0.86 across 14 leagues. Belgium, on 12 pairs, is a tie too.
- Serie B and the Scottish Premiership are where repeating last season does worst, though on small samples (10 and 8 pairs).
Nothing clever about it. Pulling a noisy number back towards its average should help most where the number is noisiest. The useful bit is knowing where that bites: a big season in a noisy league is the one most likely to be overpaid for.
Our ranges mean what they say
A range is only any use if you can trust it. Say 80% and it should happen about 80% of the time. Too often and we're hedging, too rarely and we're overconfident.
| Seasons | Inside our 80% range | |
|---|---|---|
| Every striker and midfielder pair, 14 leagues | 277 | 227 (82%) |
| Strikers whose finishing had just jumped 25%+ above their own norm | 60 | 53 (88%) |
The second row is the tough one. It's exactly the situation where a buyer is most tempted to overpay, and the range still held.
The ranges are wide, because one player's season is a few dozen shots and nobody can be precise about that. If a single-player projection looks precise, it's hiding its interval somewhere.
61 calls, written down now
A test on past seasons always invites the objection that the cases were picked after the fact. A call made before the result cannot be. So we ran the same rule on last season and froze every player it returned.
The rule, applied mechanically: every attacker (strikers and wide forwards, under either of the data feed's two labels for them) whose 2025/26 league conversion was at least 1.25 times his own established rate, with 15 or more shots on target before it and in it. That gave 61 players across 13 leagues. None were added or removed by hand.
How they settle, fixed now rather than in May:
- The outcome is his league conversion over the whole of 2026/27, in the same league as last season.
- Fewer than 15 shots on target and the call counts neither way. It is reported as insufficient evidence.
- A player who changed league is reported separately and is not in the main score. His rate is then being measured against different defences. 10 of the 61 are now in a different league: 7 transferred, 3 went up or down with their club.
- The score is how many outcomes land inside the range, and how close we get compared with simply repeating last season.
- Each call settles after the last round of its league's 2026/27 season. Every call is reported, including the ones that go against us.
Timing. The register was frozen on 22 September 2026, between four and eight rounds into the season depending on the league. No 2026/27 figures were used in any call. The live season was read only to see which club each player is at now.
The file is public and it will not change:
preregistered-calls-2026-09-22.json. Its SHA-256
fingerprint is de4967ca53e1797784ee3133fa6422dd860b7b9b2a7cf13b0582a167f90d389f. If the file is ever
altered, that fingerprint changes.
| Player | League 2025/26 | 2025/26 | His norm | Our call | 80% range | |
|---|---|---|---|---|---|---|
| A. Ezzalzouli | La Liga | 38% | 14% | 26% | 13% to 39% | |
| C. Daferner | 2. Bundesliga | 41% | 18% | 32% | 18% to 47% | |
| A. Muja | Belgian Pro League | 38% | 17% | 29% | 12% to 47% | |
| Y. Gboho | Ligue 1 | 50% | 24% | 33% | 18% to 47% | |
| Guilherme Schettine | Primeira Liga | 53% | 25% | 36% | 22% to 56% | |
| D. Calvert-Lewin | Premier League | 41% | 20% | 28% | 17% to 42% | |
| D. Durand | Ligue 2 | 62% | 32% | 43% | 27% to 59% | |
| Z. Vipotnik | Championship | 64% | 33% | 49% | 36% to 64% | |
| A. Trezza | Primeira Liga | 47% | 25% | 35% | 15% to 54% | |
| Dani Gómez | Segunda División | 47% | 25% | 33% | 17% to 50% | |
| I. Prtajin | 2. Bundesliga | 50% | 27% | 34% | 22% to 47% | |
| F. Pittarello | Serie B | 38% | 22% | 30% | 17% to 43% | |
| Beto | Premier League | 53% | 31% | 37% | 22% to 56% | now Fiorentina, Serie A |
| Gerard Moreno | La Liga | 62% | 36% | 41% | 22% to 56% | |
| Z. Davitashvili | Ligue 2 | 56% | 33% | 41% | 24% to 57% | |
| Y. Sor | Belgian Pro League | 24% | 14% | 22% | 7% to 40% | |
| L. Sinayoko | Ligue 1 | 32% | 19% | 28% | 13% to 43% | |
| L. Ramazani | La Liga | 32% | 19% | 27% | 12% to 41% | now Burnley, Championship |
| C. Ramírez | Primeira Liga | 40% | 25% | 35% | 23% to 50% | |
| M. Whittaker | Championship | 42% | 27% | 32% | 21% to 42% | |
| Raphinha | La Liga | 54% | 34% | 38% | 25% to 50% | |
| I. Baldé | Ligue 2 | 52% | 33% | 41% | 25% to 56% | |
| J. Doku | Premier League | 33% | 21% | 28% | 14% to 43% | |
| I. Sarr | Premier League | 43% | 28% | 34% | 20% to 48% | |
| L. Schuler | 2. Bundesliga | 50% | 32% | 38% | 17% to 58% | |
| C. Morris | Championship | 60% | 39% | 43% | 30% to 57% | |
| T. Lauritsen | Eredivisie | 44% | 29% | 32% | 21% to 42% | now Sampdoria, Serie B |
