Ground Loss at Keeneland
Keeneland opens Friday and once again handicappers will be trying to understand the nuisances of the synthetic surface where the races in the fall meet played out as the very antithesis of American dirt racing. Deep closers ruled the days as front runners routinely packed it in despite moderate fractions. One of the most unusual races was the Spinster where Summerly raced on the lead with Spun Sugar not far behind through pedestrian fractions of :25, :50.2, and 1:16.1. Both horses faded and finish 7th and 8th respectively.
I had no clue what to make of Keeneland so I decided to study the fall meet and utilize the data compiled from Trakus (some features available for free on their website). The actual distance run by every horse was recorded so we can measure ground loss with 100% accuracy. Ground loss is one of the fundamental components to trip handicapping and with these tools perhaps we could gain some insights for the upcoming spring meet. The results leave me more perplexed.
Ground Loss
If somehow you were able to divine which horse was going to cover the least ground over the course of a race and bet that horse blindly you would have lost badly. Of the 136 races run over the dirt track only 3 were won by horses who got the shortest trip. Luckily one of these three winners was a 47-1 bomb but the rate of return was still an abysmal -53%. None of the 28 horses at 5-1 or less and aided by the shortest trip won. On the other hand, horses who took the longest way around won 16 of 136 races and earned a 23% return. Included among these wide running winners was Patricia who looped a field of allowance sprinters to win by a neck at 10-1 despite covering 86 and 110 more feet than the second and third place finishers. She lost more than ten lengths by going seven wide on the turn still came up the winner.
Here is the data for ground loss. I assume eight feet equals one length.
| lengths lost | Dirt Sprints | Dirt Routes | ||||
| freq | wins | return | freq | wins | return | |
| 0 | 84 | 1 | -90% | 53 | 2 | 5% |
| 1 | 137 | 10 | -31% | 56 | 4 | -15% |
| 2 | 152 | 18 | 50% | 45 | 4 | -53% |
| 3 | 175 | 24 | -5% | 53 | 7 | 20% |
| 4 | 133 | 14 | 18% | 57 | 6 | -60% |
| 5 | 95 | 13 | -14% | 68 | 8 | 19% |
| 6 | 45 | 2 | -80% | 48 | 6 | -38% |
| 7 | 14 | 0 | -100% | 59 | 5 | -43% |
| 8 | 4 | 0 | -100% | 43 | 4 | 68% |
| 9 | 2 | 0 | -100% | 27 | 4 | 127% |
| >9 | 5 | 1 | 120% | 26 | 3 | -17% |
Post Positions
One thing that Trakus allows me to do is to evaluate the ground loss from post position. Unfortunately Keeneland might not be the ideal setting to calculate these numbers. Given the paucity of winners who ran near the rail, perhaps the rail was dead and thus the fastest way around the track might not have been the shortest way. Instead I’ll assume the rail was not dead and that the reason wide runners were more successful was because the nature of the racing surface favored closers who had to run wide to avoid traffic. (While I assume this I am more inclined to believe the rail was dead).
Moving one post position wider increases predicted distance traveled by approximately 2 feet (or ¼ length) for sprint races and 4 feet (½ length) for dirt races. A longer run into the first turn does reduce the post position effect.
| Distance | PP effect |
| 6.0 f | 2.66 ft |
| 6.5 | 2.02 |
| 7.0 | 1.98 |
| 7.28 | 1.81 |
| 8.5 | 4.19 |
| >8.5 | 4.12 |
To conclude, I still have no clue about Keeneland – and I am not sure how any of this is useful.
I suppose the track really is unbiased as publicized because drawing and running wide seem to have no negative effect on a runner’s chances.
Comments