Of frog legs and derby preps...

[derby1592]

A group of trainers tried an experiment, they gave a horse 1 prep and entered him in the derby, then they gave a horse 2 preps and entered him in the derby, and then they gave a horse 3 preps and entered him in the derby...

As Mark Twain did once say – \"Get your facts first, and then you can distort them as much as you please.\"

So I tried to do the best I could with the facts to try and see what they tell us about 2 or fewer preps vs 3 or more preps for the Derby.

I used the TG stat categories of top/pair/off/X and I looked at all the Derbies for 1997 to 2004. (I did not have any sheets before 1997.)

Note: I included Aptitude and Real Quiet in the \"3 preps or more\" group and I left out a few horses that did not have figures to establish their pre-derby top.

The sample size is still small for the 2 or fewer preps but the sample size is pretty large for the other group.

Here it is:

First for the \"3 or more preps\" group:

New top: 5%
Pair: 25%
Off: 29%
X: 41%

And, now ...drum roll please... here are the results for the \"less than 3 preps\" group:

New top: 17%
Pair: 22%
Off: 22%
X: 39%

I will let you all draw your own conclusions and I am sure both camps can \"distort\" or rationalize things in either direction but I would have a hard time making a case that 2 preps is a bad thing, at least based on these \"facts.\" Given that the prevailing wisdom is that the less than 3 preps is bad, I would think that you will be getting some built in value with any horse that has 2 preps or less going into the Derby.

Cheers.

Chris

[TGJB]

Chris-- how big was each sample, and which ones did you use and not use for the 2 preps or less?

[mikemd]

isn\'t this what you would expect?  by nature, lighter raced horses have more upside (in their next race) versus horses that have raced more.

although they have more \"upside\", it still doesn\'t mean that they are equal to the more established horses.

plus, the sample size is not big enough.

Chris,

That\'s great data, the right way to think about it, and certainly evidence that 1-2 preps is not a problem. Thanks.

I have one question.

Did you include the imports?

If you did, I think the data could be heavily biased in favor of the 1-2 prep horses.

1. We do not know if the quality of speed figures for foreign races is as good as for US races.

2. Foreign horses could be improving their figures a much higher percentage of the time because of differences in drug laws overseas relative to the US.

3. The studies I have heard about indicated that foreign horses (all, not just 3YOs) do in fact improve their figures (or performance) 1st time they run in the US much more often than the average horse (and then tend to bounce) No explanation given.

4. The foreign horses usually have fewer overall starts and would therefore be much more likely to improve.  

The other issue is that we don\'t know if the 2 pools are identical. So we can\'t tell if the horses are outperforming or underperforming \"expectations\" even if they are improving more often.  

Unfortunately, the only way to determine \"performance relative to expectations\" without getting subjective is profit/loss. However, we have all agreed that the sample size is too small to draw any firm conclusions despite the fact that the evidence so far suggests that they are underperforming very significantly.  

Just as an example:

If the pool of high quality horses that had 3 or more preps contained more horses that had already reached a figure of 1 or 2, their chances of improving further would probably be lower than the chances of a pool of high quality horses with two preps whose best figure so far was only a 7 because they were more lightly raced.

There are other possible examples of pool differences that would impact % improvement without indicating whether horses were underperforming or overperforming expectations.  

That\'s why even though I think the figures are a really useful tool in this analysis, the key is not just new tops, faster figures etc..

The key is performance vs. expectations or ROI (return on investment).

It doesn\'t do me much good if I find a pool of horses that improves more often than another if that expected improvement is built into the odds and/or my expectations to an even greater degree than is actually achieved on the track.

Thanks again.



Post Edited (03-30-05 20:46)

[derby1592]

Sample size for 2-or-less-preps was 23 (I did not include Worldly Manner since I knew he had a 3yo Dubai prep but there was no fig for it - I think he probably either paired or ran a new top in the Derby so that would make the stats even stronger if included).

Sample size for more-than-2-preps was 107.

I did this manually and could easily have made a mistake. Anyone else can repeat the process since all the sheets are available in the Derby archives.

At the highest level you could say that about 40% of the 2-prep horses ran a \"good\" race in the Derby (within 1 pt of top or better) while about 30% of the more-than-2-prep horses ran a \"good\" race.

If you start splitting it out in any other ways you get less than 20 in the sample, which is pretty meaningless but I think the  30% good races for the 3-or-more types is probably pretty solid with the 107 sample size and,even with the small sample size, the 40% good for the 2-prep horses would probably lead you to conclude with decent confidence that the 3-prep horses do NOT do better than the 2-prep horses if you used a standard, statistical test (I am not going to do that but someone else is welcome to do so.)

