Richiebee –

You're right about the sample sizes. That's one nice thing about finding that the pattern strengths are similar for all three age groups. Even if you have a small sample for a given pattern for one age group, you can learn something by looking at the strength of the same pattern for the other age groups.

I think a bigger problem than sample size is that many (most?) horses don't fall within one of the defined ThoroPatterns, due to layoffs or distance or surface changes. In that respect, as TGJB has indicated, the ThoroPattern is only a starting point for handicapping.

Moreover, I think that by looking at which ThoroPatterns are strong, you can learn something about what's true and not true about the form cycles of horses. I think there are a lot of misconceptions floating around out there. I've certainly partaken of a few, and have been able to improve my results not so much by doing anything brilliant as by doing fewer stupid things.

With regard to finding "winning" patterns, the flippant answer would be that if you'll tell me how fast they're all going to run, I'll take my chances on being able to figure out who will win. A more serious answer is that my analysis so far has focused on who will run either a Pair or Top. I suspect that the data will get messier (due to sample size problems) when I look only at new Tops, but finding horses likely to run new Tops may be where the real value lies. Beyond that, ThoroPattern doesn't tell you how big a Top you might expect, which may affect whether a new Top will suffice to win the race.