Horse Racing Forum - PaceAdvantage.Com - Horse Racing Message Board

Go Back   Horse Racing Forum - PaceAdvantage.Com - Horse Racing Message Board > Thoroughbred Horse Racing Discussion > General Handicapping Discussion


Reply
 
Thread Tools Rate Thread
Old 02-13-2024, 02:04 PM   #16
Poindexter
Registered User
 
Join Date: Mar 2007
Posts: 1,945
Quote:
Originally Posted by classhandicapper View Post
It's computer models that are making the odds more efficient, not the rebates. The rebates are just allowing the biggest teams to profit despite the odds being more efficient.
Okay, we will call them both contributing factors. You cant say rebates do not make odds more efficient.If somebody gets a 6 percent rebate and bets 10 percent of the pool at xyz downs and loses 3 percent before rebate, he just contributed 10 percent of the money in the pool at -3 percent. Thus the rebate of 6 percent caused him to make his large bets and make the pools more efficient. Any bet made that loses less than the track takeout makes pools more efficient.

It is my belief that far more people can become winning players with large rebates and no computer models than the other way around.
Poindexter is offline   Reply With Quote Reply
Old 02-13-2024, 02:36 PM   #17
Poindexter
Registered User
 
Join Date: Mar 2007
Posts: 1,945
Quote:
Originally Posted by classhandicapper View Post
I'm doing it since late 2014.

I know WAY more about a lot of things now than prior to having all that data to study, but I think the game got a lot tougher over the same period. Whatever edge I still have, it's smaller now than it was in the 90s when I knew less.
What was, was, and what is, is. So it sounds like the endeavor has been worth the time. My guess is that over the next 5 years, you will learn even more stuff and you will find the game a lot harder, 5 years from now than you do today. That is the nature of the beast. You can't knock out all the dumb money and encourage the smartest money (as this industry does) and expect anything else. Either this remains/becomes a money making voyage for you or not. Or maybe, regardless, you just don't find it worth your time.
Poindexter is offline   Reply With Quote Reply
Old 02-13-2024, 03:22 PM   #18
Dave Schwartz
 
Dave Schwartz's Avatar
 
Join Date: Mar 2001
Location: Reno, NV
Posts: 16,831
Quote:
Originally Posted by classhandicapper View Post
1. I currently use computer generated metrics and reports to help me handicap and also look through my data to learn and try to find inefficiencies. However, over the long haul, despite being a former computer programmer, I don't think I'm going to be able to use computers more effectively than teams of the most advanced computer scientists and mathematicians with far greater resources. It's unlikely I'm going to find much they also aren't going to find.

2. Just an example. A human can look at the horses in the paddock and potentially see something significant. He can then feed that information into a computer that will weigh it better than I can. However, my judgement in the paddock may be better than that other human. So I may see things he/she does not. If I do, those bits of information will never make it to the computer. That would still give an edge. The problem is there aren't many areas like that and the teams are going to employ high level staff.
That's a really good answer.

I don't think I'm going to be able to use computers more effectively than teams of the most advanced computer scientists and mathematicians with far greater resources.

If "More effectively" is replaced by "differently" it could be a whole, new ballgame.


If I do, those bits of information will never make it to the computer.

This is precisely what I am suggesting - that those bits of information MAKE THEIR WAY TO YOUR COMPUTER!

Analyzing one's OWN, UNIQUE OUTLOOK with proper tools could really make a difference.





.
Dave Schwartz is online now   Reply With Quote Reply
Old 02-13-2024, 03:57 PM   #19
denniswilliams
Registered User
 
Join Date: Dec 2023
Posts: 125
Just started reading Deep Learning With Python 2nd Edition by Francois Chollet, the creator of the Keras deep learning library.






Only Chapter 1 and I'm already regretting all the time I spent reading all that other nonsense on the topic out there.


'The most precise, unambiguous description of a mathematical operation is its executable code'
denniswilliams is offline   Reply With Quote Reply
Old 02-13-2024, 06:28 PM   #20
classhandicapper
Registered User
 
classhandicapper's Avatar
 
Join Date: Mar 2005
Location: Queens, NY
Posts: 20,443
Quote:
Originally Posted by Poindexter View Post
Either this remains/becomes a money making voyage for you or not. Or maybe, regardless, you just don't find it worth your time.
It's already not worth my time.

