29 March 2016. Just in time for the beginning of the final spurt of the 2015/16 Bundesliga season, the forecasts will be published from 29 March 2016 at 10 a.m. at www.sew.unisg.ch/soccer_analytics. Subsequently, the forecasts will be updated with the latest development two days after every round.
What will be forecast?
The main objective is to forecast the final league table of the Bundesliga. By way of a “by-product”, the match results of each individual day will be forecast a few days in advance. In order to reflect the uncertainty about the actual final league table, it will additionally be observed which teams will end up in a certain area of the league table with what probability.
For the 2015/16 season, and as expected, the researchers think that the race for first and second places is quite obvious. The direct relegation of Hannover 96 also seems to be pretty clear. In between, things are distinctly tighter. Thus six sides still have justified hopes of a Champions League qualification. The teams in places 3 to 8 have at least a 10 per cent chance of qualifying for the top league. In the relegation battle, there are another five teams besides Hannover which, with a probability of at least 20 per cent, will either be relegated immediately or after the relegation matches.
Moreover, the academics will compare the teams’ actual performance with what was to be expected before the season. There, you can see what a great surprise Berlin is. The pre-season forecast awarded Berlin a score of 28 points on the 27th match day. The actual number of points scored by the team, namely 48, shows that Hertha BSC is the positive surprise of the season. The negative side is represented by Wolfsburg and Hannover, which on the 27th match day have 14 and 12 fewer points against them, respectively, than was expected by the forecasting model before the season.
How does the forecast work?
For the forecast, the researchers draw on an extensive data basis of information including previous match results and style of play, the current members of the team, the environment and the match schedule of the current season. The forecasting model is produced by means of an econometric method which uses past seasons for filtering out information that has a demonstrable predictive efficiency for the result of football matches and the teams’ places in the final Bundesliga table.
This results in a forecasting model which, on the basis of the information taken into consideration, computes the probabilities of a home win, a draw and an away win. This will serve as a basis on which the expected score of a team from each match can be calculated. If the expected scores of these matches are added up, this will result in the forecast for the final league table. However, such forecasts are fraught with uncertainty since, of course, it is not always the favourites that win. The researchers take this uncertainty into account with the help of many thousands of simulations and are thus able to determine the probabilities of a certain season result for the individual sides.
Who is behind this?
This project is conducted by the Sports Economics research group led by Prof. Dr. Michael Lechner, Chair of Empirical Economic Research at the University of St.Gallen. This chair always makes an effort to apply the latest statistical methods, including the methods of “machine learning”, which have become increasingly popular in recent years when it comes to drawing up the best possible forecasts for uncertain events. An interest in the football Bundesliga and the new methods gave rise to the idea of producing a Bundesliga forecast and sharing it with the general public. In this way, a data environment was created which is well suited to testing the new kinds of statistical methods before these methods are then used for an analysis of issues that are relevant to economic policy.
Photo: fotolia.de/p!xel 66