1.
The model outputs home goals and away goals for soccer matches.
Sample is 3114 games played in the MLS and Finland.
Model mean error: 0.916
If I set every prediction to the league average home goals and away goals, so for MLS, home goals = 1.52 and away goals = 1.06
Mean Error = 0.943
So I'm barely ahead of the most primitive prediction method known to man. Astrology probably outperforms what I've made. Should I be as distressed as I am?
2.
And about mean error, am I right in thinking that if a method of prediction has a mean error higher than that indicated by the odds, after converting the moneyline and spread to a number of points/goals for each team, then the predictions will not make a profit? I was checking out the results at http://www.thepredictiontracker.com/ and it appears that almost no rating system consistently beats the opening line and the closing line is almost impossible to beat. Injuries are probably not accounted for in the rating system, that's one fault. Still, the oddsmakers are pretty damn good at what they do.