RDSS 2.1 Factor Statistical Study

[Ted Craven]

Bill Lyster volunteered to gather statistics from RDSS2.1 Factors, some existing and some new ones, and present them in a way which might give some validation that the factors represent either an improvement over what we have now (e.g. the new CR+ Class Rating compared to the traditional CR Class Rating) or new factors which we have not seen before: the Rx Line Score and the Rx1 variant of it (which excludes the external 'Power Number' component of it, currently BRIS ProfitLine).

I can't keep straight (sorry Bill ) how many races are involved, because some races from 2015 did not involve Rx (only Rx1) and we added VDC only after a large number of races has already been captured. But it 2000+ races in all from 2015 and 2016 were input (Bill, feel free to clarify).

(Note: in future, this is what the Modeler is for - statistical including Impact Value studies).

Attached here in Excel spreadsheets are some of the results of what Bill has presented to me, broken down in various ways. The sheets show Hit Rate, average Mutuel and ROI by rank, and a subtotal by Top 4 combined ranks (also Rank 5 in the overall Summary sheet). Also a combined ROI% betting ALL Top 4 ranks (i.e. the loss involved)

The sheet named '...by distance...' is broken down by distance groupings. It has separate lines for races from Southern California (SA and DMR) in 2015 and for all other races (2016 GP, TAM - possibly other tracks).

The other Summary sheet is broken down by Winners/Maidens, then Field Size groupings, then Surface.

The overall takeaway is:

1. The new CR+ as a Class factor is several percentage points better than the old CR factor.

2. CSR Composite Speed Rating is a very predictive factor - 74% Top 4 and > 80% Top 5. CSR ranks > 5 hit less than 20% of the time (all distances, tracks, surfaces, field sizes)

3. Of the new Rx and Rx1 factors (Rx1 excludes the external 'Power Number' / ProfitLine), Rx has a slightly higher Hit Rate (74.8%), and Rx=1 (Rx rank 1) average mutuel than Rx1, though lower average mutuel from Ranks 2 and 3. The Top 4 combined ROI for Rx is slightly better than Rx1 (4% loss versus 6% loss). However for day-to-day use, the Rx and Rx1 line Scores can probably be used interchangeably (i.e. you don't need TwinSpires' ProfitLine number - except perhaps in Maiden races with FTS, where it is a definite help along with Workout Patterns and Tote Analysis).

4. VDC Top 4 Hit Rate is 73% - always reliable, although a fair number of tied ranks.

Notes: These initial studies are imperfect in that they often have multiple ties, which means that in someraces, there are more than 4 horses in the Top 4 ranks. VDC has more tied ranks than CSR which has more ties than Rx and Rx1. CR+ has almost no ties. Further studies need to be done (i.e. in the Modeler) to create proper Impact Values for these and other Factors.

However, I have found that there is almost NEVER a doubt in identifying which are the Top 4 ranked Rx or CR+ horses. Also - the BL/BL ranks/tiers are almost always unique, though not yet part of these studies.

FWIW, I almost never include CSR horses with rank > 5 as Primary Contenders - unless there are extenuating circumstances such as trouble or unsuitable distance/surface or other legitimate excuse in usually the last 2 lines, being the most highly weighted in the CSR formula (in which case there is a method of recalculating CSR without that line).

All in all - I found these stats sufficiently encouraging to create a new Analysis screen around them - the Rx Screen.

The Studies:

* RDSS Summary Stats-Rev2.xlsx
* RDSS Summary Stats - by Distance-Rev2.xlsx

If Bill Lyster is available to comment on these studies or correct any mis-statements I have made, I will appreciate his help with that while I am gone the rest of the month.

Again, thank you Bill - it was a monumental amount of hand data entry (though I know you love it ). In the future, the Modeler should make this and much more, much easier.

Enjoy!

Ted