FOUND A WAY TO BEAT THE FAVORITE? SO WHAT VALUE IS THAT?

[formula_2002]

I develope what may be a reliable model to "significantly" beat the favorite.
All it does it does is cut down an absurd take out down from 17% to 11%, leaving much heavy lifting to be done.

(see my posts in the selections forum)

A bet on all 385 horses that ran in DBF races (don’t bet the favorite) returned a .83 roi, a loss of 17 cents on the dollar
The 45 horse designated “DBF” returned a .69 roi, a loss of 31 cents on the dollar.

The 340 horses not designated “DBF” returned a .89 roi, a loss of 11 cents on the dollar, cutting down the track vig from 17% to 11%.

Roi’s are the ratio for actual winners/expected winners. In a fair game this ratio should be 1.

A bet on every horse returned a loss of 385 x 17 cents = $65.45
A bet every “DBF” horse returned a loss of 45 x .31 cents = $13.95
A bet on every horse not designated “DBF returned a loss of 340 x .11 = $37.40


However,if you lay odds +15% , (sounds like a nice offer) on the 45 “DBF” horses, you should profit by 21 cents on the dollar

In this case (11 wins x 1.15)/16 expected wins =.79
1-.79 = 21 cents
(wins are sum 1/(odds+1))

[formula_2002]

A little twist to Fabricand's "theory of maximum confusion",

I modeled 76 plays where the 2nd choice had characteristics
That were very similar to the favorite.

In 493 races the model designated the favorite "DBF”, that is, don’t bet the favorite. In those races, there were 76, 2nd choices that were similar to the favorite.

When two horses are similar, Fabricand would say to bet the favorite because of "inside" or "late information. He goes a bit further, demanding certain odds criteria and some other things, but that’s the nut of it.

I my case, I'm assuming the public is betting on the favorite because it does not see what I see, that is, the 2nd choice is similar to the favorite and the favorite is being overbet, while the 2nd choice is being under bet.

In the 493 races the favorite returned a .66 roi.
The 2nd favorite was, to me, similar to the favorite in 76 of those races and returned a 1.21 roi!

Obviously, further testing is required.

JUST A NOTE, I CORRECT MYSELF, IT'S NOT JUST THE 2ND CHOICE. MOST OF THE TIME IT IS, BUT SOME TIMES ITS THE 3RD OF 4TH CHOICE.
THE ODDS ON THE SIMILIAR HORSE ARE CAPPED AT 4-1, BE IT 2ND, 3RD OR 4TH CHOICE.

[jamey1977]

A little twist to Fabricand's "theory of maximum confusion",

I modeled 76 plays where the 2nd choice had characteristics
That were very similar to the favorite.

In 493 races the model designated the favorite "DBF”, that is, don’t bet the favorite. In those races, there were 76, 2nd choices that were similar to the favorite.

When two horses are similar, Fabricand would say to bet the favorite because of "inside" or "late information. He goes a bit further, demanding certain odds criteria and some other things, but that’s the nut of it.

I my case, I'm assuming the public is betting on the favorite because it does not see what I see, that is, the 2nd choice is similar to the favorite and the favorite is being overbet, while the 2nd choice is being under bet.

In the 493 races the favorite returned a .66 roi.
The 2nd favorite was, to me, similar to the favorite in 76 of those races and returned a 1.21 roi!

Obviously, further testing is required.

JUST A NOTE, I CORRECT MYSELF, IT'S NOT JUST THE 2ND CHOICE. MOST OF THE TIME IT IS, BUT SOME TIMES ITS THE 3RD OF 4TH CHOICE.
THE ODDS ON THE SIMILIAR HORSE ARE CAPPED AT 4-1, BE IT 2ND, 3RD OR 4TH CHOICE. What was your criteria for determining similarity ? . Your concept is the same way all of us find a way to get that profit. You have demonstrated to all of these Doubting Thomas 's that beating the track can be done. Break down those races and find that common denominator. You've said it all.

[formula_2002]

What was your criteria for determining similarity ? . Your concept is the same way all of us find a way to get that profit. You have demonstrated to all of these Doubting Thomas 's that beating the track can be done. Break down those races and find that common denominator. You've said it all.

