The development of the Trakus electronic tracking technology has allowed horse players to look at a race in different ways than the traditional methods of charts and video replays. One of the biggest aspects of the technology is the collection of data relating to ground loss, fractional times, and more accurate measurements of horse separation at intervals around the track. Currently, only a few tracks utilize Trakus technology on a daily basis - Keeneland, Del Mar, Churchill Downs and Woodbine. Gulfstream Park has announced that they will add the technology for their 2011-2012 winter meet.
Earlier today I wrote a post on the Pacific Classic and Travers where I looked at the data from the Equibase and DRF Formulator charts in order to analyze and compare the performances of the horses in each race That data was based on the times recorded by the timer at the track, with extrapolations based on the chart-callers beaten lengths estimates. Almost all of the past performance data, speed figures, performance ratings, and pace figures we see in the horse racing world are based off of this data.
Trakus data, since it is based on radio frequencies and timing for each and every horse in the field (as opposed to estimates based off one horse) can provide a much clearer picture than estimated running positions of traditional charts. Since Del Mar utilizes Trakus technology, and provides the results on their website, I'd thought I would dive into the data for yesterday's Pacific Classic.
According to the Trakus data from Del Mar, Acclimation covered the 10 furlong Pacific Classic in 1:59.96, which is slightly faster than the track timer result of 2:00.16. In terms of distance traveled, rate of speed, and final times, the race broke down in the following manner:
|Dist. (ft)||+/-||Avg. MPH||Final Time||Beaten Lengths
||Time Diff.||Per Length|
|4||Game On Dude||6687||11||37.8||120.69||4.25*||0.730||0.172|
The Trakus information indicates that Acclimation traveled the shortest distance around the track, with the exception of Tres Borrachos who traveled 20 fewer feet during his journey. That's not a surprising revelation since we know that Acclimation was all alone on the lead and was not forced to go wide at any point in the race.
The rate of speed data is interesting for a couple reasons. First, Twirling Candy was actually the fastest horse in the race when you consider his final time and the amount of ground covered. I don't believe this suggests that he would have won had he hugged the rail the entire time (he would likely have encountered additional issues in that position), but it does suggest that the question of which horse is the "fastest" is somewhat relative. Additionally, ground loss and trip are critical components to winning. Something we know already but struggle to quantify.
The second thing I notice with the data is the relatively minor differences in speed over the course of the entire race. Setsuko finished ten lengths back of the winner, yet his average speed for the race was only a half a mile an hour slower. The difference between winning and losing a top level race is very small in terms of how fast a horse is actually running.
An additional piece of information that we can take from this has to do with the old adage that one length equals one fifth of a second on the race track. With Trakus data measuring both speed and distance we can test that theory to a much higher degree of accuracy than in the past. As it turns out, the old adage is remarkably consistent. In yesterday's Pacific Classic, one length equaled an average of 0.18 seconds, or slightly less than 1/5th.
Now let's compare the times from the DRF and Equibase charts to the Trakus charts. First are the internal splits for each horse in the race:
|4||Game On Dude||24.74||24.08||24.23||23.73||24.54|
|4||Game On Dude||24.19||24.24||24.07||23.72||24.47|
There is not a huge difference between the data from each system; both illustrate that Twirling Candy and Stately Victory ran the fastest final quarter miles in the Pacific Classic and that Acclimation was one of the next fastest finishers (although they differ on whether Quindici Man or Acclimation was faster in the final quarter mile). However, there is a difference in the fraction-to-fraction relationship for each horse, as is illustrated in the next chart, which is the comparison of the internal splits to measure whether a horse ran a quarter faster or slower than the previous one. Negative numbers indicate that the quarter of a mile was run faster than the previous.
|4||Game On Dude||-0.66||0.15||-0.50||0.81|
|4||Game On Dude||0.05||-0.17||-0.35||0.75|
These two charts see a bit more divergence than the first two. The biggest difference is the in times between the 1/2 mile and 3/4 mile marks. The DRF/Equibase charts suggest that almost every horse in the field ran SLOWER in the 3rd quarter than in the 2nd. The Trakus charts suggests the exact opposite - everyone went faster in their 3rd quarter split (and their 4th).
There is also a difference with respect to Acclamation. Trakus data suggests he ran slower in the 2nd quarter than the first. The Formulator data states that he went faster.
In the end it doesn't matter to many players the decimal point details of which horse ran what time in what quarter mile. But the Trakus data does shed some light on the variances that can occur when times are estimated based on one clocking and then extrapolated by running position and beaten lengths. Consider the calculation of speed or pace figures using beaten lengths; would they be more accurate if they were to simply on final time or individual clockings? And how much more accurate would they be? With regards to final time - the Formulator charts have Game On Dude running the Pacific Classic in 2:01.32, while Trakus clocked him at 2:00.69. That's a half second difference, or roughly 2+ lengths. If you are comparing times from track to track using figures based off chart data, that's a significant variance to overcome.
More than anything, the Trakus data shines a light on questions we should have with much of the data we are presented with in the horse racing world. It's not that the data is bad. Rather, technology exists to time individual horses in a much more precise manner than ever before, which increases our understanding of how races are run and won. A deeper understanding of the game is something that is important to handicappers, owners, trainers and jockeys.
As additional tracks begin to implement Trakus, or similar electronic tracking technology, the volume of data and the ability to analyze it in a broader context will likely increase. And that potentially puts us at the tip of a great expansion of horse racing knowledge.
*The Equibase chart called the distance between Stately Victor and Game On Dude to be ¼ length in their running line preview, but stated the difference as a "nose" in the result chart. This highlights the inconsistency of beaten lengths data and the difficulty reconciling past performance lines with results charts.