Statistical association football predictions

From Wikipedia, the free encyclopedia
Jump to: navigation, search

Statistical Football prediction is a method used in sports betting, to predict the outcome of football (soccer) matches by means of statistical tools. The goal of statistical match prediction is to outperform the predictions of bookmakers, who use them to set odds on the outcome of football matches.

The most widely used statistical approach to prediction is ranking. Football ranking systems assign a rank to each team based on their past game results, so that the highest rank is assigned to the strongest team. The outcome of the match can be predicted by comparing the opponents’ ranks. Today, eight different football ranking systems exist: FIFA World Rankings, World Football Elo Ratings, AQB Sports Ratings, The Roon Ba, InternationalMark, Rsoccer, Mondfoot and Chance de Gol.

There are three main drawbacks to football match predictions that are based on ranking systems:

  1. Ranks assigned to the teams do not differentiate between their attacking and defensive strengths.
  2. Ranks are accumulated averages which do not account for skill changes in football teams.
  3. The main goal of a ranking system is not to predict the results of football games, but to sort the teams according to their average strength.

Another approach to football prediction is known as rating systems.While ranking refers only to team order, rating systems assign to each team a continuously scaled strength indicator. Moreover, rating can be assigned not only to a team but to its attacking and defensive strengths, home field advantage or even to the skills of each team player (according to Stern [1]). An example of a football rating system is the pi-rating system[2] which provides relative measures of superiority between football teams (also applicable to other sports), and which is said to outperform considerably (in terms of profitability against the betting market) the widely accepted ELO rating system.

History[edit]

Publications about statistical models for football predictions started appearing from the 90s, but the first model was proposed much earlier by Moroney,[3] who published his first statistical analysis of soccer match results in 1956. According to his analysis, both Poisson distribution and negative binomial distribution provided an adequate fit to results of football games. The series of ball passing between players during football matches was successfully analyzed using negative binomial distribution by Reep and Benjamin [4] in 1968. They improved this method in 1971, and in 1974 Hill [5] indicated that soccer game results are to some degree predictable and not simply a matter of chance.

The first model predicting outcomes of football matches between teams with different skills was proposed by Michael Maher [6] in 1982. According to his model, the goals, which the opponents score during the game, are drawn from the Poisson distribution. The model parameters are defined by the difference between attacking and defensive skills, adjusted by the home field advantage factor. The methods for modeling the home field advantage factor were summarized in an article by Caurneya and Carron [7] in 1992. Time-dependency of team strengths was analyzed by Knorr-Held [8] in 1999. He used recursive Bayesian estimation to rate football teams: this method was more realistic in comparison to soccer prediction based on common average statistics.

Football Prediction Methods[edit]

All the prediction methods can be categorized according to tournament type, time-dependence and regression algorithm. Football prediction methods vary between Round-robin tournament and Knockout competition. The methods for Knockout competition are summarized in an article by Diego Kuonen.[9]

The table below summarizes the methods related to Round-robin tournament.

# Code Prediction Method Regression Algorithm Time Dependence Performance
1. TILS Time Independent Least Squares Rating Linear Least Squares Regression No Poor
2. TIPR Time Independent Poisson Regression Maximum Likelihood No Medium
3. TISR Time Independent Skellam Regression Maximum Likelihood No Medium
4. TDPR Time-Dependent Poisson Regression Maximum Likelihood Time dumping factor High
5. TDMC Time-Dependent Markov Chain Monte-Carlo Markov Chain model High

Time Independent Least Squares Rating[edit]

This method intends to assign to each team in the tournament a continuously scaled rating value, so that the strongest team will have the highest rating. The method is based on the assumption that the rating assigned to the rival teams is proportional to the outcome of each match.

Assume that the teams A, B, C and D are playing in a tournament and the match outcomes are as follows:

Match # Home Team Score Away Team Y
1 A 3 - 1 B y_{1}=3-1
2 C 2 - 1 D y_{2}=2-1
3 D 1 - 4 B y_{3}=1-4
4 A 3 - 1 D y_{4}=3-1
5 B 2 - 0 C y_{5}=2-0

Though the ratings r_{A}, r_{B}, r_{C} and r_{D} of teams A, B, C and D respectively are unknown, it may be assumed that the outcome of match #1 is proportional to the difference between the ranks of teams A and B: y_{1}=r_{A}-r_{B}+\varepsilon _1. In this way, y_{1} corresponds to the score difference and \varepsilon _1 is the noise observation. The same assumption can be made for all the matches in the tournament:

\begin{matrix}
   y_{1}=r_{A}-r_{B}+\varepsilon _{1}  \\
   y_{2}=r_{C}-r_{D}+\varepsilon _{2}  \\
   ...  \\
   y_{5}=r_{B}-r_{C}+\varepsilon _{5}  \\
\end{matrix}

By introducing a selection matrix X, the equations above can be rewritten in a compact form:

