BC Results Using TG Figures

[JimP]

Because I’m interested in such things, I decided to take a look at how some single variant system approaches to play selections would have worked out for the 11 BC races using the TG figures. (Please note that I would never recommend approaching the handicapping of a race or a series of races using a single factor, but I thought it would be an interesting exercise to look at it.)

I looked at the following factors:
1. Improved or paired figures between last two races,
2. New top in last race,
3. Paired top in last race,
4. Best TG figure in last race,
5. Best TG figure for this year,
6. Best TG figure for career,
7. Best TG figure for career at distance and surface.

I looked at the number of wins, places, and shows that would have been produced by each factor. And finally I computed a Win Impact Value The ratio of the actual winners to the expected number of winners) for each play.

Here are a few of the more interesting results:
a. Using the “improved/paired figures in last two races” produced 9 wins, 8 places, and 4 shows with a Win Impact Value of 1.52. That’s 9 winners from 11 races with this single factor.
b. Using the “new top in last” produced 3 wins, 2 places, and 3 shows with a Win Impact Value of 0.91. (If you exclude the 3 juvenile races the Win Impact Value rises to 1.33 for this factor.)
c. Using the “paired top in last” produced 4 wins, 4 places, and 1 show for a Win Impact Value of 1.57. (If you exclude the 3 juvenile races the Win Impact Value rises to 2.00 for this factor.)
d. A combination of “new top” and “paired top” produced 7 wins, 6 places, and 4 shows with a Win Impact Value of 1.43.  
e. All of the other factors produced very poor results as a single variant selection system.

Related observations:
a. The “improved/paired figures in last two races” missed only 2 winners, Indian Blessing in the Juvenile Fillies and the English Channel in the Turf. It’s interesting that Indian Blessing only had 2 starts. That would seem to make the failure of the factor for her of much less significance. English Channel experienced trouble in his last race, so that calls into question whether the figure he earned might have actually been better. Perhaps without the trouble he encountered he would have shown an improving line as well.
b. The combined “new top” and “paired top” missed only 4 of the 11 winners, Indian Blessing in the Juvenile Fillies, English Channel in the Turf, Kip Deville in the Mile, and Ginger Punch in the Distaff. See comments immediately above regarding the questions around the Indian Blessing and English Channel figures. It’s interesting that while Kip Deville and Ginger Punch did not better or equal their prior tops in their last race coming into the BC, the figure they each earned in their prior race was actually their 2nd best figure of their career. So both were actually very close to achieving their prior best and qualifying under this factor.  

Here are the race by race details:
F&M Sprint - Maryfield ($18.00) - Last paired prior top, improved by 1 point
             from prior race      
Juv Turf - Nownownow ($27.20) - Last was new top, improved by 1 point from prior
           race         
Dirt Mile - Corinthian ($9.40) - Last paired prior top, improved by 5 points from
            prior race       
Juv Fillies - Indian Blessing ($5.40) - Only 2 starts for career, first at sprint
              and second at route, won both in hand, the figures degraded between
              the two prior races but both figures were possibly not the best she
              could have earned in each of those races with a full effort
Juv - War Pass ($6.40) - Last paired prior top, which was in the prior race,
      only 3 starts for career       
F&M Turf - Lahudood ($25.40) - Last was new top, improved by 1.75 points from
           prior race         
Sprint - Midnight Lute ($7.00) - Last was new top, improved by 7 points from
         prior race         
Mile - Kip Deville ($18.40) - Last was 2nd best career figure (by 1 point),
       improved by 1.75 points from prior race      
Distaff - Ginger Punch ($11.00) - Last was 2nd best career figure (by 1.75
          points), paired prior race which was a career top for a route race   
Turf - English Channel ($8.00) – Last was 2 points off from prior race, trouble
       in last makes figure questionable, prior race was 2nd best career figure
       (only .25 point off career top)      
Classic - Curlin ($10.80) - Last paired prior top, improved by .75 point from
          prior and prior was 2nd best career figure (only .50 point off career
          top)       

Again: This is not how I handicap, so please don\'t point out the fallacies in using a single factor to handicap a race. I understnad that point completely. But I would be interested in hearing what others conclude from looking at these results. Admittedly from a very small sample.

