You've probably noticed that one of the rules appearing in many handicapping angles or systems states that you should eliminate any horse that won its last race. This is usually predicated on the assumption that a horse that won its last race will likely be overbet today, making it an underlay./p>
However, there may be more to it than that. Think about it. On the one hand, most books tell you that the horse's last race is usually the best indicator of how it may perform today. On the other hand, it seems that horses rarely win two races in row.
In order to look further into this, I checked the past performances of all horses that were entered at 35 different tracks on Saturday, 8/9/2014. I limited the check to one day to eliminate the possibility of a horse being scratched one day and running the next, in which case the same set of past performance lines would appear twice. Still, on just this one day, 2,559 horses were entered in various races.
I checked the overall lifetime performances of these horses and found they accumulated 5,973 wins in 41,051 races. That amounts to a 14.6% win rate or about 1 out of every 7 races. I also checked their performances in the current year, and found 1,677 wins in 12,040 races or 13.9%, still about 1 out of every 7 races. This seems to make sense. After all, in a field of 6 or 8 horses, only one horse can win. However, it also amounts to only about 2.3 career wins per horse.
The really striking result, however, applies to winning two races in a row. Of the 2,559 horses that were entered on that one Saturday, only 311 had EVER won two in a row in the races shown in their past performance lines. That's just over 12% or about 1 out of 8 horses, and it refers to having done it just ONCE in the past year or so, not a regular occurrence.
So it appears that, with the exception of stakes races, where only the best horses meet, betting on a horse that won its last race is a losing proposition. I verified this by checking the results of my own Key Race program, that finds horses coming out of a race in which 3 or more horses have gone on to win their next race. Overall, ten months data at 20 tracks showed a 26% win rate and a 16% profit. But if I eliminated bets on horses that actually won the key race, the results jumped to a 29% win rate and 46% profit! So, it definitely paid to avoid a horse that won its last race!