76 plays is not a statistically proven demonstration. It's only a hope for the future.:)


the criteria is about 50 factor of hundreds of factors generated by JCapper, some statistical analysis, thousands for Bris single file format data structures, and a lot of time..
Other than that,, anyone can do it :)

[formula_2002]

In a series of races from 6-30-2010 through 07-09-2010, 65 DBF plays returned a .65 roi, inline with the above study.
However when the races include todays 10 DBF plays, 75 DBF plays returned a .84 roi.
Hopefully, this one day, where the DBF plays returned a 1.80+ roi and all favorites returned a 1.30 roi is an anomaly, and the world will return to its normal spin tomorrow :)

[BC_Sprint_Guy]

However... what does DBF stand for? :confused:

[PhantomOnTour]

However... what does DBF stand for? :confused:
Don't Bet Favorite...which means if this horse is the favorite then don't bet it.

[formula_2002]

DBF horses

While the concept has not changed, the program has.

The idea is to find favorites that are similar to other horses.
The other horses are similar to the favorite based on MY rules of similarity, which also includes the public’s odds
Fabricand also required a minimum 1/ (odds+1) for the other horse or sum 1/ (odds+1) for the other horses.
My program did not provide for that and it should have.

To help compensate for that requirement, at this time I will insist on at least 2 other horse that are similar to the favorite based on ML odds.
Therefore, at least three horses must carry the tag , “DBF” in any one race.
If one of these turns out to be the favorite, then bet the remaining “DBF” horses providing they are of odds <=4-1.
Preliminary testing indicates the favorite roi in this case to be less than .50
Positive roi’s for the remaining DBF horses <=4-1 is a bit spotty.

[BetHorses!]

I feel the best value is to toss the bad fav from the tri's and supers.

minimum of 10 horse fields and I sometimes box up to 6 horses

[formula_2002]

I feel the best value is to toss the bad fav from the tri's and supers.

minimum of 10 horse fields and I sometimes box up to 6 horses

I TESTED THE DBF PLAYS ON A FRESH DATA BASE.
IN A 177 PLAYS, WHEN DBF IS THE FAVORITE, IT RETURNED A .58 ROI.
I'M CONFORTABLE WITH NOT BETTING THE DBF, I JUST CAN'T FIND A RELIABLE WAY OF BEATING IT WITH A POSITIVE ROI.


I recently read something in the book “Quantum”.
Many renowned scientists will only study things that can be tested and verified.
Einstein was one of the few that did not follow that line of thinking.
Now I know why I’m no Einstein :)

[BetHorses!]

I TESTED THE DBF PLAYS ON A FRESH DATA BASE.
IN A 177 PLAYS, WHEN DBF IS THE FAVORITE, IT RETURNED A .58 ROI.
I'M CONFORTABLE WITH NOT BETTING THE DBF, I JUST CAN'T FIND A RELIABLE WAY OF BEATING IT WITH A POSITIVE ROI.




How many times was the DBF off the board regardless if it was actual Fav?

[formula_2002]

[QUOTE=BetHorses!]How many times was the DBF off the board regardless if it was actual Fav?[/QUOTE

This may be more helpful..let me know.

In a fresh data base (a bit larger than last reported), there were 1806 favorites that returned a actual wins/ expected win roi of .87

there were 226 races where I designated the favorite as a bad bet (DBF). that a/e roi was .64.

there were 776 horses designated ,DBF and not the favorite that a/e roi was .75

the significance is the 23 poinf difference between all favorites and DBF favorites.

[formula_2002]

the 1st note of this thread;
The 45 horse designated “DBF” returned a .69 roi, a loss of 31 cents on the dollar.


Ive added to the data base with 18,000 races as far back as 2007 and get these new totals which are very similiar to the origonal study


[I]poor betting favorites, "DBF", returned a 30% loss (732 plays)


3rd choices , B3, return a 3% loss (680 play)

betting horses >=10-1, returned a 21% profit (808 plays, generally more than 1 plays per race.)

betting favorites, odds <=1-1, returned a 2% loss (333 plays)

LTCP plays returned a 5% profit (863 plays)


rules of play and information on the study, can be found on my web site.
I'll be adding commentary and recaps of the plays posted in the selections forum there.