[girly]

Wow! Very interesting stuff! Of course, bouncing happens, but what\'s the harm of doing some win/place bets based on those variants- Now I\'m going to have to study the sheets harder. I got the Seminar and I thought it was great as a guide. Honestly, all those figures just start swimming in my head- I actually draw on the sheets. I\'m never going to be a math wiz. But I\'m grateful that your curiosity got you looking at the information and thank you for the post.
   In the end though, I tend to win when I just like a horse, but intuition is helped mightily by knowledge.

[stillinger]

girly Wrote:
-------------------------------------------------------
I tend to win when I just
> like a horse, but intuition is helped mightily by
> knowledge.
Girly,
I am going to be presumptuous and ask you to remember that I told you this would happen if you just followed along. Makes it like Multiple Choice, doesn\'t it? And you can always pick, other.
skip
Good advice you got about going to the track if and when you can, as well.

[girly]

yeah, but I live in upstate NY and the only time I go is during Saratoga summer. But then, Saratoga is one of the most beautiful tracks in the country or so I hear-

[stillinger]

girly Wrote:
-------------------------------------------------------
> yeah, but I live in upstate NY and the only time I
> go is during Saratoga summer. But then, Saratoga
> is one of the most beautiful tracks in the country
> or so I hear-
Girly-
Tell me you at least watch recorded reruns every day.
If you do, the sheets will present a picture that is crystal.
And with that feminine intuition !! Good God.
Notice the boys start thinkin\' about Dom, and whiz, and hookers!!
They might be distracted. I wanted to move to Saratoga Springs,
but when I was on vacation, my wife bought a new house, so not for
a while, I guess
skip

[girly]

I will try to watch re-runs sometimes, but I like to think I have a balanced life and have so many other things to do all the time. Enjoy your new house!

[inkyredbone]

An important fact you have not provided is how many horses qualified as \"improved/paired figures in last two races”.  The system \"bet on horses with four legs\" gave all 11 winners.....

[JimP]

Ok, I posted this and went away to my real life, expecting to come back in 24 hours and read all kinds of interesting conclusions and observations. But alas, there seems to be a dearth of responses. Does that mean that no one,other than me and girly, finds this stuff interesting?

Girly, mentioned \"bounce\". And that is one thing I was thinking about as I assembled the data. Since I don\'t know what figures TG assigned to the BC runners, I can\'t say whether any of these win, place, and show horses \"bounced\". Maybe they \"bounced\" and still managed to win or hit the board. Or maybe they didn\'t bounce. Irrespective of the \"bounce\" possibility, it appears that improvement between the last two races before the BC was a strong indicator of an impending good performance in the BC. And it appears that even if the last race was a \"new top\" or a \"paired top\" there was no negative indication for the coming performance. At least that is how I read it regarding the finish position performance. Maybe someone at TG can tell us whether these winners \"bounced\". All I can see is that they won at a very high rate. This time. And I can see that the win rate was strongly correlated to the performance in their last race prior to the BC.  

I was expecting someone might mention ROI, since all I mentioned in my post was Win Impact Value. I also looked at the ROI on these factors, but decided that I would exclude that analysis from the post for the sake of simplicity.

[JimP]

Inkyredbone, I assumed that everyone would understand that Impact Value was based on that. When the Impact Value is greater than 1.00 it means more horses won than would be expected to win in a random sample of the population. Therefore the selection system that you postulated (i.e., \"horses with 4 legs\") would have an Impact Value of 1.00 and therefore not be of much interest. As you can see from the data I presented, the \"improved last race\" selection and the \"new or paired top in last\" did much better than 1.00. The other selection systems that I looked at, such as \"career best TG figure\", all had Impact Values less than 1.00. In other words, they were negative indicators for this population. I don\'t find that too surprising. It\'s just common sense (whatever that has to do with handicapping) that recent performance would be more significant than some long ago career best. I was somewhat surprised (ok, let\'s say very surprised) at how strong the Impact Values are for the \"improved\" and \"new/paired\" last race systems. It\'s a small sample. But still, at least a curiosity to me.