\mathbf{y}=\mathbf{Xr}+\mathbf{e}

Entries of the selection matrix can be either 1, 0 or -1, with 1 corresponding to home teams and -1 to away teams:

\begin{matrix}
   \mathbf{y}=\left[ \begin{matrix}
   2  \\
   1  \\
   -3  \\
   2  \\
   2  \\
\end{matrix} \right], & \mathbf{X}=\left[ \begin{matrix}
   1 & -1 & 0 & 0  \\
   0 & 0 & 1 & -1  \\
   0 & -1 & 0 & 1  \\
   1 & 0 & 0 & -1  \\
   0 & 1 & -1 & 0  \\
\end{matrix} \right], & \mathbf{r}=\left[ \begin{matrix}
   r_{A}  \\
   r_{B}  \\
   r_{C}  \\
   r_{D}  \\
\end{matrix} \right], & \mathbf{e}=\left[ \begin{matrix}
   \varepsilon _{1}  \\
   \varepsilon _{2}  \\
   \varepsilon _{3}  \\
   \varepsilon _{4}  \\
   \varepsilon _{5}  \\
\end{matrix} \right]  \\
\end{matrix}

If the matrix \mathbf{X}^{T}\mathbf{X} has full rank, the algebraic solution of the system may be found via the Least squares method:

\mathbf{r}=\left( \mathbf{X}^{T}\mathbf{X} \right)^{-1}\mathbf{X}^{T}\mathbf{y}

If not, one can use the Moore–Penrose pseudoinverse to get:

\mathbf{r}=\mathbf{X}^+ \mathbf{y}

The final rating parameters are \mathbf{r}=[1.625,\ 0.75,\ -0.875,\ -1.5]^{T}. In this case, the strongest team has the highest rating. The advantage of this rating method compared to the standard ranking systems is that the numbers are continuously scaled, defining the precise difference between the teams’ strengths.

Time-Independent Poisson Regression[edit]

According to this model (Maher [6]), if X_{i,j} and Y_{i,j} are the goals scored in the match where team i plays against team j, then:

\begin{align}
 X_{i,j} &\sim \text{Poisson}(\lambda ) \\ 
 Y_{i,j} &\sim \text{Poisson}(\mu ) \\ 
\end{align}

X_{i,j} and Y_{i,j} are independent random variables with means \lambda and \mu . Thus, the joint probability of the home team scoring x goals and the away team scoring y goals is a product of the two independent probabilities:

P\left( X_{i,j}=x,Y_{i,j}=y \right)=\frac{\lambda ^{x}\exp (-\lambda )}{x!}\frac{\mu ^{y}\exp (-\mu )}{y!}

while the generalized log-linear model for \lambda and \mu according to Kuonen [9] and Lee [10] is defined as: \log \left( \lambda  \right)=c^{\lambda }+a_{i}-d_{j}+h and \log \left( \mu  \right)=c^{\mu }+a_{j}-d_{i}, where a_{i},d_{i},h > 0 refers to attacking and defensive strengths and to home field advantage respectively. c^{\lambda } and c^{\mu } are correction factors which represent the means of goals scored during the season by home and away teams.

Assuming that C signifies the number of teams participating in a season and N stands for the number of matches played until now, the team strengths can be estimated by minimizing the negative log-likelihood function with respect to \lambda and \mu :


\begin{align}
  & L(a_{i},d_{i},h;\ i=1,..C)=-\log \prod\limits_{n=1}^{N}{\frac{\lambda _{n}^{x_{n}}\exp (-\lambda _{n})}{x_{n}!}\frac{\mu _{n}^{y_{n}}\exp (-\mu _{n})}{y_{n}!}} \\
 & =-\sum\limits_{n=1}^{N}{\log \left( \frac{\lambda _{n}^{x_{n}}\exp (-\lambda _{n})}{x_{n}!}\frac{\mu _{n}^{y_{n}}\exp (-\mu _{n})}{y_{n}!} \right)} \\ 
 & =\sum\limits_{n=1}^{N}{\lambda _{n}}+\sum\limits_{n=1}^{N}{\mu _{n}}-\left( \sum\limits_{n=1}^{N}{x_{n}\log \left( \lambda _{n} \right)} \right)-\left( \sum\limits_{n=1}^{N}{y_{n}\log \left( \mu _{n} \right)} \right)+\sum\limits_{n=1}^{N}{\log \left( x_{n}! \right)}+\sum\limits_{n=1}^{N}{\log \left( y_{n}! \right)} \\ 
\end{align}

Given that x_{n} and y_{n} are known, the team attacking and defensive strengths \left( a_{i},d_{i} \right) and home ground advantage \left( h \right) that minimize the negative log-likelihood can be estimated by Expectation Maximization:

\underset{a_{i},d_{i},h}{\mathop{\min }}\,L(a_{i},d_{i},h,i=1,..C)

Improvements for this model were suggested by Mark Dixon (statistician) and Stuart Coles.[11] They invented a correlation factor for low scores 0-0, 1-0, 0-1 and 1-1, where the independent Poisson model doesn't hold. Dimitris Karlis and Ioannis Ntzoufras [12] built a Time-Independent Skellam distribution model. Unlike the Poisson model that fits the distribution of scores, the Skellam model fits the difference between home and away scores.

