Question for data base guys

[markgoldie]

Recently, I've been looking at what appears to be something of a promising angle. This takes a somewhat contrarian view aimed at the overplay given to early speed.

The angle is to play S-types or P types whose last two speed numbers are higher than any last two of other contenders. Also, there must be 20 or more total Quirin points in the balance of the field, excluding horses with less than 5 Quirin points. Dirt races are preferred.

Another way of approaching the qualifications would be using Bris Prime Power numbers. This play would be any S or P type in the above pace profile who have a 2 or more point advantage over the field.

I have only anecdotal evidence that this may work. However, I have noticed a consistent wagering bias to speed types such that it appears many bettors will not play S types under most or all circumstances. In addition, there seems to be unusual and unwarranted betting action given to any horse who it is thought may prove to be the lone speed, irrespective to class and/or final speed numbers.

We know that there is a reasonable correlation between favortism and high Beyer speed relative to the field. We also know that there is a reasonable correlation to the top Bris Prime Power horse and betting favortism. But, anecdotally, I am not seeing the same degree of favortism (or favortism at all) in these late-running types. On the other hand, I am anecdotally seeing what appears to be a reasonably close win percentage.

At any rate, it would be interesting to see if some guys with large data bases could back-run this scenario without too much trouble. My gut says there could well be a straight up positive ROI in such situations.

[Sham]

Interesting premise...I can get pretty close to your parameters with HSH, so I took a swing. I don't believe I can isolate horses with the highest FT figure in both of the last two races...but I can look at rank for "Final Time,Average of the last two" = 1

First, I created a filter with the following restrictions:

Calendar Years 08-09
Fast dirt only
No MDN
No 3yo before May
Distance 5.5 to 9.0
Betting Interests 6+
QSP Total of horses with 5+ QSP >=20


Here are the "rank Final Time Avg Last Two=1" horses, without regard to running style
201-PubCh
----------------------------------------------------------------
WIN BETS
Field1 Field2 Starts Pays Pct $Net IV PIV HV
---------------------------------------------------------------
3/5 761 431 56.6 $1.75 4.30 1.05 1.52
6/5 1,860 728 39.1 $1.60 3.07 0.96 1.31
5/2 4,443 1,233 27.8 $1.69 2.24 1.01 1.26
9/2 2,743 506 18.4 $1.75 1.53 1.04 1.17
6 1,180 143 12.1 $1.62 1.04 0.97 1.01
9 835 94 11.3 $2.02 0.98 1.22 1.17
15 554 42 7.6 $2.02 0.66 1.17 1.04
24 240 11 4.6 $1.86 0.40 1.10 0.87
35 108 0.0 $0.00 0.00 0.00 0.00
Above 49 0.0 $0.00 0.00 0.00 0.00

Total 12,773 3,188 25.0 $1.70 2.04 1.01

Here are only those horses with QSP < 5
201-PubCh
----------------------------------------------------------------
WIN BETS
Field1 Field2 Starts Pays Pct $Net IV PIV HV
---------------------------------------------------------------
3/5 191 112 58.6 $1.84 4.65 1.10 1.61
6/5 516 209 40.5 $1.67 3.25 1.01 1.37
5/2 1,538 430 28.0 $1.73 2.31 1.02 1.28
9/2 995 177 17.8 $1.65 1.53 1.00 1.13
6 455 61 13.4 $1.80 1.18 1.07 1.12
9 335 34 10.1 $1.89 0.90 1.10 1.07
15 254 19 7.5 $1.96 0.68 1.16 1.04
24 117 6 5.1 $2.13 0.47 1.23 0.99
35 43 0.0 $0.00 0.00 0.00 0.00
Above 29 0.0 $0.00 0.00 0.00 0.00

Total 4,473 1,048 23.4 $1.72 1.99 1.03

From those numbers, it looks like there is some merit to your observations, although not a large difference. Lots of room for tweaking though.

I've done a fair bit of work trying to isolate races that will favor one running style over another...and the results have been a bit disappointing. One conclusion I've come to....the standard tools we use to describe running style are really lacking. Sure...you can be pretty confident that a horse with 8 QSP is a need-to-lead type. I can also find a number of need-to-lead horses on any racing day that have fewer than 5 QSP.

[Sham]

...One conclusion I've come to....the standard tools we use to describe running style are really lacking...

And now I see your post in the other thread where you state the same thing, only more eloquently. So, um, yeah! I agree!

[markgoldie]

Sham: Thanks for those two reports. Notice that you were able to break down the fig-rank in last two which is kind of what I had in mind and may be ballpark what I was after. Actually, I adjust figs to the class in which they were earned and if I had to point to one insight or achievement in my long handicapping career, that would be it. Never learned it in a book, least of all Beyer (whom I avidly read as they were published), since he said that the strength of a fig was never influenced by the class in which the horse competed. Suffice it to say that after many many years of doing this, the fact that horses tend to run faster in weaker races is settled law for me. I use a formula which I have explained on the forum many months ago (if you're interested, PM me for a reiteration).

At any rate, your other breakdown (if I'm reading this correctly) was for horses with less than 5 QSPs in a field where those with 5 or greater added to a total of 20 or more. But irrespective to running style, that would allow for E4's or less or E/P4's or less. And that's not what I was envisioning. More like S1 or S2 tops, as well as P1 or P2, because these are the runners that I think are routinely shunned at the mutuel windows.

Any way to break these two categories out from your data base? Thanks again.

