How can I predict football draws ?
Unlike sports that generally have higher scores, such as rugby and American football, football has a good chance of ending in draws. In this article, we'll look at how to predict which matches are more likely than others to end in a draw. Easier said than done? Read this article to find out.
During the 2022/23 season, let's take a look at what happened, in terms of draws, in the 5 major European leagues in the 1st Division (Level 1) and 2nd Division (Level 2).
Overall, draws accounted for around 25% of matches, or 1/4 of games, with the most common scores being, in order, 1 - 1, then 0 - 0 and finally 2 - 2.
Intuitively, a draw is more likely between two teams of relatively equal quality, taking into account the venue.
A title contender playing host to a struggling team would have an implied probability of a draw estimated at around 14%, while the probabilities would rise to around 30% in the case of a meeting between two mid-table clubs.
We can confirm this by applying the widely used Poisson distribution to two theoretically equivalent teams.
While a purely Poisson distribution approach slightly underestimates the probability of a draw in football, which requires adjustment, the methodology is nonetheless relatively straightforward.
Calculating the chances of a draw
Let's take the example of a match with an expected total of 2.5 goals between two teams of equal strength. In this scenario, each team will score an average of 1.25 goals against its opponent.
Poisson's law estimates that both teams have a 29% chance of not scoring, so the unadjusted probability of a 0-0 tie is obtained by multiplying these two probabilities.
0,29 X 0,29 = 0,08
In line with the actual match data, a 1-1 draw has a greater chance of happening, almost 13%.
Once calculated, the probabilities of all possible ties (0-0, 1-1, 2-2, etc.) are added to obtain the overall probability of a draw for a typical game between two teams of the same level.
In this example, without the correction for the slight deviation between the Poisson distribution and reality, a draw is expected to occur about 27% of the time.
While the propensity of two balanced teams to draw is easy to assess, the probability of a club scoring more or fewer goals is often overlooked when it comes to predicting draws.
Inevitably, the lower the total number of goals expected in a match, the more likely each team is to score few points.
The probability of scoring no goals in a situation that is less conducive to points (2.2 total goals expected in an evenly-matched match) rises from 0.29 in the previous case to more than 0.33.
Conversely, the odds of a 0-0 tie rise from 0.08 to 0.11, and the unadjusted cumulative probability of a draw exceeds 0.29.
If we select the matches most likely to end in a draw, we inevitably find teams of a similar level, who may have shown a certain tendency to be good in defence but not very effective in attack.
Predicting which matches are likely to end in a draw
However, knowing how to identify the prospects of a draw in no way guarantees that this high probability has not been taken into account in the available odds.
It is also important to bear in mind that a team that puts in extreme performances, sometimes even over an entire season, is likely to have more moderate results in the future.
If a club seems to specialise in draws, partly because the games in which it participates have few goals, the persistence of this characteristic could be overestimated, making future ties seem more interesting than they actually are.
The teams with the most draws in a season between 2006 and 2023 were generally in the middle of the table.
Together, these teams averaged almost 16 draws in a season, compared with a league average of just 10 over the same period.
The following season, however, the number of draws recorded by these potential specialists dropped from 16 to just over 10, which is very close to the competition average.
Regular winners such as Manchester City, Tottenham and Liverpool, who seemed to shun draws, suffered a similar fate.
From an average of less than five draws each per season during the course of a single campaign, they too regressed the following season: an average of eight draws between them, a figure that brings them close to the league average of 10 draws.
It is essential to understand the factors that can influence draws in football, although, as always, it is important to bear in mind the regression to the mean of the extreme values; moreover, as the climax of the season approaches, it is also interesting to consider the situations in which a draw can benefit both teams, particularly in Italian Serie A.
Thursday, February 8, 2024
In my same category
Can xG be used to predict soccer matches ?
Saturday, February 24, 2024
In this article, we will attempt to understand what expected goals are, use xGs to predict football matches and, finally, analyse the model created by the duo Mark Dixon and Stuart Coles from the University of Lancaster in the North West of England.Sports...See the article
Calculating Expected Value for sports betting
Wednesday, February 14, 2024
Get to grips with a basic Expected Value formulaLearn how to work out whether you should make a bet or notExpected ValueThe amount a player can expect to win or lose if they were to place a bet on the same odds many times over, calculated through a simple...See the article
How to use a live odds screen ?
Wednesday, January 31, 2024
Every bettor should use a live odds screen. Unfortunately, a lack of available information results in many never realising the merits of live odds software. With multiple free options now available, this article will serve as an introduction to using a li...See the article