[stillinger]

JimP Wrote:
-------------------------------------------------------
> Ok, I posted this and went away to my real life,
> expecting to come back in 24 hours and read all
> kinds of interesting conclusions and observations.
> But alas, there seems to be a dearth of responses.
> Does that mean that no one,other than me and
> girly, finds this stuff interesting?
>
> Girly, mentioned \"bounce\". And that is one thing I
> was thinking about as I assembled the data. Since
> I don\'t know what figures TG assigned to the BC
> runners, I can\'t say whether any of these win,
> place, and show horses \"bounced\". Maybe they
> \"bounced\" and still managed to win or hit the
> board. Or maybe they didn\'t bounce. Irrespective
> of the \"bounce\" possibility, it appears that
> improvement between the last two races before the
> BC was a strong indicator of an impending good
> performance in the BC. And it appears that even if
> the last race was a \"new top\" or a \"paired top\"
> there was no negative indication for the coming
> performance. At least that is how I read it
> regarding the finish position performance. Maybe
> someone at TG can tell us whether these winners
> \"bounced\". All I can see is that they won at a
> very high rate. This time. And I can see that the
> win rate was strongly correlated to the
> performance in their last race prior to the BC.  
>
> I was expecting someone might mention ROI, since
> all I mentioned in my post was Win Impact Value. I
> also looked at the ROI on these factors, but
> decided that I would exclude that analysis from
> the post for the sake of simplicity.

I would like to say something interesting but this sounds complicated.
I make no conculsions, just stay open and play.

[BitPlayer]

Jim -

Your data is consistent with what the ThoroPattern data shows:  that a Paired Top is most likely to produce a good next-out performance, and a New Top is next best.

As you indicate, the obvious problem with an analysis like yours is that the sample size is so small.  I thought about asking about ROI, but small ROI samples are even more susceptible to being skewed by a single large payout.  Did you calculate Hit-the-Board impact values?

[JimP]

Bitplayer,
It is true that these results for the “new top” and “paired top” selections align well with the ThoroPattern trends. However what is interesting about these results is that the “improved last race” selections do even better than the “new top” and “paired top” selections. At least for this sample, a simple improvement in the last race seems to have more impact than reaching a new top or pairing a prior top in that last race.

I did look at the The Win-Place-Show Impact Values. They were not quite as strong as the Win Impact Values. The W-P-S Impact Value for the “improved last race” selection factor was 1.18 and the W-P-S Impact value for the “combined new top plus paired top” selection factor was only 0.96.

[bobphilo]

Jim P, well done. I just saw your post because; I too was back in my “real life”. This is interesting information. It shows that one has to be very careful in predicting bounces since a top in the last race can be a very positive factor rather than a liability. Of course, as you point out, this is only one of several variables to consider but this is an important one.

I had my best BC yet, due in part to betting a lot of horses coming off of tops and seeing this as a generally positive sign, unless there was some OTHER ADDITIONAL reason to expect them to bounce. I know that the BC races are a small sample but I\"ve also found this to be true in general and am glad to see some additional research to support it. Thanks.

Bob

[JimP]

Thanks.

What would be really interesting is to see how these factors would have performed with a bigger sample. Maybe somebody out there has the BC sheets for the last several years and would care to undertake a muli-year analysis.

And it will be interesting to see how many, if any, of the winners this year actually bounced on the TG figures even though they won the race. That\'s one of the elusive things about playing for bounces. A bounce can still produce a winner. If TG publishes the figures they assigned to the BC winners this year, we\'ll know whether there were any bounces involved this year.

[fkach]

I also seem to do quite well betting horses coming off tops UNLESS I have a very good reason to think the figure didn\'t reflect the horse\'s actual ability well (bias aided, especially easy trip, surface preference et..)