MPRanger
04-28-2012, 10:59 PM
I'm hoping someone smarter than me can help me figure this out;
In Ziemba and Hausch's Dr. Z's Beat The Ractrack Chapter 3, entitled
Applying Stock Market Efficiency Concepts to Horseracing Betting Markets,
the authors present the case of the efficiency of the toteboard.
They also point out a certain inefficiency as far as the toteboards fractional
odds and the natural decimal percentages. However, if I understand this
right, you can account for this longshot - favorite bias and adjust for it so as
to bring the toteboard odds back into a higher measure of efficiency.
I'm alright with that part. I believe in the efficient market hypothesis. I
believe the toteboard is efficient, not as tight as an NFL point spread but
generally very accurate.
One of my problems is that I don't naturally think in the same terms as the
way they wrote. For example, I think in terms of Track Take Out as opposed
their reference to Track Pay Back. GP has a 17% takeout. I go 100-17=83.
To get true odds for a 2-1 shot I'll take .83*33.33 to get 27.66.
On page 38 there is table 3.2, it shows adjustments for the bias but
it includes the "Track Payback" figure. They got these adjustment numbers
by comparing actual results with post time odds. But even their own "Track
Payback" numbers don't make sense to me. I can't reproduce their numbers
by what I see in the book. There are several other charts preceding fig 3.2
which are interesting but they don't seem to be connected in a way that
one leads to the next and this is the result. In fact they are from several
different studies. They seem to be presented more as examples than as
premises.
It could be that I'm just missing something because I'm not really a math
guy. And this is what I'm asking for assistance with or at least some ideas
if anyone wants to discuss it.
To keep it simple:
1.) 1-1 fractional odds decimal equivalent = 50%
2.) Track Take Out = 17%
3.) Their LS-Fav Bias adjustment number is 8.2
Ok, I would go like this:
100-17=83
.83 x 50 = 41.5
Then add their adjustment back in?
41.5 + 8.2 = 49.7
But they don't speak this language.
They show "Track Payback" (NY 17%) for the above as 90.2.
I don't see how they get this number or even why they would use
this number as opposed to true odds.
Also, I'm aware that Barry Meadow found errors in their predictions
for place and show due to using a the Harville formulas which do
not account for horse and jockey's giving up or other living being
factors. But this is different from that. I respect these authors and think
they have done a great study even if flawed in some ways. I'm not saying
this LS-Fav bias thing is flawed, just that I don't quite understand it.
So, what better place to come with this than PaceAdvantage.com
In Ziemba and Hausch's Dr. Z's Beat The Ractrack Chapter 3, entitled
Applying Stock Market Efficiency Concepts to Horseracing Betting Markets,
the authors present the case of the efficiency of the toteboard.
They also point out a certain inefficiency as far as the toteboards fractional
odds and the natural decimal percentages. However, if I understand this
right, you can account for this longshot - favorite bias and adjust for it so as
to bring the toteboard odds back into a higher measure of efficiency.
I'm alright with that part. I believe in the efficient market hypothesis. I
believe the toteboard is efficient, not as tight as an NFL point spread but
generally very accurate.
One of my problems is that I don't naturally think in the same terms as the
way they wrote. For example, I think in terms of Track Take Out as opposed
their reference to Track Pay Back. GP has a 17% takeout. I go 100-17=83.
To get true odds for a 2-1 shot I'll take .83*33.33 to get 27.66.
On page 38 there is table 3.2, it shows adjustments for the bias but
it includes the "Track Payback" figure. They got these adjustment numbers
by comparing actual results with post time odds. But even their own "Track
Payback" numbers don't make sense to me. I can't reproduce their numbers
by what I see in the book. There are several other charts preceding fig 3.2
which are interesting but they don't seem to be connected in a way that
one leads to the next and this is the result. In fact they are from several
different studies. They seem to be presented more as examples than as
premises.
It could be that I'm just missing something because I'm not really a math
guy. And this is what I'm asking for assistance with or at least some ideas
if anyone wants to discuss it.
To keep it simple:
1.) 1-1 fractional odds decimal equivalent = 50%
2.) Track Take Out = 17%
3.) Their LS-Fav Bias adjustment number is 8.2
Ok, I would go like this:
100-17=83
.83 x 50 = 41.5
Then add their adjustment back in?
41.5 + 8.2 = 49.7
But they don't speak this language.
They show "Track Payback" (NY 17%) for the above as 90.2.
I don't see how they get this number or even why they would use
this number as opposed to true odds.
Also, I'm aware that Barry Meadow found errors in their predictions
for place and show due to using a the Harville formulas which do
not account for horse and jockey's giving up or other living being
factors. But this is different from that. I respect these authors and think
they have done a great study even if flawed in some ways. I'm not saying
this LS-Fav bias thing is flawed, just that I don't quite understand it.
So, what better place to come with this than PaceAdvantage.com