| M. Godts | Eredivisie | 40% | 27% | 37% | 21% to 50% | now Paris Saint-Germain, Ligue 1 |
| O. Dembélé | Ligue 1 | 56% | 37% | 40% | 25% to 54% | |
| S. Mito | Eredivisie | 39% | 26% | 32% | 13% to 53% | |
| R. Højlund | Serie A | 48% | 32% | 38% | 23% to 54% | |
| André Luiz | Primeira Liga | 37% | 25% | 31% | 14% to 50% | |
| R. Jiménez | Premier League | 47% | 32% | 36% | 20% to 50% | now Wolves, Championship |
| Toni Martínez | La Liga | 39% | 27% | 35% | 23% to 50% | |
| L. Boyé | La Liga | 52% | 36% | 41% | 26% to 58% | |
| A. Ueda | Eredivisie | 54% | 38% | 45% | 31% to 62% | |
| A. Gouiri | Ligue 1 | 47% | 33% | 36% | 22% to 52% | |
| Gabriel Silva | Primeira Liga | 33% | 24% | 28% | 14% to 43% | |
| D. Welbeck | Premier League | 46% | 33% | 37% | 22% to 52% | |
| O. McBurnie | Championship | 44% | 32% | 37% | 27% to 49% | now Hull City, Premier League |
| G. Moncini | Serie B | 52% | 38% | 40% | 27% to 55% | |
| F. Abiuso | Serie B | 44% | 32% | 38% | 19% to 56% | |
| Richarlison | Premier League | 42% | 31% | 35% | 23% to 50% | |
| A. Novaković | Serie B | 31% | 23% | 28% | 14% to 43% | |
| Murilo | Primeira Liga | 34% | 26% | 30% | 14% to 43% | |
| Jorge de Frutos | La Liga | 36% | 27% | 32% | 16% to 47% | |
| E. Gliozzi | Serie B | 35% | 27% | 33% | 13% to 53% | |
| B. Thomas-Asante | Championship | 50% | 38% | 41% | 24% to 57% | now Coventry, Premier League |
| N. Futkeu | 2. Bundesliga | 69% | 52% | 56% | 40% to 72% | now SV Elversberg, Bundesliga |
| J. Pohjanpalo | Serie B | 68% | 52% | 54% | 42% to 68% | |
| G. Hirst | Championship | 38% | 29% | 34% | 19% to 50% | |
| Ronald | Championship | 25% | 19% | 25% | 7% to 43% | |
| João Pedro | Premier League | 54% | 41% | 44% | 28% to 60% | |
| E. Demirović | Bundesliga | 55% | 42% | 44% | 32% to 57% | |
| Clayton | Primeira Liga | 43% | 34% | 36% | 24% to 48% | |
| C. Tzolis | Belgian Pro League | 39% | 30% | 34% | 23% to 46% | now Arsenal, Premier League |
| G. Isaksen | Serie A | 26% | 21% | 25% | 11% to 39% | |
| V. Vermeij | 2. Bundesliga | 61% | 48% | 49% | 32% to 64% | |
| F. Schleusener | 2. Bundesliga | 46% | 37% | 39% | 25% to 54% | |
| F. Diédhiou | Ligue 2 | 50% | 40% | 42% | 28% to 61% | |
| F. Bonazzoli | Serie A | 31% | 25% | 30% | 12% to 47% | now Cremonese, Serie B |
Christos Tzolis is on the list through the rule, separately from our piece on his assists. On his own finishing we have him at 34%, with a range of 23% to 46%. He has since moved to Arsenal, so under rule 3 he is reported separately.
Where this stops
- We're not claiming to predict better than a club's own analysts. In the Premier League we don't even beat the simplest guess.
- No verdict on any transfer. Fee, fit, role and minutes all matter more than this.
- Nothing beyond finishing. One number, measured carefully.
Method
League football only, 14 leagues, the four completed seasons from 2022/23 to 2025/26, players with 600 or more minutes. Player seasons are season totals (sources below), with no event or tracking data. Seasons are compared only within the same league.
Each estimate starts from the league's typical striker, fitted only on seasons up to the decision point, then moves towards the player's own record in proportion to how much of it there is. The 80% range allows for both how little we know about the player and how few shots a season contains.
Every call gets written up here in May, misses included.
Data sources
- API-Football: per-season player statistics for league football (minutes, shots, shots on target, goals, assists, key passes, passes, dribbles, duels, tackles, fouls, blocks, interceptions, match rating) and the position label. 14 leagues, 2022/23 to 2025/26.
- API-Football: transfer records and 2026/27 appearances, used only to say which club each player is at now.
Edited for style on 22 September 2026. The calls, the settlement rules and the register file are unchanged.
Corrected on 25 September 2026. When a player split a season between two clubs, we had dropped the shorter spell if it was under 600 minutes, and left out players whose two spells only reached 600 together. Counting the whole season moved the real share of finishing up by one to three points in most leagues, and by seven in Serie B: the range is now 19% to 34% (was 18% to 32%). Our ranges held 227 of 277 seasons (82%, was 224 of 275) and 53 of 60 spike cases (was 51 of 58). The Premier League is still a tie against repeating last season, now a fraction behind rather than a fraction ahead, and not by more than chance. The link between how noisy a league is and how much we gain strengthened, from 0.77 to 0.86. The 61 calls, their rules and the register file are unchanged, as promised.
Want this run on a player you're working on? We'll do the first one free. How it works →