I do it because I love the top end of the sport and the intellectual challenge of trying to win. The money is just a way for me to keep score.

I mostly handicap and bet high quality races. Those are the fields that are getting smaller and less competitive all the time. It's tough for me to find value in 5 horse fields in NY and CA, especially if there's an odds on favorite from a top barn that looks legitimate. Advanced computer models are only going to make it tougher.

I could switch up my whole game and start studying maiden races on the turf or something else where the fields tend to be larger. There might be more value available, but I'm not going to enjoy it.

I'm not there yet, but I'm getting closer to saying "It's not worth all this effort. I proved to myself I was capable of being a winner player long term, but it's time to put my time and energy into something else I enjoy like shooting pool or playing guitar".
__________________
"Unlearning is the highest form of learning"
classhandicapper is offline   Reply With Quote Reply
Old 02-13-2024, 10:49 PM   #21
steveb
Registered User
 
Join Date: Jan 2008
Location: Melbourne Australia
Posts: 886
Quote:
Originally Posted by Nitro View Post
I’m not sure if I read your proposal correctly. I would think that the primary goal of any worthwhile computer program (related to the horse racing game) would be for humans to concentrate on providing OBJECTIVE information to produce OBJECTIVE results.
By providing SUBJECTIVE information you can only hope to achieve a range of probabilities that once again have to be (subjectively) interpreted and rationalized by humans or statistically analyzed to arrive at any number of possibilities of how likely certain outcomes will occur.

1) I believe computers will always play an important role in better understanding of our game as long as we maintain that it’s only a game we’re playing. As multifaceted as it is, I would think that the programs used must utilize the most relevant and specific data in order to produce reasonable results.

2) Concerning “discernment”, I would question how a computer program can better perceive, recognize, and evaluate things that are not obvious or straightforward. Again, I have to bring up basic factors like overall physicality and intentions.
.
.

as regards your last sentence, they(computer teams) will have people on the ground assessing how the nags present, and fed into their models.
they'll also have factors with either +ve or -ve values for commentators that make public selections(like Jenny Chapman in HK does)
steveb is offline   Reply With Quote Reply
Old 02-13-2024, 11:22 PM   #22
denniswilliams
Registered User
 
Join Date: Dec 2023
Posts: 125
Quote:
Originally Posted by steveb View Post
as regards your last sentence, they(computer teams) will have people on the ground assessing how the nags present, and fed into their models.
they'll also have factors with either +ve or -ve values for commentators that make public selections(like Jenny Chapman in HK does)

His assumption, obviously, since he constantly brings up the same issue, is that the only data being used is what's available publically. This might be because all these racing 'models' that are displayed on the web use the 'simplest' parameters. This is like grad school, where they give you a simple example in class and an exponentially more difficult problem for homework or on an exam.




BTW: I consider it the kiss of death when Chapman picks a horse I like. Though, she's actually good.

Last edited by denniswilliams; 02-13-2024 at 11:23 PM.
denniswilliams is offline   Reply With Quote Reply
Old 02-14-2024, 12:27 AM   #23
steveb
Registered User
 
Join Date: Jan 2008
Location: Melbourne Australia
Posts: 886
Quote:
Originally Posted by denniswilliams View Post
His assumption, obviously, since he constantly brings up the same issue, is that the only data being used is what's available publically. This might be because all these racing 'models' that are displayed on the web use the 'simplest' parameters. This is like grad school, where they give you a simple example in class and an exponentially more difficult problem for homework or on an exam.




BTW: I consider it the kiss of death when Chapman picks a horse I like. Though, she's actually good.
i have a list of all the factors that one group had about 18 years or so years back.
one factor that grabbed my attention, was one pertaining to how deep the track was, and how much had been added to it for any particular meeting.
don't know how valuable it was, but it was one of about 130 different factors that they had coefficients for(a lot of which were over my head, as i am more or less just a times guy).
nothing was left to chance.

still, AW told me that i could probably do better than them on any one race, but overall i would have no chance.
steveb is offline   Reply With Quote Reply
Old 02-14-2024, 01:29 AM   #24
Jeff P
Registered User
 
Jeff P's Avatar
 
Join Date: Dec 2001
Location: JCapper Platinum: Kind of like Deep Blue... but for horses.
Posts: 5,208
Steve, was that track weight factor derived (at least in part) from your work with time?

