Guest post written by AM
One of the things that distinguish soccer among other sports is the fluidity and dynamic nature of the game. It is a team game. The success or failure of the whole team has to be the collective effort of the 11 first team players and reserves. Players improve and dwindle, teams acquire players, teams sell players. A club can go bankrupt and be forced to sell its start players just to stay afloat (like the wonderful Lazio team of 1997 to 1999).
When confronted with this sort of dynamism in football, a bettor who chooses to rely on statistical data to make a betting decision on a match is going to have a tough time indeed. Soccer is not like horse racing when it comes to statistical prediction. A horse’s individual performance can be followed with better accuracy. How do you follow the performance of a team with over 22 players? It is surely a more daunting task.
Despite these challenges, a number of statistical programs have been developed in an attempt to increase a bettor’s chances in soccer betting. Some of these programs are simple, while others are downright complex. The aim of these statistical programs is to provide a ranking system which can place teams on a scale of performance. The ranking is based on past results, and so the strongest teams are assigned top rankings. Using this ranking system, a bettor can predict the outcome of a soccer match, usually in favour of the stronger teams.
Using statistical ranking can have some drawbacks. Some of these drawbacks are as follows:
1.   Ranks assigned to the teams do not differentiate between their attacking and defensive strengths. It is possible for a strongly attacking side that scores goals to have a leaky defence. This factor can come to play when two closely ranked teams square off against each other, producing surprise results.
2.   Ranks are accumulated averages which do not account for skill changes in football teams, for example if a team brings in a lethal striker mid-season.
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 method of statistical analysis gives teams ratings. Ratings usually take cognizance of factors such as attacking and defensive prowess, home advantages and player strengths.
The use of statistical analysis in soccer betting can be said to have originated in 1956 when Moroney first published his version of statistical analysis of soccer matches. Since then, they have been modified and have enjoyed patronage since the early 90s.
Several statistical methods of soccer betting analysis are now in use and are used to predict results in various competition types such as in round-robin or knockout competitions.
Some of the popular statistical methods are:
- Time Independent Least Squares Rating
- Time Independent Poisson Regression
- Time Independent Skellam Regression
- Time-Dependent Poisson Regression
- Time-Dependent Markov Chain
Of these methods, the Poisson regression and Markov Chain are the best performing, but they are also very complex to use. The Least Squares rating tends to give poor results.
I agree that using stats to predict results can be daunting but if you are willing to spend the time you can find patterns in goals scored, goals conceeded, average scorelines and outright results which can result in much more profitable bets.