In this article, we explore a model developed to predict the number of corners for a game of soccer using the method of Poisson distribution and statistical data. This formula has been developed by Sports Loci and is used to make daily predictions for soccer games.
Most Common Corner Outcomes
The average match features anywhere from 9 to 11 corners and lower numbers such as 4 to 7 for more
competitive leagues, as found by FootyStats. Therefore, it is within this range that we expect to find results.
Factors for Predicting Corners
There are many different factors that can affect the number of corners seen throughout the course of a match, including the attacking styles of both the home and away sides, as well as the tempo of the match, for example. The formula should take into account the performance of a given team in three different areas, the corner strength of a team in recent performances, over the course of the current season, and historically vs. their opponent for the match in question. Different weights must be given to the situations mentioned above, totaling 100%. Recent performances are the most telling for a match, as well as seasonal performance, whereas historical data plays a smaller role in predicting outcomes.
Weights:
Recent Weight: 0.45 or 45%
Seasonal Weight: 0.35 or 35%
Historical Weight: 0.20 or 20%
Obtained Corner Strength of a Team
In order to create a model to predict the number of corners that will be seen in a game, a sense of the strength of a team in terms of corners obtained each match must be developed. This, of course, will be developed using recent, seasonal & historical data.
The first step is determining the average number of corners obtained in each game by a team in each respective situation.
Example:
Team Recent – 5.43 Corners
Team Seasonal – 4.6 Corners
Team Historical (against an opponent) – 5.1 Corners
Next, determine the average number of corners obtained by any team in each situation. For historical data, this number is the average corners by both teams over the last 5 historical encounters.
Example:
Recent Average – 4.7 Corners
Seasonal Average (League Average) – 4.623 Corners
Historical Average – 5.12 Corners
The strength of a team is then taken for each situation using the formula below:
Strength = Team Average / Overall Average
Example:
Recent Def Strength: 5.43 / 4.7 = 1.1553
Seasonal Def Strength: 4.6 / 4.623 = .9950
Historical Def Strength: 5.1 / 5.12 = .9961
The overall strength of a team is then the weighted sums of each situation using the weights specified in the top paragraph.
(Recent Strength * .45) + (Season Strength * .35) + (Historical Strength * .2)
Example:
(1.1553 * .45) + (.9950 * .35) + (.9961 * .2) = 1.0534
Defensive Strength
The defensive strength of a team in terms of corners is how good a given team is at not conceding corners over the course of a game. This can be determined similarly to the method explained above for calculating a team’s attacking or corner obtainment strength.
In order to develop defensive strength, we must determine the average number of corners conceded by the team each game and the average of concerns conceded in each situation, respectively.
Example:
Team Recent – 3.2 Corners
Team Seasonal – 4.1 Corners
Team Historical (against an opponent) – 3.87 Corners
Recent Average – 5.1 Corners
Seasonal Average (League Average) – 3.98 Corners
Historical Average – 6.23 Corners
The defensive strength of a team is then taken for each situation using the formula below:
Strength = Team Average / Overall Average
Example:
Recent Def Strength: 3.2 / 5.1 = .6275
Seasonal Def Strength: 4.1 / 3.98 = 1.030
Historical Def Strength: 3.87 / 6.23 = .6212
A sum should again be developed using the weights for each given situation as done above and for each team, respectively.
(Recent Strength * .45) + (Season Strength * .35) + (Historical Strength * .2)
Example:
(.6275 * .45) + (1.030 * .35) + (.6212 * .2) = .7671
Overall Average
The overall average of corners for teams across all situations is found using the same formula as before:
(Recent Avg. * .45) + (Season Avg. * .35) + (Historical Avg. * .2)
Example:
(5.1 * .45) + (3.98 * .35) + (6.23 * .2) = 4.934 Corners
Possible Number of Corers for Each Team & Totals
The predicted number of corners for either team can be calculated using the formula below:
Overall Attacking Strength * Overall Defensive Strength * Overall Average
Example:
1.0534 * .7671 * 4.934 = 3.987 Corners
The above is done for both teams, and their respective probable number of corners can be combined to find the probable number of total corners for the entire game.
Poisson Distribution
Using the statistical method of Poisson Distribution and the mean number of corners (probable corners for each team) developed above: the likelihood of each total of corners for either team can be obtained. Poisson distribution delivers a likelihood from 0% to 100% of each outcome. Since no team can end with 3.987 corners, for example, Poisson distribution is used with this number found to find the probability of an exact number of corners.
For Example:
Using a possible mean corner for Team A of 3.987 Corners
Poisson Distribution (Lamba = 3.987 & X = 3) for 3 corners = .196 * 100 = 19.6 %
Poisson Distribution (Lamba = 3.987 & X = 4) for 4 corners = .19536 * 100 = 19.536 %
Poisson Distribution (Lamba = 3.987 & X = 5) for 5 corners = .15578 * 100 = 15.578 %
Probability of Total Corners for Both Teams
Now that the likelihood of total corners for either team has been created. These totals for each team can be added together to obtain a likelihood of the total number of corners for both teams on the pitch and hence the match. Combining two totals and probabilities, the likelihood of the entire total for a match is explained below:
For Example:
Team A Probability for 3 Corners = .196
Team B Probability for 6 Corners = .0987
Probability of 9 Corners = .196 * .0987 = .0194 * 100 = 1.94%
Of course, this is just the probability of team A obtaining 3 corners and Team B obtaining 6 Corners, but 9 Corners can also be achieved in total if, for example, Team A obtained 4 corners and Team B 5. Therefore, the summated probability (As done above) must be done for each possible way to achieve a total and summed together for a final completed probability of a total.