-jp
.
__________________
Team JCapper: 2011 PAIHL Regular Season ROI Leader after 15 weeks
www.JCapper.com
Jeff P is offline   Reply With Quote Reply
Old 02-14-2024, 02:23 AM   #25
steveb
Registered User
 
Join Date: Jan 2008
Location: Melbourne Australia
Posts: 886
Quote:
Originally Posted by Jeff P View Post
Steve, was that track weight factor derived (at least in part) from your work with time?

-jp
.
from memory no.
i think done independently from me.

with my way of thinking it would not be a big deal, because my way it would just be part of the variant for that particular day.
i think they placed more stock in it than i did.
steveb is offline   Reply With Quote Reply
Old 02-14-2024, 03:43 AM   #26
Nitro
Registered User
 
Nitro's Avatar
 
Join Date: Feb 2009
Location: NY
Posts: 18,811
Well after absorbing the expose’ that started this thread about statistical models, and the other recent thread involving a dissertation covering various forms of regression comparisons used in statistical analysis, I’ve come to a personal conclusion that my version of using statistics for this game is about 180 degrees off of what’s been presented so far. I say this only because it seems clear to me that these articles (and probably many others) present an attack on the game as if you knew nothing. Other than the key generic factors used to profile each entry for purposes of statistically attempting to find the best approach for finding potential winners. It dawned on me that while Benter was extremely successful, he was not a handicapper at all! He used his exceptional mathematical and programming skills to eventually utilize only the most important variables that he deemed would produce the best prediction for a race’s outcome. So, like many other situations some have been trying to use his example by trying to emulate his type of statistical analysis.

The type of statistical evaluation I’ve used over the years focuses on a different objective. It attempts to prove the legitimacy of specific criteria to discover which recognized parameters lead to an improved hit frequency. So rather than trying to observe whether something works or not by employing a common array of both subjective and objective profiling factors, it starts off with a solid premise that some known predetermined form of handicapping technique already works to some degree. By following a logical argument this valid major premise is subsequently considered true. Then logically not only will the conclusion be true but by introducing specific (rather than generic) racing criteria to test its overall validity, this conclusion will be further augmented by producing enhanced results that are more easily applied and relied upon. In other words, the credibility of the entire exercise will become self-evident, and its continued use can be safely relied upon.
I’ll provide a couple very successful examples:

Many years ago (probably sometime in the early 80’s) while still pursuing and perfecting my speed handicapping techniques I came across the name of someone who later became recognized as quite a professional handicapper and entrepreneur, Michael Pizzolla. At the time I was always intrigued by unique information about the game. He was offering a pamphlet titled something like “Profiting by Only Betting the Top 2 Betting Favorites to Win”. After receiving this nicely formatted 5 x 7 glossy paged pamphlet, it opened my eyes to another interesting aspect of the game. Not only did he provide the justification for just betting the top 2 betting choices in a race, but it also included several unique betting strategies based solely on an established hit frequency. Although most people at the time primarily referred to the favorite in a race as having an approximate 33% Win frequency, the information in the pamphlet also declared something I was unaware of: The Top 2 betting choices win 55 to 57% of the time.

I was a bit taken back by this statement and somewhat skeptical, so I decided to put some of my basic statistical knowledge to work. Rather than take his written word at face value I set up a basic algorithm that would not only confirm the truth about this, but because I noticed an average hit range mentioned I decided to take this true premise a step further. Not only did I investigate the betting of every race and their results, I also introduced all of the specifically familiar race conditions for each of those races. These conditions defined the type of race it was, by class, distance, surface, and type of entrant by age and sex. I also intentionally used only those races from a single racing jurisdiction: NYRA. (BTW, I think that simultaneously using multiple jurisdictions for any single statistical analysis is a huge mistake).

Anyway, as it turned out not only was this premise true based on an average, my little exercise covering 100’s of races over a 3 year period categorically broken down by specifics also showed some remarkable swings in the hit frequency from one type of race to another. Some were over 80% while others fell into the midrange average and others were as low a 20%. So, if you’re a player relying on just these conclusions which type of races would you play wholeheartedly?

