How Many Games Before the League Table Tells the Truth?
How much luck sits in the league table after 5, 10, 19 and 38 games, how often the weaker team is ahead, and what that means for early-season statistics.
By Adam from Poland · · 5 min read
After five rounds, newspapers print the league table as if it meant something. Sometimes it does. Often it is mostly luck. This guide puts numbers on how quickly the table starts to reflect how good teams really are, and what that means for reading early-season statistics.
Why one match tells you little
A football match has only three outcomes and few goals, so a single result carries a lot of chance. Take an ordinary mid-table side that wins 40% of its games, draws 27% and loses the rest. It averages about 1.47 points per game, but the points it takes in any one match vary a lot: the standard deviation is about 1.3 points.
Averaging over more matches shrinks that uncertainty, but slowly, with the square root of the number of games:
| Games played | 5 | 10 | 19 | 38 |
|---|---|---|---|---|
| Typical error in points per game | ± 0.58 | ± 0.41 | ± 0.30 | ± 0.21 |
After five games, a team’s points per game could easily be half a point away from its true level. That is the difference between a European place and a relegation fight.
How often is the worse team ahead?
A more intuitive test: take two teams, one genuinely good (about 1.8 points per game) and one ordinary (about 1.4). Simulate their seasons many times and count how often the weaker team is strictly ahead in the table:
| After | 5 games | 10 games | 19 games | 38 games |
|---|---|---|---|---|
| Weaker team ahead | 27% | 22% | 16% | 9% |
After five games the weaker side is ahead more than a quarter of the time. Even at the halfway point it happens about one time in six. Over a whole season, the table mostly gets it right, but not always: roughly one season in eleven, the worse team still finishes above the better one.
What this means in practice
- Early tables exaggerate. Surprise leaders after a month are usually good teams that also got lucky, and they tend to slip back.
- Home and away splits are thinner still. After ten games a team has played only about five at home. Big gaps between home and away records in the first half of a season are mostly noise.
- Goals settle faster than points. Goal difference uses more information than win-draw-loss, so it separates teams sooner. Shot and chance numbers settle faster again.
- Last season still matters. This is why models carry a team’s previous form into a new season and let the new results take over gradually.
How this affects our predictions
Every match page shows a confidence level. It is lower when the model has little recent data to work with, for example at the start of a season or for a newly promoted side. Treat low-confidence figures as rougher estimates, and the form guide explains why short runs deserve the same caution.
The same arithmetic applies to judging predictions themselves: a short record says little about a model. See how to judge football predictions.
The figures above come from a simple simulation with fixed win, draw and loss chances. Real seasons add injuries, transfers and changes of manager, which make tables noisier, not less.