How often do the best performers win in football and in other sports?
Words of wisdom from soccer boffin Kevin Pullein
A tennis player with a 60 per cent chance of success in each point would win nearly all of their matches. A football team with a 60 per cent chance of scoring each goal that is scored will win half of their matches.
The best competitors win less often in football than in tennis and many other sports.
This is why football is so often frustrating to watch. Perhaps the best way to wager on football, though, is by remembering how often it is frustrating to watch. Many times the score mocks the play.
I studied expected goals stats from three sources for Premier League games since the start of last season. Expected goals stats are an estimate of how many goals would be scored on average with the same number of attempts from the same positions.
There was a strong correlation between expected goals and actual goals. That is to say, teams with 1.0 expected goals in a match averaged 1.0 goals, teams with 2.0 expected goals in a match averaged 2.0 goals, and so on.
But those averages came from all sorts of different individual makeups. Teams with three expected goals for a match sometimes scored three goals. Often, though, they scored two or four, or one or five, or nought or six and upwards.
A lot of times teams scored fewer than their expected goals and lost. They had not played badly. They just got less than they deserved. A lot of times teams scored more than their expected goals and won. They had not played well. They just got more than they deserved.
I expressed each team’s expected goals as a percentage of both teams’ expected goals. Imagine, for example, a fairly commonplace match in which the home team created 1.5 expected goals and the away team created 1.0 expected goals. The total expected goals for the match were 2.5. The home team created 60 per cent of the expected match goals and the away team created 40 per cent of the expected match goals.
I did this for all matches. Then I compared the percentage of expected match goals with the percentage chance of winning. A team had to have created 60 per cent of the expected match goals to have a 50 per cent chance of winning.
My technical models tell me what in effect is the same thing. A team must have a 60 per cent chance of scoring each goal that is scored in a match to have a 50 per cent chance of winning.
When such teams do not win they draw slightly more often than they lose. But if you are better than your opponents even drawing is a disappointment.
A team with 60 per cent of the expected goals in a match will take on average 60 of the shots, as you might have anticipated.
In a typical Premier League match there will be about 25 shots. If one team have 60 per cent of the shots and the other team 40 per cent then one team will take 15 shots and the other team ten.
Suppose a team have 15 shots for and ten against, match after match after match. Half of the time they will not win. How frustrating will that be, for them, for anyone who supports them and for anyone who has backed them?
Why do the best competitors win less often in football than in so many other sports? Why would a football team with a 60 per cent chance of scoring each goal win much less often than a tennis player with a 60 per cent chance of prevailing in each point? (In practice, the tennis player would have a higher chance when serving and a lower chance when receiving. I am simplifying.)
Perhaps the most suitable explanation in this context is to say that there are more points than goals. In a football match there might not be any goals. If there are no goals nobody can win. In a best of three sets tennis match there must be at least 48 points. In a best of five sets tennis match there must be at least 72 points.
A tennis match is broken down into sets and games rather than being merely the first to so-many points. This adds uncertainty, but not as much as is added to a football match by the possibility of no goals and the probability that there will not be many goals. A large sample is more likely to give an accurate impression of the underlying reality than a small sample.
Sixty per cent of cars are black, white or grey. It is more likely that most of the cars you will see are black, white or grey if you are driving on a busy motorway than if you are parked in a quiet side street.
Follow us on Twitter @racingpostsport
Like us on Facebook RacingPostSport