Today I use a similar platform for playing a 3 entry Dutch Win bet in Hong Kong. I first qualify the race by establishing a play or pass rule based on the % of return profit, knowing exactly what my previous hit frequency is for the last 8 years. This hit frequency is not based on an average because every playable race is also broken down by the similar race conditions mentioned previously to identify the best race prospects for success.

So, in a nutshell, if you as a handicapper can’t come up with a realistic winning and thoroughly tested betting strategy based on accepting and including objective conditions, then perhaps the (aforementioned) statistical analysis using generic entry profile factors is the way to go. I would think that any
(even marginally) competent player with a sincere interest in making money could graduate to a higher level of play by really putting all that specific data to work.
.
.
Nitro is offline   Reply With Quote Reply
Old 02-14-2024, 10:05 AM   #27
classhandicapper
Registered User
 
classhandicapper's Avatar
 
Join Date: Mar 2005
Location: Queens, NY
Posts: 20,443
Quote:
Originally Posted by steveb View Post
as regards your last sentence, they(computer teams) will have people on the ground assessing how the nags present, and fed into their models.
Right, but at least there's a theoretically possibility someone could be better at it than whoever the teams having doing it. IMO, we are going to have to find situations where the information gathering is human vs human. That's where it's possible to get an edge over computers. It's unlikely any human is going to process information both they and the teams have better than the computers.
__________________
"Unlearning is the highest form of learning"

Last edited by classhandicapper; 02-14-2024 at 10:08 AM.
classhandicapper is offline   Reply With Quote Reply
Old 02-14-2024, 11:27 AM   #28
denniswilliams
Registered User
 
Join Date: Dec 2023
Posts: 125
Quote:
Originally Posted by steveb View Post
from memory no.
i think done independently from me.

with my way of thinking it would not be a big deal, because my way it would just be part of the variant for that particular day.
i think they placed more stock in it than i did.

Pronounced track bias last meet @ Sha Tin (the 12th) - though the last couple of races played true. I haven't done the charts for the day yet but I'm assuming they'll capture it. Are you still making figures for HKG? If so, notice any abnormalities?
denniswilliams is offline   Reply With Quote Reply
Old 02-14-2024, 05:33 PM   #29
steveb
Registered User
 
Join Date: Jan 2008
Location: Melbourne Australia
Posts: 886
Quote:
Originally Posted by denniswilliams View Post
Pronounced track bias last meet @ Sha Tin (the 12th) - though the last couple of races played true. I haven't done the charts for the day yet but I'm assuming they'll capture it. Are you still making figures for HKG? If so, notice any abnormalities?
i make figues auto generated(no imput from me), so it won't pick up bias.
from what i can tell there's not much generally.
i have coded up lots of checks that tell me when things look wrong though.
shatin generally presents no problems, in that it's mostly always within acceptable range.
HV is much more random, and much more difficult to predict.
Pretty sure that will show up in the public probs, but I have not bothered to check, it's just my opinion(will get around to it one day as it's a simple check)

it will vary according to rail possy, and which horses are drawn
where......personally i think bias in HK is overstated.
different courses favour different running styles
against that though, is that i am not very observant these days.
i am a yesterday man!
steveb is offline   Reply With Quote Reply
Reply




Thread Tools
Rate This Thread
Rate This Thread:

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

» Advertisement
» Current Polls
How do you primarily find bettable horses?
Make a formal odds line - 21.70%
23 Votes
Rank the horses in order of preference & compare it to the odds - 21.70%
23 Votes
Feel - 12.26%
13 Votes
Only play against short priced horses I dislike - 5.66%
6 Votes
Look for value oriented trips, angles, situations - 33.02%
35 Votes
Other (describe) - 5.66%
6 Votes
Total Votes: 106
This poll is closed.
Powered by vBadvanced CMPS v3.2.3

All times are GMT -4. The time now is 03:05 AM.


Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2024, vBulletin Solutions, Inc.
Copyright 1999 - 2023 -- PaceAdvantage.Com -- All Rights Reserved
We are a participant in the Amazon Services LLC Associates Program, an affiliate advertising program
designed to provide a means for us to earn fees by linking to Amazon.com and affiliated sites.