Wed Sar 3rd - interesting jockey change

[HP]

[Michael D.]

[HP]

[TGJB]

Those interested in rebates.

[kev]

toocool505@aol.com  

Thanks

[Alydar in California]

Catalin:

There is no disagreement here. We must look elsewhere. In the case of a turf horse again running on turf, do you adjust your percentages based on late speed? Why? If not, has something changed?

[Catalin]

Alydar:

To me, R.O.I. as it is being employed here is a pretty meaningless concept in and of itself. It is a compound statistic dependent upon how the crowd bets a race (in addition to the win %).  If you employ an odds line for each race you\'re capturing the effect of the crowd by passing any horse below your estimate of fair value.

To know that a particular jockey has a positive ROI in a given situation is only useful if you think that your estimate of  fair value in that situation is no better than the public\'s, in which case you are really just second guessing the probablity estimates you\'ve previously assigned to the race.  I hope that made sense.


Re final fraction adjustments to turf figures, the answer is no I do not.  Nothing has changed in my mind.  I just have not found a satisfcatory way to do it.  I think I might have the math worked out, something along the lines of a sustained pace figure adjusted for trips and weight.  The problem is that it\'s too time consuming to do it for every race (or at least the last ten races)on a horses sheet, something you need to do to discern the pattern.

[Alydar in California]

Catalin wrote: \"To me, R.O.I. as it is being employed here is a pretty meaningless concept in and of itself. It is a compound statistic dependent upon how the crowd bets a race (in addition to the win %). If you employ an odds line for each race you\'re capturing the effect of the crowd by passing any horse below your estimate of fair value.\"

As I have written several times, I\'m interested in ROI to the extent that it tells us something about jockey competence. I could not care less how the crowd figures to bet a particular jockey in today\'s race. See my 8-24 \"David Patent\" post (it is about a foot above me as I write), and see my other posts on this subject, the ones that haven\'t been deleted, that is.

\"To know that a particular jockey has a positive ROI in a given situation is only useful if you think that your estimate of fair value in that situation is no better than the public\'s, in which case you are really just second guessing the probablity estimates you\'ve previously assigned to the race. I hope that made sense.\"

Not if, like me, you use ROI for nothing but what it can tell you about jockey competence. Let\'s say you have done all your handicapping (jockeys aside) and estimated the chances that each horse would run a particular figure. You have two horses at 2-1 on your line. Then you see that over a big sample, one rider has an ROI of 1.90 and the other has an ROI of 1.10. I consider this evidence that the first rider is more competent. Again, I could not care less about what ROI tells us about the crowd\'s tendency to bet a certain rider in today\'s race. That would amount to double counting. I discussed this at great length a foot above.

\"Re final fraction adjustments to turf figures, the answer is no I do not. Nothing has changed in my mind. I just have not found a satisfcatory way to do it. I think I might have the math worked out, something along the lines of a sustained pace figure adjusted for trips and weight. The problem is that it\'s too time consuming to do it for every race (or at least the last ten races)on a horses sheet, something you need to do to discern the pattern.\"

     Here is where I wanted to go with this: A horse\'s late speed is accounted for in his figure. How does a figure cheat a horse who closes fast? I know how you can answer that from a Sartin angle, but I think it would be immensely complicated to combine this with what you\'re doing now. And the temporary rails are a huge pain in the ass.

[Catalin]

Alydar wrote \"Then you see that over a big sample, one rider has an ROI of 1.90 and the other has an ROI of 1.10. I consider this evidence that the first rider is more competent.\"

Or that the second rider is dramatically overbet.  There\'s almost no way to comb the errors from that stat.  I think two more useful stats would be something like

A) How frequently a rider gets a horse to run it\'s effective top/2 pts from its top/4 pts from its top etc. and

B)Average ground loss by jockey by post by running style relative to a standard you set. (e.g.  Horses that show an \"Early\" running style breaking from post 5 loss on average 2.10 paths.  How many paths does a Jerry Bailey \"Early\" lose from post 5?)

I realize this is ridiculously cumbersome, I\'m just trying to make a point.   I do think though that those stats if available would give you a better handle on who is a good rider (coaxes a top effort more frequently / loses less than expected amount of ground) from who is not (doesn\'t often get horses to run their best / loses more ground than PP and running style should dictate).  Chris has some really good ideas on how to measure the effectiveness of jockeys (and trainers even more so) using things other than ROI.


Alydar wrote \"Here is where I wanted to go with this: A horse\'s late speed is accounted for in his figure. How does a figure cheat a horse who closes fast? I know how you can answer that from a Sartin angle, but I think it would be immensely complicated to combine this with what you\'re doing now. And the temporary rails are a huge pain in the ass.\"


What I\'m talking about doing is in effect recalculating a figure where some % of the figure is derived from early pace, and some from late speed where the mix is closer to 50/50 even though the last fraction might only account for 30% of the final time.  Something along the lines of weight/ground adjusted sustained pace numbers expressed in sheet format.  I agree that it is immensely difficult to combine pace, whether its dirt early pace or grass late pace with what we are using now.  And the rails are a pain in the ass to deal with but since few people make such turf figures they would still have value even if they weren\'t as accurate as final time numbers.

[Michael D.]