Cheers.

Chris

P.S. One final note - for those of you thinking of betting the Dubai horse in this year\'s Derby, I could find 4 horses in the sample that I know came from Dubai and 3 of the 4 ran good races in the Derby (if you assume that Wordly Manner paired or ran a new top, which I am guessing he did). Only Express Tour ran a poor race. So I would be careful about discounting the Dubai horse if it is fast relative to the field. Certainly not conclusive but definitely no reason to view them negatively with regards to running a good race.

Were the imports included in the sample?

[jimbo66]

Chris,

I am sorry, but your figures for the two prep horses since 1997 don\'t match the sheets that Jerry posted yesterday.  No way that 39% of those horses paired up or ran tops.  

Anybody else besides Chris look at the sheets Jerry posted?

[mikemd]

i did a quick look at the sheets.  my count could be off 1 plus/minus, but no impact on the conclusion.

i had 8/22 horses with 2 or fewer starts as a 3yo run either lifetime tops or an equal to a lifetime top. (36%)

i had 4 of these 22 hit the board. (~18%)
 - lion heart 5/1
 - proud accolade 23/1
 - victory gallop 15/1
 - peace rules 7/1

i had 24/106 others run either a lifetime top or an equal to a lifetime top. (23%)

i had 20/108 others hit the board. (~18%)

i had 20/22 past derby winners ran a lifetime top or an equal to a lifetime top win the derby.

like i said, i\'m 1 off here or there (i did this on the back of an notecard)

my conclusion: although horses with 2 or fewer starts as a 3yo are more likely to run new tops, they are no more likely to hit the board.  however, it must be noted that the sample size for lightly raced horses is way too small.  you really can\'t draw firm conclusions from it.  but when is the last time you had a 95% confidence level on a race?

i remember chris doing some nice work on derby winners last year or the year before.  it would be worth another read.

[HP]

I looked at the sheets and...if the TIMING is good, and the last prep comes three weeks before the Derby, the horse has a shot to run well.  The further back from three weeks that last prep is, the worse the result...

It looks to me like two preps, properly timed, could do the trick.  I certainly wouldn\'t throw a horse out that had a properly timed two prep campaign.

I know I keep raising the same issue, but I think it is highly likely that the foreign imports are heavily biasing the results for 1-2 preps for the reasons described earlier in the thread.

At a minimum, I think it makes sense to note that it could be an issue and look at the stats with and without the foreign imports before coming to a firm conclusion.  

I also have a possible alternative to Jerry\'s suggestion that we look at \"in the money finishes\" instead of profit/losses for all the horses with 1-2 preps. I think it was a good idea, but one possible problem is that many of the horses could be in \"just good enogh shape to hit the board\", but not win.

We might eliminate all the horses that went off at odds of greater than \"X\" (maybe 15-1?). Then we could eliminate all the horrible horses that just happened to have 1 or 2 preps from negatively distorting the results for the sample as a whole. We could then focus just on horses that seemed to have some shot coming in and see how they did \"relative to odds\". That would remove subjectiveness from the study and eliminate any biases in the samples like foreign imports or more lightly raced horses that haven\'t shown as much having a greater probability of improving.

Ex 1. - if a horse went off at 7-2 and finished 9th that would be bad.

Ex 2. - if a horse went off at 15-1 and finshed 2nd, that would be good even though he didn\'t win.

Ex 3. - if a horse went off at 6-1 and he finished 3rd-4th that would be neutral.

Unfortunately, I\'m just not sure how we can get a list of all the horses with 1-2 preps going all the way back to 1937 and what their odds and finish positions were.

However, if we had that list, we could certainly get a better feel for whether these horses were underperforming or overperforming handicapper\'s \"expectations\" (which IMHO is more important than which group is improving more).



Post Edited (03-31-05 09:40)

[TGJB]

I took a very fast look at the imports for the time period Chris tested (97 on) and could have missed some, but it looks like the foreign horses don\'t affect the study too much one way or the other, although it could depend on what you consider a \"foreign horse\". Of the 6 I found, 4 had run in this country before the Derby (Castledale, Express Tour, Indian Express, Johannesburg), only Castle Gandalfo and China Visit had not. Don\'t know about foreign horses with more than 2 preps.

By foreign I meant horses coming off a foreign prep. I\'ll take a look at it when I have some time.

[TGJB]

Jimbo-- I just did it fast, same way that Chris did (throwing out only Worldly Manner), and got the same 4 new tops/5 pairs out of 23 stat. Keep in mind that the definition of new top is more than a point better, and pair is within a point either way of the previous top-- as it is for the ones with 3 or more preps.