[stats725]

your article was very interesting.my friends did lots of research on Fabricands system . Here is the actual results they got; eastern tracks only 434 plays 190 won 43.7% at a 5.66 average price for a profit of 19.4% 257 placed ,59.2% at a 3.52 average price, for a profit of 4.2%. this test was done a long time ago , but I think it will still work. anyone who would like to discuss this can just call me at 702 877 2468 7am to 11pm, best time 8pm to 11pm vegas time, good luck. Steve

[CBedo]

betting horses >=10-1, returned a 21% profit (808 plays, generally more than 1 plays per race.)I have a similar model that I use as part of a more comprehensive approach that I arrived at from a logical qualitative standpoint.

In my handicapping, I found some longshots that seemed to jump off the page as being live, but when betting them, many (most actually) did not win or even hit the exacta. Originally I thought I must have underestimated their flaws or overestimated the good qualities that stood out to me in the first place.

After further research, I found that most of the time, they got beat by one of the top two or three public choices, and possibly I had found good horses, but had failed to relate them to how they fit in the race itself. Favorites were killing me.

So after doing some work to try to identify false or more usually, vulnerable favorites (and 2nd & 3rd choices as well), I had two sets of races, races with vulnerable favorites, and a set of races that contained my positive long shots. When I began to bet the longshots in the intersect of these two sets, my roi jumped dramatically. Individually, both were important, but I needed both together to get into positive territory.

Long winded way of saying "vulnerable top picks plus something to like about some other horse equals positive value."

[cnollfan]

I have a similar model that I use as part of a more comprehensive approach that I arrived at from a logical qualitative standpoint.

In my handicapping, I found some longshots that seemed to jump off the page as being live, but when betting them, many (most actually) did not win or even hit the exacta. Originally I thought I must have underestimated their flaws or overestimated the good qualities that stood out to me in the first place.

After further research, I found that most of the time, they got beat by one of the top two or three public choices, and possibly I had found good horses, but had failed to relate them to how they fit in the race itself. Favorites were killing me.

So after doing some work to try to identify false or more usually, vulnerable favorites (and 2nd & 3rd choices as well), I had two sets of races, races with vulnerable favorites, and a set of races that contained my positive long shots. When I began to bet the longshots in the intersect of these two sets, my roi jumped dramatically. Individually, both were important, but I needed both together to get into positive territory.

Long winded way of saying "vulnerable top picks plus something to like about some other horse equals positive value."

This sounds good.

[Robert Fischer]

when you have vulnerable heavy favorites, they don't really require a single special pick elsewhere in the same race.

obviously by definition that true contending horse will be overlayed.

[levinmpa]

I TESTED THE DBF PLAYS ON A FRESH DATA BASE.
IN A 177 PLAYS, WHEN DBF IS THE FAVORITE, IT RETURNED A .58 ROI.
I'M CONFORTABLE WITH NOT BETTING THE DBF, I JUST CAN'T FIND A RELIABLE WAY OF BEATING IT WITH A POSITIVE ROI.


I recently read something in the book “Quantum”.
Many renowned scientists will only study things that can be tested and verified.
Einstein was one of the few that did not follow that line of thinking.
Now I know why I’m no Einstein :)

I was using HTR several years ago and identified a pattern on certain ML favorites that would only win about 14% of the time. Like you I thought this was great information, but never identified a profitable way to play those races. I always thought the only way to make a profit on these DBFs was to literally bet against them, like with Exchange wagering. Offer 5/2 on the 2/1 DBF or something of that nature and you should make a nice profit over the long run. You're laying odds, but you're getting the entire field against the DBF. Of course, exchange wagering is not available to most of us legally, so I never pursued.

[formula_2002]

Beating the favorite and the track

Example of three horse field race in an 18% take out track.

Let’s say the favorite odds are 1-1 then;
A.
Three real horses + one track horse
Real horse #1 1-1 track odds .50 probability
Real horse #2 3-1 track odds .25 probability
Real horse #3 3-1 track odds .25 probability
Track horse 3.5-1 track odds .22 probability
Total book 1.22

The track has a winning horse in every race and no one can bet on that horse, and it pays 3.5-1 odds.

B.
The real probability of the hoses #1, #2 and #3 are
Real horse #1 1-1 track odds .39 probability
Real horse #2 3-1 track odds .195 probability
Real horse #3 3-1 track odds .195 probability
Track horse 3.5-1 track odds .22 probability
Total book 1.00


To break even beating the favorite horse and the track horse, the public must over bet the favorite.