Somebody please explain this \"double counting\" argument for me. It seems to me this line of reasoning assumes that all jockeys have the same ROI, which obviously is not correct. There is clearly an advantage betting jockeys with higher ROI\'s, even after the public has spoken. This is not double counting, it\'s simply telling the market that it has marked something incorrectly. Try betting Rojas twenty times at 20-1, think you will come out even in the end?........ nooooooooo, you will not (even if I give you a check equal to the takeout). Try betting Chavez or JR ten times in the mud at 10-1, I am pretty sure you will have better luck (yes, even after the public has factored their opinion into the odds).

[derby1592]

Catalin said

\"I think a lot of what\'s been posted in this string misses the point. What is it exactly that we are trying to \"predict\"? For all those that respond \"the winner\", do not pass go, do not collect $200.

What we are really trying to forecast (at least for all those that profess to be sheet handicappers) is the probability that a horse will run a given number, how his trip (i.e ground loss, pace, etc.) will factor into that number and what is the likelihood that a given # will translate into a 1st place finish, 2nd place finish, etc. based upon the expected figures and the trips of his competitors. When we upgrade a horse based upon a trainer switch to Baffert or Frankel, or a jockey switch to Prado or Velasquez all we are really saying is that the horse is MORE likely to run his \"good number\" (trainer) and LESS likely to waste energy (ground loss, pace duels, etc.) doing it (jockey), and that these factors (increased % of good numbers / less wasted energy) translate into an increased likelihood of a win.\"

Well stated.

Chris

[Alydar in California]

Michael D. writes: \"Somebody please explain this \'double counting\' argument for me.\"

When you place a bet, you\'re purchasing the horse and the rider, among others. Let\'s say a rider has an incredible ROI, 2.40. He (not to mention the trainers to whom he is linked) has been underrated in aggregate, as your enemy Keynes might have put it, but this doesn\'t mean that all of his horses were overlays. Some were surely underlays. They had terrible patterns, etc. The egotistical, comprehensive handicapping favored by Catalin, Chris, and me would lead us, or me, at least, to say: \"Screw Bailey\'s ROI (except to the extent, if any, that it tells us something about his competence). ROI is a HINT about how the crowd will rate one factor, Bailey\'s ability, but hints are eclipsed by facts (the tote board), and we will have the facts soon enough. And since we try to take every factor into consideration, we will be able to locate those instances where this underrated rider is on a horse that is being overrated so much that the package is an underlay.\"

In short, to count the hint (ROI) when you\'re going to be counting the fact (tote odds) is to double count.

You go about things differently, Michael, but what you do is perfectly logical. In fact, one could be so bold as to say that you may still call yourself a sheet handicapper.

[Alydar in California]

Catalin wrote: \"Or that the second rider is dramatically overbet.\"

As I wrote earlier, that is why popularity and lack of same should be taken into consideration. (I believe this affects relatively few riders in a serious way, and I believe you believe the same.) That is also why I wrote \"evidence\" instead of \"proof.\"

\"There\'s almost no way to comb the errors from that stat.\"

It is desirable but not imperative to comb all the errors out of the stat. Le mieux est l\'ennemi du bien. But why not drain any jockey adjustments out of your personal line, substitute your line for the crowd\'s line, and calculate your own ROI? Call it a ROME (return on my investment), but don\'t expect to build it in a day.

\"A) How frequently a rider gets a horse to run it\'s effective top/2 pts from its top/4 pts from its top etc. and

B)Average ground loss by jockey by post by running style relative to a standard you set. (e.g. Horses that show an \"Early\" running style breaking from post 5 loss on average 2.10 paths. How many paths does a Jerry Bailey \"Early\" lose from post 5?)\"

In general, I am partial to this sort of thing. In specific, I think there are good things about this, and, in specific, I think Chris is brilliant. But please remember that riders try to WIN the race, or at least ought to be trying to win. And good riders are on good horses, horses who often OUGHT to be taken outside. The biggest problem with this sort of thing is that the trees obscure your view of the forest. Think of a method for judging pitchers that fails to give credit to pitchers who don\'t allow runners to cross home plate. In other words, Catalin, cross these ideas with ROI.

\"What I\'m talking about doing is in effect recalculating a figure where some % of the figure is derived from early pace, and some from late speed where the mix is closer to 50/50 even though the last fraction might only account for 30% of the final time. Something along the lines of weight/ground adjusted sustained pace numbers expressed in sheet format. I agree that it is immensely difficult to combine pace, whether its dirt early pace or grass late pace with what we are using now. And the rails are a pain in the ass to deal with but since few people make such turf figures they would still have value even if they weren\'t as accurate as final time numbers.\"

Please trust me. I was already extremely familiar with how this worked, but you did not answer my question.

[Michael D.]

Alydar,
First you call me a Keynesian, now a \"sheet handicapper.\" You are a cruel man. Very interesting points, but we do look at this differently. Busy morning here, check out my post later on if you are around. Cheers

[TGJB]

Again with the pitcher thing.
The concept of performance figures is to measure ability, not accomplishment, with the idea that going forward it will lead to accomplishment. Otherwise we would just keep track of wins. Variables in baseball are even more complicated (for one, pitching with a 5 run lead is different than in a tie game) but the idea behind that baserunner per inning stat (which could be refined further, but it\'s a quick check easily made on stats when the pitcher comes into the game) is the same, and usually turns out to be a very good guide to accomplishment.
The Rome in a day thing was good.