Hello,
The hockey seasons are already in play in Europe and the NHL season is starting to warm up.
I have a special interest in game total betting and I think the totals are easier to predict for hockey and soccer games since the scoring process is a Poisson process.
The scoring in hockey can be fairly enough estimated using the Poisson distribution, because the goals are rare events for most of the leagues, they happen one at a time and knowing the average per unit of time we could calculate the expected number of goals for 60 minutes.
I made a little study today on the games played in european major leagues (Czech Republic Extraliga, Austrian EHL, Switzerland league A, Germany DEL, Slovakia Extraliga, Finnish SM Liiga and Russian KHL).
There were 37 games played in these leagues and I estimated the game total taking into account only the offensive of both teams. I didn't care about the other's team defensive line.
I estimated the total of the game by summing the average goal scoring of the teams. E.g. team A has played 10 games and scored 30 goals wich means 3 goals/game and team B played 12 games an scored 24 golas wich means 2 goals/game; the expected total is 3+2 = 5 goals. Following the probabilities estimated by the Poisson distribution at 5 goals/game, the odds are 61.5% under 5.5.
I didn't use the exact estimation of odds, but instead I used the next rule: if the sum of average goals scored was equal or over 5.7 I predicted over 5.5 and bellow 5.7 I predicted under 5.5.
According to Poisson distribution, the odds are bigger for the under 5.5 when the expected average is less the 5.7 and bigger for the over when the expected average is more or equal 5.7.
Following these rules I got an impressive result: 24 bets of 37 were winners. That means a 64% strike rate at an 1.7 (decimal oods) average. The ROI was 14%.
I will digg more the next days and post the findings. If the strike rate remains over 60% I think it's something.