Hi,
First time on here, so hello to the hopeful hordes.
I have to say that, though I have used a relatively small data sample (testing for a week), I am nevertheless quite disappointed with the Football Predictor!
I would say that 90% of my predicted values have not been reflected in actual values taken approx 3 days prior and observed again 5 min before kickoff.
I am mindful of Gavin's reply to the Football Predictor post/thread some dozen or so below this one where he says that it is likely that the statistical base people are likely to adjust values closer to an event due to volatility but the older data (say 2 or 3 days pre) should supply adequate pointer information to odds movement towards True Value.
For instance, today's Premiership matches. The Predictor (or the statistical base?) was woefully off kilter in predicting the odds values for such as C Palace v Fulham, Brighton V Watford, Chelsea v Huddersfield, etc etc. Not one gave a profitable prediction.
If I was betting at £50 or £100 values, instead of testing, then I would have been sorely tried.
I average both statistics sites Superiority and Total Goals values for a finer input.
I am at a loss to understand why this is. As some wiseguy said, the greater the statistical base the greater the regression to the mean.
Still, ever optimistic, I will plough on.
Anyone any ideas?
Cheers