A $1 bet on the #2 and the #3 horse in 100 races cost $200
Given that horse #2 and #3 run at track odds of 3-1, then in order to get a return of $200 in 100 races, any one of the two horses must win a combined total of 50%.

Therefore;
C.
Real horse #1 1-1 track odds .28 probability
Real horse #2 3-1 track odds .25 probability
Real horse #3 3-1 track odds .25 probability
Track horse 3.5-1 track odds .22 probability
Total book 1.00

In this case, the even money favorite must win no more than 28% of the time, compared to the 39% of the time it would “normally” win,
72% of normal




Now let’s say the favorite odds are 1.5-1 then;

A1
Real horse #1 1.5 -1 track odds .40 probability
Real horse #2 2.33-11 track odds .30 probability
Real horse #3 2.33-1 track odds .30 probability
Track horse 3.5-1 track odds .22 probability
Total book 1.22



The real probability of the hoses #1, #2 and #3 are;

B1
Real horse #1 1.5 -1 track odds .33 probability
Real horse #2 2.33-1 track odds .225 probability
Real horse #3 2.33-1 track odds .225 probability
Track horse 3.5-1 track odds .22 probability
Total book 1.00


To break even beating the favorite horse and the track horse, the public must over bet the favorite.


A $1 bet on the #2 and the #3 horse in 100 races cost $200
Given that horse #2 and #3 run at track odds of 2.33-1, then in order to get a return of $200 in 100 races, any one of the two horses must win a combined total of 60%.


Therefore;
C1
Real horse #1 1.5-1 track odds .18 probability
Real horse #2 2.33-1 track odds .30 probability
Real horse #3 2.33-1 track odds .30 probability
Track horse 3.5-1 track odds . .22 probabilities
Total book 1.00

In this case, the even money favorite must win no more than 18% of the time, compared to the 33% of the time it would “normally” win,
54.5% of normal


No accommodation is made here for the long to short odds biases.

[classhandicapper]

I more or less agree with the contention that you need "TWO" value oriented opinions in a race to turn solid profits. That's been my experience too. The take is so darn large that even if you find a mistake on the board, it usually only large enough to get you in the ballpark of break even plus or minus a little. It's rare that you find a favorite at 8-5 that should be 8-1 or something that extreme. You usually find 8-5 shots that should be 3-1 or 7-2 (things like that). So not only do you have to find a favorite you aren't very keen on, you have to find a horse in the same race that appears under bet relative to the rest of them.

[Robert Fischer]

I more or less agree with the contention that you need "TWO" value oriented opinions in a race to turn solid profits. That's been my experience too. The take is so darn large that even if you find a mistake on the board, it usually only large enough to get you in the ballpark of break even plus or minus a little. It's rare that you find a favorite at 8-5 that should be 8-1 or something that extreme. You usually find 8-5 shots that should be 3-1 or 7-2 (things like that). So not only do you have to find a favorite you aren't very keen on, you have to find a horse in the same race that appears under bet relative to the rest of them.

there are different types of wagering situations

sounds like you are describing a type you have had some consistency with.

[classhandicapper]

there are different types of wagering situations

sounds like you are describing a type you have had some consistency with.

Agreed. In general this is the way I see it.

It's hard to cover the entire take if your only opinion is that the favorite is over bet unless that horse is extremely over bet (which doesn't happen often).

Usually you have to like something about another horse that is not reflected on the board too. Whether it's win, exactas, multi race bets etc... you really want to have more than one opinion if your approach is to bet against bet favorites.

[davew]

the value would be MUCH greater if you had a method to pick favorites who would not make the superfecta

it must happen in the 15-20% range depending on field size

[CBedo]

the value would be MUCH greater if you had a method to pick favorites who would not make the superfecta

it must happen in the 15-20% range depending on field sizeThere's greedy, and then there's GREEDY! :lol:

[pondman]

However,if you lay odds +15% , (sounds like a nice offer) on the 45 “DBF” horses, you should profit by 21 cents on the dollar


Hate to say it-- the returns a little low for me. Be a little tight for me. I'd rather sit and hit a few longer shots and not have to be at the window for pennies.