Time-Dependent Markov Chain Monte Carlo[edit]

On the one hand, statistical models require a large number of observations to make an accurate estimation of its parameters. And when there are not enough observations available during a season (as is usually the situation), working with average statistics makes sense. On the other hand, it is well known that team skills change during the season, making model parameters time-dependent. Mark Dixon (statistician) and Coles [11] tried to solve this trade-off by assigning a larger weight to the latest match results. Rue and Salvesen [13] introduced a novel time-dependent rating method using the Markov Chain model.

They suggested modifying the generalized linear model above for \lambda and \mu :

\begin{align}
  & \log \left( \lambda  \right)=c^{\lambda }+a_{i}-d_{j}-\gamma \cdot \Delta _{i,j} \\ 
 & \log \left( \mu  \right)=c^{\mu }+a_{j}-d_{i}+\gamma \cdot \Delta _{i,j} \\ 
\end{align}

given that \Delta _{i,j}=\frac{\left( a_{i}-d_{j} \right)+\left( d_{i}-a_{j} \right)}{2} corresponds to the strength difference between teams i and j. The parameter \gamma >0 then represents the psychological effects caused by underestimation of the opposing teams’ strength.

According to the model, the attacking strength \left( a \right) of team A can be described by the standard equations of Brownian motion, B_{a,A}\left( t \right), for time t_{1}>t_{0}:

a_{A}^{t_{1}}=a_{A}^{t_{0}}+\left( B_{a,A}\left( t_{1}/\tau  \right)-B_{a,A}\left( t_{0}/\tau  \right) \right)\cdot \frac{\sigma _{a,A}}{\sqrt{1-\gamma \left( 1-{\gamma }/{2}\; \right)}}

where \tau and \sigma _{a,A}^{2} refer to the loss of memory rate and to the prior attack variance respectively.

This model is based on the assumption that:

{a_{A}^{t_{1}}}  | {a_{A}^{t_{0}}}\; \sim N\left( a_{A}^{t_{0}},\ \frac{t_{1}-t_{0}}{\tau }\sigma _{a,A}^{2} \right)

Assuming that three teams A, B and C are playing in the tournament and the matches are played in the following order: t_{0}: A-B; t_{0}: A-C; t_{1}: B-C, the joint probability density can be expressed as:

\begin{align}
  & P(a_{i},d_{i},\gamma ,\,\tau ;\ A,B,C)=P\left( \lambda _{A},t_{0} \right)\cdot P\left( \lambda _{B},t_{0} \right)\cdot P\left( \lambda _{C},t_{0} \right) \\ 
 & \times P\left( X_{A,B}=x,Y_{A,B}=y|\lambda _{A},\mu _{B},t_{0} \right)\cdot P\left( X_{A,C}=x,Y_{A,C}=y|\lambda _{A},\mu _{C},t_{0} \right) \\ 
 & \times P\left( \lambda _{A},t_{1}|\lambda _{A},t_{0} \right)\cdot P\left( \mu _{C},t_{1}|\mu _{C},t_{0} \right) \\ 
\end{align}

Since analytical estimation of the parameters is difficult in this case, the Monte Carlo method is applied to estimate the parameters of the model.

References[edit]

  1. ^ Stern Hal. (1995) Who's Number 1 in College Football?...And How Might We Decide? Chance, Summer, 7-14.
  2. ^ Constantinou, Anthony; Fenton, N. (2013). "Determining the level of ability of football teams by dynamic ratings based on the relative discrepancies in scores between adversaries.". Journal of Quantitative Analysis in Sports. 9 (1): 37–50. Retrieved 25 March 2014. 
  3. ^ Moroney M. J. (1956) Facts from figures. 3rd edition, Penguin, London.
  4. ^ Reep C. Benjamin B. (1968) Skill and chance in association football. Journal of the Royal Statistical Society, Series A, 131, 581-585.
  5. ^ Hill I.D. (1974), Association football and statistical inference. Applied statistics, 23, 203-208.
  6. ^ a b Maher M.J. (1982), Modelling Association Football scores. Statistica Neerlandica, 36, 109-118
  7. ^ Caurneya K.S. and Carron A.V. (1992) The home advantage in sports competitions: a literature review. Journal of Sport and Exercise Physiology, 14, 13-27.
  8. ^ Knorr-Held, Leonhard (1997) Dynamic Rating of Sports Teams. (REVISED 1999). Collaborative Research Center 386, Discussion Paper 98
  9. ^ a b Diego Kuonen (1996) Statistical Models for Knock-out Soccer Tournaments
  10. ^ Lee A. J. (1997) Modeling scores in Premier League: is Manchester United really the best. Chance, 10, 15-19
  11. ^ a b Mark J. Dixon and Coles S.G. (1997) Modeling Association Football Scores and Inefficiencies in the Football Betting Market, Applied Statistics, Volume 46, Issue 2, 265-280
  12. ^ Dimitris Karlis and Ioannis Ntzoufras (2007) Bayesian modelling of football outcomes: Using the Skellam’s distribution for the goal difference
  13. ^ Rue H. and Salvesen Ø. (1999) Predicting and retrospective analysis of soccer matches in a league. Technical Report. Norwegian University of Science and Technology, Trondheim.