Open stanra opened 8 years ago
Aymeric Technique
I heard aymeric is using some basic statistics to guide his choices : he makes the score average for the x previous games for the two teams.
Could either be a pure average of goals scored, could also be first result average for the winner, score average for the score.
Could take into account the home/away in a naive way by putting higher weights on results in same configuration as now.
Last Ranking
Could use the last (=current) ranking of the teams in competition to predict the result. For games in the first matchday, the rankings are too variable, so they should use last seasons ranking.
For the score, a stupid way would be to use the average of goals scored and taken for each team, and average it.
Could chose a range that decides a draw
Example: Paris 1er - 64 for - 10 against - Lyon 4eme - 39 for - 19 against -
Winner is Paris. They score 2,3 g/m and Lyon take 0.8 g/m so Paris score 1.6=2 goals. Paris take 0.4g/m and lyon puts 2g.m so Paris take 1.2=1 goals. Prono : Paris 1 - 2 Lyon
Need to do first some 'dumb' predictors, that only use some simple statistics to predict the score.
They are fast to use, and can be quite efficient already. This goes in pair with some data exploration, and should give ideas for real predictors