Elo
flagship metric · data to Mon 12 Oct 2026A rating of how strong a club (or a national team) has proved to be, computed by 11footy from results and goals only.
Every side has a number. After each match the winner takes points from the loser, more for a wider margin and for a result the ratings did not expect. A rating difference of 100 points means the stronger side is expected to score about 64% of the available points.
- Club rating: per league, on that league's scale
- National-team rating: one scale for all internationals
What goes in
our base data only- Clubs: the final score of every league match since 2014-15, in the order played; each league has its own scale.
- National teams: the final score, the tournament and whether the ground was neutral, for every international since 1872 (the World Football Elo method).
How it is computed
the formula in words- Expected score = 1 / (1 + 10^(-d / 400)), d = own rating - opponent's (+ the home advantage for the home side).
- New rating = old + K x margin factor x (result - expected score); result 1, 0.5 or 0.
- Clubs: K = 20, home advantage 60, margin factor 1 for 0-1 goals, 1.5 for 2, (11 + N) / 8 from 3; a new season keeps 75% of the distance from 1500. National teams: K from 20 (friendly) to 60 (World Cup finals), home advantage 100.
- The rating used for a match is always the one before it.
| Constant | Value |
|---|---|
| Clubs: rating at the first season of the data | 1,500 |
| Clubs: rating of a newcomer to a league | 1,450 |
| Clubs: K (points at stake per match) | 20 |
| Clubs: home advantage (points) | 60 |
| Clubs: share of the distance from 1,500 kept into a new season | 75% |
| National teams: first rating | 1,500 |
| National teams: home advantage (points, not on neutral ground) | 100 |
| National teams: K, World Cup finals | 60 |
| National teams: K, continental finals and major tournaments | 50 |
| National teams: K, qualifiers and Nations Leagues | 40 |
| National teams: K, other tournaments | 30 |
| National teams: K, friendlies | 20 |
The constants are fixed beforehand and never fitted on the results they are later checked on.
When to trust it
sample and threshold policy- A rating needs a history: the clubs' ratings start at 1500 in 2014-15, so the first two seasons only warm them up and are not part of the check.
- A club that was not in the league the season before starts at 1450.
- National teams start at 1500 at their first recorded match; the check begins in 2010, long after every established side has a history.
How well it works
worked out from the data, not typed in- Clubs, 2016-17 to 2026-27: on 18,234 matches, each scored with a mapping fitted on the seasons before it, log loss 0.985 against 1.071 for the league's home/draw/away frequencies alone.
- National teams, 2010 to 2026: on 16,278 matches, log loss 0.888 against 1.048 for the home/neutral frequencies alone.
Clubs
21,886 league matches rated since 2014-15| Season | Matches | Log loss, Elo | Log loss, frequencies | RPS, Elo | RPS, frequencies |
|---|---|---|---|---|---|
| 2016-17 | 1,826 | 0.946 | 1.051 | 0.192 | 0.227 |
| 2017-18 | 1,826 | 0.970 | 1.065 | 0.197 | 0.229 |
| 2018-19 | 1,826 | 0.985 | 1.071 | 0.199 | 0.228 |
| 2019-20 | 1,725 | 0.993 | 1.070 | 0.205 | 0.232 |
| 2020-21 | 1,826 | 1.006 | 1.090 | 0.206 | 0.236 |
| 2021-22 | 1,826 | 0.997 | 1.076 | 0.203 | 0.230 |
| 2022-23 | 1,826 | 0.990 | 1.063 | 0.204 | 0.229 |
| 2023-24 | 1,752 | 0.977 | 1.076 | 0.195 | 0.229 |
| 2024-25 | 1,752 | 0.988 | 1.079 | 0.201 | 0.233 |
| 2025-26 | 1,752 | 0.997 | 1.071 | 0.203 | 0.229 |
| 2026-27 | 297 | 0.973 | 1.068 | 0.197 | 0.228 |
| All | 18,234 | 0.985 | 1.071 | 0.200 | 0.230 |
Each season is scored with a mapping from the Elo difference to home win, draw and away win (one slope, two thresholds) fitted only on the seasons before it. The comparison is the league's own frequencies of home wins, draws and away wins in those seasons, which know the home advantage but nothing about the teams. Lower is better for both scores; the last season is still under way. 2014-15 and 2015-16 only warm the ratings up.
National teams
all internationals with a rating| Year | Matches | Log loss, Elo | Log loss, frequencies | RPS, Elo | RPS, frequencies |
|---|---|---|---|---|---|
| 2010 | 863 | 0.948 | 1.069 | 0.190 | 0.229 |
| 2011 | 1,119 | 0.859 | 1.042 | 0.167 | 0.225 |
| 2012 | 1,025 | 0.920 | 1.036 | 0.189 | 0.227 |
| 2013 | 963 | 0.903 | 1.047 | 0.175 | 0.222 |
| 2014 | 856 | 0.949 | 1.052 | 0.190 | 0.226 |
| 2015 | 1,039 | 0.874 | 1.046 | 0.175 | 0.232 |
| 2016 | 920 | 0.880 | 1.034 | 0.173 | 0.223 |
| 2017 | 924 | 0.916 | 1.063 | 0.178 | 0.224 |
| 2018 | 929 | 0.935 | 1.059 | 0.184 | 0.225 |
| 2019 | 1,149 | 0.837 | 1.035 | 0.165 | 0.227 |
| 2020 | 347 | 0.986 | 1.083 | 0.196 | 0.229 |
| 2021 | 1,115 | 0.802 | 1.036 | 0.150 | 0.224 |
| 2022 | 970 | 0.910 | 1.037 | 0.182 | 0.224 |
| 2023 | 1,054 | 0.866 | 1.047 | 0.173 | 0.232 |
| 2024 | 1,231 | 0.899 | 1.059 | 0.174 | 0.226 |
| 2025 | 1,002 | 0.819 | 1.042 | 0.158 | 0.228 |
| 2026 | 772 | 0.896 | 1.057 | 0.179 | 0.233 |
| All | 16,278 | 0.888 | 1.048 | 0.175 | 0.227 |
The same walk, by calendar year, with the mapping fitted on the matches from 1990 to the year before. The comparison is the frequencies on home and on neutral ground.
What it does not tell you
limits- Club ratings are computed league by league: a 1600 in one league is not a 1600 in another, so the ranking of the five leagues together is a rough order, not a measurement.
- Only results and goals: no line-ups, injuries, transfers or expected goals.
- Clubs start at 1500 in 2014-15; the first seasons are a warm-up.
- The European cups have their own rating of the cup clubs, on its own scale, shown on the Europe pages.
Where you see it
each one has a ? beside itTeams: strength rankingTeam pages: strength over timeMatch pages: Elo ranking of the two sidesNational teams: strength ranking
Expected goals source: 11footy's own expected goals. The same description as data: /lab/api/metrics/elo.
Data to (the latest league match played), built . Unless a season is chosen, every view shows 2025-26 (2026-27 is still young: 0 players have 900 league minutes so far, so the numbers are of 2025-26 until half as many as last season do.). Percentiles and per-90 ranks from 900 league minutes, within the player's role, five leagues together; under 450 minutes a small sample.