Testarossa,

I tend to agree that many people either don't understand the statistics they are trying to use or make mistakes when trying to interpret them. Certainly the stats on days since last run or TAB number are totally useless in my opinion. The days since last run is too dependant on the type of race, what the previous run was, the individual horse, etc - trying to generalise across all races/horses will not work.

The biggest mistake by far is looking at a particular statistic on winners without comparing it to the non-winners. For example they see a statistic that 70% of winners have run in the last 21 days (I just made that one up!) and think this is a good indicator. The problem is that if 70% of losers also ran in the last 21 days then the statistic is absolutely useless.

Another common mistake is coming up with a number of statistical criteria with various percentages (eg. 50% of winners have Criteria A, 65% of winners have criteria B, etc) and then if they find a horse that satisfies all these they assume it has an even greater chance of winning. In actual fact the chance is no higher than the highest of the criteria (and this only applies in the long run anyway).

Another big problem is that all this type of analysis often just ends up selecting the favourite.

I do use statistics as part of my analysis but not as the sole basis. Also I do not try to "come up with the winner" for every race - instead I try to work out a fair price for the runners that I think have a chance and then look for value when placing my bets (ie. horses that are paying over my calculated price).