On Line Travers Day Seminar

[TGJB]

Wait, 70 and 100 aren\'t the same thing?

[TGJB]

Still trying to understand this. So if I take 1/5 100 times, and win 70, that means I\'m going to think I\'m really smart 70 times, right? What\'s wrong with that?

[miff]

Guess you know,you need to win 84 out of 100 to break even @ 1-5. Since you like stats, 1-5 shots underperform vs their mathematical probability and do not produce break even(unfiltered,of course,large sample size done 7-8 years back)

[TGJB]

I don\'t especially like stats, except when the results are extreme, and I agree that everything has to be taken in context. (The place they really go nuts with taking stats out of context is the commercial breeding industry. Total BS). Stats are one data point, among many.

But I\'m not shocked about the 1/5 shots-- I would say anyone who\'s been betting horses for a while wouldn\'t be, and wouldn\'t need a study.

The \"gun to your head\" guy was great. Yes, if it\'s a binary choice, 1/5 shots are more likely to win than lose, and bullet or no bullet is pretty binary. Pari-mutuel betting offers many more choices, and corresponding odds.

[miff]

The mistake people make is assuming that AP is 100% to show up and run his best and that no other horse will run faster than it ever did before and beat AP.Very speculative to start putting probabilities on those two possibilities which are contradictory to the present pp\'s of the field.

Stick to my take that the only conceivable reason to try to beat AP is a risk/reward prop vs a creative racing reason.

[TGJB]

They\'re always risk/reward questions.

[Mathcapper]

miff Wrote:
-------------------------------------------------------
> 1-5 shots underperform vs their mathematical
> probability and do not produce break even
> (unfiltered,of course,large sample size done 7-8 years back)

Not sure if you\'r referring to this, but Steve Klein did the definitive study on odds-based ROI about 10 years ago in his book, The Power of Early Speed, covering 8 years worth of races and over 1.6 million horses: Summary results from The Power of Early Speed

He found that heavy chalks (1-9 to 1/2) actually outperform their mathematical probabilities, although as you noted, not by a large enough degree to break even. On average, they produce an ROI of around -13% as compared to an average track take/breakage of around -18.5%.

The pools have gotten more efficient since then, so not sure how pronounced the fave/longshot bias still is in the win pools today.

The other notable thing, and the focus of Klein\'s book, was the incredible ROI produced by early leaders, at any odds, as I\'m sure any past-posters can attest to.

fwiw

Rocky R

[miff]

Thanks Rocky,

Believe what I saw came out of some British stat geeks. Recall that, as you noted, longshots performed worse than odds on horses vs probability.

Mike