Prices read a league table better than results do. And the range matters more than the middle.
- Within a season, ten games in. Built from betting prices and scored on four past seasons, our 80% range for a club's final position held 78% of the time in the Premier League, 79% in the Championship and 84% in League One. Built from results alone: 66%, 51% and 59%. Read straight off the table: 21%, 11%, 11%.
- One, two and three seasons ahead. On 815 club forecasts from 2012/13 to 2023/24, we beat the historical base rate for which division a club ends up in, at every horizon.
- The weak spot. For clubs we call fairly safe, two and three seasons out, we overstate the chance they change division. We say about one in five. It happened about one in eight (137 and 61 clubs).
The rules
Every forecast uses only what was known at the time: the table after ten games, the fixtures left, and the Bet365 closing prices of the games already played. Team strengths come from those prices. The rest of the season is played out thousands of times, play-offs included.
Seasons ahead work from the same start. Where a club finishes this season sets where it starts next. After that, history decides: how often clubs in that spot went up, stayed or went down, using only seasons that had finished before the forecast was made.
This does not beat the betting market. It turns the market's view into division probabilities with a range you can check. That is the point.
Within a season
Four seasons, 2022/23 to 2025/26, forecast from ten and twenty games in. An 80% range should catch the final position eight times in ten. Position error is the ranked probability score: lower is better.
| Division, games played | From prices: error | Inside 80% range | From results: error | Inside 80% range | Table as it stands: error |
|---|---|---|---|---|---|
| Premier League, 10 | 0.092 | 78% | 0.102 | 66% | 0.134 |
| Premier League, 20 | 0.061 | 83% | 0.071 | 74% | 0.097 |
| Championship, 10 | 0.117 | 79% | 0.176 | 51% | 0.233 |
| Championship, 20 | 0.105 | 75% | 0.114 | 66% | 0.177 |
| League One, 10 | 0.097 | 84% | 0.119 | 59% | 0.172 |
| League One, 20 | 0.080 | 82% | 0.087 | 71% | 0.131 |
80 Premier League club-seasons per row, 96 in the other two. Prices win every row. Results alone get close by twenty games but are well behind at ten, and their ranges are too narrow: they look sure and miss.
Seasons ahead
Which division will a club be in one, two and three seasons from now? We scored the probabilities with the Brier score, lower is better, against two simple answers: the club stays where it is, and the historical rate of moving from that division.
| Seasons ahead | Forecasts | Ours | Historical rate | Stays put |
|---|---|---|---|---|
| One | 815 | 0.285 | 0.378 | 0.469 |
| Two | 815 | 0.402 | 0.458 | 0.614 |
| Three | 747 | 0.470 | 0.522 | 0.742 |
The gap narrows as you look further out, as it should. About three clubs in ten are in a different division two seasons later, and "they'll stay put" is the worst answer at every horizon.
Where it goes wrong
We checked whether the probabilities mean what they say, by grouping clubs on the chance we gave them of being in a different division.
| Chance we gave of a different division | Seasons ahead | Clubs | We said | Happened |
|---|---|---|---|---|
| Under 10% | One | 308 | 4% | 6% |
| 10% to 25% | One | 237 | 16% | 20% |
| 25% to 50% | One | 137 | 36% | 34% |
| 50% or more | One | 133 | 68% | 59% |
| Under 10% | Two | 91 | 4% | 2% |
| 10% to 25% | Two | 137 | 20% | 12% |
| 25% to 50% | Two | 456 | 35% | 34% |
| 50% or more | Two | 131 | 59% | 57% |
| Under 10% | Three | 63 | 4% | 0% |
| 10% to 25% | Three | 61 | 18% | 13% |
| 25% to 50% | Three | 418 | 41% | 36% |
| 50% or more | Three | 205 | 59% | 57% |
All twelve groups are shown. Three misses to know about. One season out, clubs we give a 10% to 25% chance of moving move a little more often than we say: 16% against 20%. Clubs we call likely to move next season move less often than we say: 68% against 59%. Two and three seasons out, clubs we call fairly safe are safer than we say.
Change runs both ways, up and down, so these errors are not automatically on the cautious side for a lender. Read the range, not the middle.
What it's worth
Division sets central income. From the Premier League's own 2024/25 figures, the club that finished 20th was paid £109.2m and the champions £174.9m. Below that the drop is steep.
| Where a club plays | Premier League money, 2024/25 |
|---|---|
| Premier League, 20th | £109.2m |
| Championship, first season after relegation | £48.9m |
| Championship, second season after relegation | £40.1m |
| Championship, no parachute (solidarity) | £5.4m |
| League One (solidarity) | £0.8m |
The two parachutes are 55% and 45% of the Premier League's equal share, and the Championship figure is 11% of the first parachute. All three reproduce from its table to the rounding shown. The League One figure comes from secondary sources. EFL central money, matchday and commercial income are not included.
Put the two together and a promotion chance becomes a number in pounds, with a cautious case beside it. That is the question a buyer, a lender or a licence application is really asking.
Data sources
- football-data.co.uk: results and closing prices (Bet365, and Pinnacle where available) for the Premier League, Championship, League One, League Two and National League, 2005/06 onwards.
- API-Football: current fixtures and results.
- Premier League: central payments to clubs, 2024/25.
One change after the first run. Our first three-season forecast forgot how strong a club was once it stayed in the same division, and gave the best Premier League sides about a one in ten chance of being in the Championship by 2029/30. We now carry each club's standing forward. The three-season score moved from 0.473 to 0.470; the one and two season scores did not change. We are telling you because we changed it after seeing the result.
Pricing a club, a loan or a licence application? We'll run the division forecast for the club you're looking at, with the range and the money beside it. Email advanceddata7@gmail.com.