[Sham]

At any rate, your other breakdown (if I'm reading this correctly) was for horses with less than 5 QSPs in a field where those with 5 or greater added to a total of 20 or more.
That's correct


But irrespective to running style, that would allow for E4's or less or E/P4's or less. And that's not what I was envisioning. More like S1 or S2 tops, as well as P1 or P2, because these are the runners that I think are routinely shunned at the mutuel windows.
Any way to break these two categories out from your data base? Thanks again.
Is the "E4, S1, etc" designation a BRIS rating? My data is all in HSH, using HDW data. HSH has a couple of ways to look at running style:

-a running style designation that's been in the program for as long as I can remember, which I believe is simply based on QSP's. (7-8=E, 5-6=P, etc)
-Cramer running style, which considers the call at which the horse took the lead (or within 1 length if I remember correctly). An "S" horse would be one who wasn't within 1 length until the wire.

If any of that matches what you're looking for, I can run some more tables.

[markgoldie]

Well, the number of QSP given to a horse doesn't exactly match his running style. That is, you could, for example have an E1 horse, which means essentially that the animal has little early speed but then fades. So that's not what I'm looking for.

Yes, lower QSP numbers may indicate a running style. For example, a horse with 1 QSP is more likely to be a P or S type runner than an E or an E/P.

But if it's not too much trouble, why not run the above study where the target horse's QSP is <3. That, at least, might get us closer to what we're looking for.

[Sham]

But if it's not too much trouble, why not run the above study where the target horse's QSP is <3. That, at least, might get us closer to what we're looking for.
201-PubCh-QSP > 3
----------------------------------------------------------------
WIN BETS
Field1 Field2 Starts Pays Pct $Net IV PIV HV
---------------------------------------------------------------
3/5 79 41 51.9 $1.61 4.18 0.98 1.43
6/5 235 90 38.3 $1.59 3.11 0.96 1.31
5/2 763 207 27.1 $1.66 2.26 0.99 1.24
9/2 520 88 16.9 $1.58 1.47 0.96 1.08
6 251 35 13.9 $1.88 1.25 1.11 1.17
9 174 19 10.9 $2.01 0.98 1.19 1.16
15 147 10 6.8 $1.70 0.62 1.06 0.93
24 64 5 7.8 $3.31 0.69 1.88 1.51
35 22 0.0 $0.00 0.00 0.00 0.00
Above 18 0.0 $0.00 0.00 0.00 0.00

Total 2,273 495 21.8 $1.70 1.87 1.00
Not much help there. Here's the table I probably should have posted from the start. This is all the "rank Final Time Avg Last Two=1" horses broken down by QSP:

202-ES
----------------------------------------------------------------
WIN BETS
Field1 Field2 Starts Pays Pct $Net IV PIV HV AvgOdds
---------------------------------------------------------------
0 1,135 232 20.4 $1.76 1.75 0.99 1.18 3.02
1 304 66 21.7 $1.75 1.94 0.99 1.21 2.77
2 834 190 22.8 $1.56 1.93 0.98 1.17 2.57
3 1,091 274 25.1 $1.81 2.13 1.06 1.29 2.51
4 1,108 274 24.7 $1.67 2.05 1.05 1.25 2.51
5 2,434 616 25.3 $1.71 2.07 1.03 1.25 2.38
6 2,234 585 26.2 $1.83 2.11 1.03 1.28 2.26
7 1,727 430 24.9 $1.56 2.00 0.95 1.16 2.16
8 1,906 521 27.3 $1.65 2.15 1.01 1.24 2.06

Total 12,773 3,188 25.0 $1.70 2.04 1.01
This is a breakdown of the same horses by Cramer Running Style (1=early, 6=late)
390-CRSE
----------------------------------------------------------------
WIN BETS
Field1 Field2 Starts Pays Pct $Net IV PIV HV AvgOdds
---------------------------------------------------------------
1 2,508 634 25.3 $1.57 2.06 0.98 1.19 2.22
2 1,472 413 28.1 $1.78 2.27 1.08 1.32 2.19
3 1,422 361 25.4 $1.72 2.07 1.00 1.23 2.26
4 3,607 872 24.2 $1.70 1.97 0.99 1.21 2.39
5 2,626 661 25.2 $1.79 2.07 1.06 1.28 2.49
6 1,138 247 21.7 $1.68 1.84 0.98 1.18 2.75

Total 12,773 3,188 25.0 $1.70 2.04 1.01
Noticeable difference for the Earliest horses. I added a column to each table for Average Odds. No question that the earlier runners are bet harder.

It seems to me that by starting with horses who are top for the last two final time figures, it's going to be difficult to show much more than a subtle difference, especially given the broad filter we're starting with.

Just a random thought...I think it would be interesting to look at your theory for poly tracks only, but in reverse. Seems the conventional wisdom is that poly favors late runners...which would open up the possibility that early runners are undervalued. (all of the above tables are dirt races only)

[markgoldie]

Very interesting results here and not what I would have imagined at all.

If we can assume that a great deal of wagering action is drawn by top Beyer numbers (and I think we can), my premise is that the prejudice towards early speed would tend to overlook these animals if they exhibited an off-the-pace running style. This study seems to confirm that fact- that is, as the QSP number goes down, the average odds goes up, as you note.

But why doesn't the win percentage hold more steadily? This would seem to indicate a complete exoneration of pace-speed handicappers vs. final-speed handicappers. Put another way, it indicates that no matter the animal's final-speed capability, the greater the lack of even tactical speed, the less chance of winning. But why?

Why, for example, if we take two horses who can run identical Beyer numbers of, let's say 85, is the horse designated as an S(4) [or possibly a P(4)] more likely to win than the horse designated S(0)? Clearly, neither horse will be a part of the pace scenario of the race. So neither animal should be affected by it. Both run the same final times, so we can assume that the S(0) closes faster. The speed-favoring biases of individual tracks should even out over the sample. So what's the theoretical advantge held by the S(0)??

I am at something of a loss to explain this. Any ideas?