TrackMaster Variants (Evil or Misunderstood?)

[Ted Craven]

These thoughts were prompted by comments from Alydar_David and Ramon201 on this thread:

Originally Posted by alydar_David

While Val isn't dead I think it's seriously wounded. The program doesn't perform as well as it did prior to the Trackmaster change in February. I've handicapped thousands of races before and after the change and there is a discernable difference.

It's like blasting away with a shotgun now. Before it was like taking careful aim with a sniper's rifle.

One way to offset this is to box like crazy whenever there's value. I've noticed many four-horse-box trifecta hits, and some outrageous five-horse-box superfecta scores. Unfortunately, that approach is not for everyone.

The older manual programs should still work as well as ever. They still use 1/5 of a second. I'm not exactly sure which programs have suffered from the change. Certainly, Ted would know.

I know that Val2, Val3c and all versions of Speculator up to Spec150 are affected by TrackMaster/Equibase since Feb 1, 2006 now stating Daily Track Variants (DTV) in 10ths rather than 5ths (that is, 10ths normalized to 8 furlongs). I can't say about immediately previous programs like Synthesis and PaceLauncher which also used TrackMaster data files.

Doc always used only HALF the DTV when adjusting the raw pacelines from TM. This is consistent with his frequent observations that a major component of the variances in final times compared to their 3 year (or periodic) 'pars' was due to the effect of the Match Up in a given race (and also due to calibre of competition on any given day), and much less due to daily changes in the speed of the track surface. But when TM decreased the unit of measure of DTV from 2/10ths (1/5th) to 1/10 seconds this had the effect of causing the programs using their data to use the FULL daily variant instead of the historical HALF variant, thus mooting the foregoing discussion.

This doesn't mean much when DTVs are lower, say -7 to +7, but when you get wild variants like +30 or -27 (like the winner of BEL0610 race 13 in its last paceline, discussed elsewhere recently), adding or subtracting such large numbers to the raw call times can adjust a winner right out of contention, plus make the speed ratings look very bizarre compared to final time.

I'd like to make clear that the ONLY data from TrackMaster used to modify the raw call times is the Daily Track Variant. NOT the Speed Ratings. The Speed Ratings are TM's equalizing measure of Final Time - a guideline for selecting representative pacelines for today's matchup. They have NO effect on the velocities calculated from Equalized, Normalized, Adjusted times, and NO effect on subsequent factors. If you concentrate on mitigating the effects of variants gone wild, and use F6 in Validator or the Perceptor I Total ranking screen in Speculator Past Performances (plus common sense) to select pacelines, I think you end up close to the territory which Ramon discusses, and this seems right, given the best of what we understand Doc intended. And you don't have to rely as much on TM's Speed Ratings to select pacelines, if you don't want to.

Unfortunately, in Validator and previous (excepting ValM), there is no way to manually modify the DTV supplied by TrackMaster. In Speculator you can use the manual entry screen after you have selected some pacelines. As Ramon does, you could take the above mentioned -27 variant and make it be only -9 by entering +18 (-27 +18 = -9). And you could make that DTV = 0 by entering +27 (which puts the horse on top by a mile), like Richie has been experimenting with. Some will recall around February 1 we were discussing the relative merits of using the new (imposed by TM) FULL variant compared to the historical HALF variant used by the programs. The results of some tests were tentatively conclusive, so I released Spec150 with a default of the FULL variant, but made a setting whereby you could reinstate the historical HALF variant (by renaming the file _TVHALF.TXT to TVHALF.TXT, see the Spec150 Release Notes PDF). This was one of the main reasons for even doing Spec150 (along with some bug fixes).

Three things occurred on Feb 1, 2006 which affect Sartin Programs using TrackMaster data:

1. TM updated the par tables they use to calculate DTV and Speed Ratings. They have done this annually since forever.

2. They changed the scale of the Speed Ratings (forced by owner Equibase) from 30 - 100 to 0 - 130 to be in conformance with 'international' (and it seems, Beyer-like scales)

3. The Daily Track Variant (as well as Speed Rating), which is a function of a day's averaged, normalized final times for a given track/surface compared to the par they maintain for that track/surface - had its unit of measure changed from 5ths to 10ths.


Since we know TM's SRs don't affect our numbers, I believe points #1 and #3 above have the biggest impact on changed readouts, with the principal culprit being #1, the new annual pars, and the accomplice being #3 with its effect on how much variant our numbers are now subject to. But I agree with Ramon, that for now, a redoubled attention to the variants being accepted in the calculations is still a source of good mutuels. It just seems there must now be more work required to keep them realistic, in some cases - and paying close attention to what you're doing is never a bad thing.

In my work on the RDSS Windows software, I am now concentrating on the 'Adjustments' module which, when using TM data, permits you to use whatever percentage of DTV you want (e.g. 100%, 50%, 0%), put a minimum/maximum cap (e.g. like Ramon's -9 / +9); also, enter and store regular shipper adjustments from one track to another, one surface to another; use the historical Val/Spec Universal par table for distance adjustments, or DRF 3 year best times, or commercial pars such as HSH, Cynthia, or PaceFigures, or your own. Also change how DTV is applied, from simply 1/10th second per point to 1/10th second normalized to 8 furlongs, or .0125 seconds per point per furlong, which is how TM calculates them. (Amongst other things, these are my current excuses for taking so long to deliver this thing!)

I don't know whether TrackMaster data is becoming too 'dirty' or 'sloppy' to use or whether their last par upgrade had a lot of defects, or whether BRIS, or DRF or HDW data would be better. All these questions can be answered. I know I can't simplistically go to TrackMaster and say: "your data sucks - I lost the Derby, the Preakness and the Belmont for the first time in 10 years." Some people won... (and I'd need to give them a bit more detail than 'sucks'). It reminds me more than ever that we need to continue to forge into the future.

All thoughts welcomed!

Ted