At what odds?

[Capper Al]

There has been lots of talk about making wagers on low odd horses with a greater degree of certainty verse higher odds horses that pay better but are longer shots. Many system touts have written systems that cherry pick low odd horses. Most veteran handicappers are repelled by this. For to the old timers, who have survived many pari-mutual pool battles, know from their experience to wait and cherry pick the higher odds horses. Both camps had viewed the truth about wagering differently, and they were diametrically opposed. They see the situation as either/or.



The truth according to Quirin, in his classic handicapping book Winning at the Races, is both and neither. The key is being able to understand a horse's true value in today's race. Let's review overlay and underlay. (These are easy to mix up.)



An underlay horse is over bet in proportion to its true chances of winning which would be a losing premise over the long run. On the other hand, an overlay is an under bet horse that would show a profit over the long run. For example, one horse going off at 4/5 odds and still can be an overlay while a horse going off at 10/1 odds could be an underlay. Matter of fact, in general according to Quirin on page 298, horses running at odds of 1/2 or less are overlays while horses at higher odds are underlays.



The trick is in figuring a horse's true value, if it can be done. Some authors have suggested asking yourself "if this race was run a 100 times how many times would you expect your horse to win?" This is much too subjective for me.



Then there are those who just bet against the favorite period. Think again. If the average race has an 8 horse field and the favorite on average wins 1 out 3 then the other 7 horses have a 2 out of 3 change between themselves. This translates to 2/3 times 1/7 for on average or 2/21 for each horse, a little over 10/1 odds.



On other side of the fence, betting on the favorite every time is a slow death that will eventually tap out a player. Yet Quirin charts do point to overlays on chalk, and many systems have been built on separating false favorites from true favorites with a better than expected chance to win. If one could just eliminate 20% of the false favorites they should be profitable. This might be the most prudent course of action to beat the game.

[mrharness]

I knew a guy that only bet if a horse was 4/5 or lower. And if the odds were going to move before post time, he only wanted them to move in one direction - down. However, 1/9 was too low. He had money coming in from other interests but he could have lived on what he made at the track.

[Overlay]

A person should neither avoid favorites altogether, nor bet higher-odds horses indiscriminately, merely because of their odds. It's also common advice to avoid factors that the general public is keying on, but the farther away you get from handicapping fundamentals, the more your hit rate suffers. Even initially lucrative spot-play angles lose their profitability with time and increased public awareness.

I think wagers should be based on how your assessment of a horse's winning chances (however you arrive at it, although I, too, prefer using a more objective, quantitative technique, rather than subjective opinion) compares with the public's, over the full odds spectrum, and without eliminating any horse in a field outright from consideration. To me, the keys to an approach that holds up over time are taking a variety of basic handicapping factors into account (rather than keying solely on one overriding element), and (as CapperAl indicated) betting on the basis of value, which can potentially be found at any odds level.

[Capper Al]

A person should neither avoid favorites altogether, nor bet higher-odds horses indiscriminately, merely because of their odds. It's also common advice to avoid factors that the general public is keying on, but the farther away you get from handicapping fundamentals, the more your hit rate suffers. Even initially lucrative spot-play angles lose their profitability with time and increased public awareness.

I think wagers should be based on how your assessment of a horse's winning chances (however you arrive at it, although I, too, prefer using a more objective, quantitative technique, rather than subjective opinion) compares with the public's, over the full odds spectrum, and without eliminating any horse in a field outright from consideration. To me, the keys to an approach that holds up over time are taking a variety of basic handicapping factors into account (rather than keying solely on one overriding element), and (as CapperAl indicated) betting on the basis of value, which can potentially be found at any odds level.

Determining value is very difficult. How does one call a horse worthy of 4/5 odds and another worthy of 9/2? Some would jump on the 9/2 just because he's higher.

[Overlay]

Determining value is very difficult. How does one call a horse worthy of 4/5 odds and another worthy of 9/2? Some would jump on the 9/2 just because he's higher.

It's relative within each individual race field. You mentioned the "if this race were run a hundred times" method as one way. On the quantitative side, Quirin (for example) demonstrated a weighted, multiple-regression rating method, including a discussion of how the final composite total for each horse could be converted to fair odds. To me, a numerical approach like that has the advantages of being more consistent from race to race; of giving the ability to tear the handicapping model down for analysis to see where it might need adjustment if unexpected results start regularly occurring; and (to address your point) of providing greater confidence in odds assignment and wagering than if you were relying only on opinion or guess (however well-educated it might be). That helps in maintaining betting discipline, and in basing wagers on how the odds compare to winning chances, rather than (as you noted in your post) on just the odds themselves.

[jonnielu]

Determining value is very difficult. How does one call a horse worthy of 4/5 odds and another worthy of 9/2? Some would jump on the 9/2 just because he's higher.

Knowledge of ability and condition for today's race.

[Tom]

As a sinlgle bet, I would never bet a 4-5 shot. Pass or find something else. 4-5 is like playing a change machine. :rolleyes:

[oddsmaven]

I wouldn't argue with those that reject betting heavy chalk, but sometimes I will take them, though often on top of an exotic...as CapperAl suggests, a 4:5 shot could actually have even a better chance of winning than its price...to grind them out, one needs patience and not too greedy a disposition.

It's easier to find occassional plays at decent odds using certain angles, like "lone speed of the race", but if your strength is assessing accurate odds for a field, there are opportunities for finding dominant selections...not all of these "obvious standouts", merit a clearcut edge, and it is a ton of work - considering numerous factors intelligently (class, condition, distance, pace, and many more)...if you overlook one thing, your projections will be probably be out of whack...and I would not trust the morning line as a solid guide.

Many bettors may feel they can identify overlays, but very few, IMO really have a thorough grasp.

[HuggingTheRail]

Many bettors may feel they can identify overlays, but very few, IMO really have a thorough grasp.

Then, once identified, there is the whole psychology of not being swayed by the real time odds.

It took me too long to accept my work/numbers, and not second guess myself. For a long time, I would sit there thinking...."this horse is worth 8-1...he is currently 17-1..what am I missing?" Then I would second guess myself, and often times talk myself out of my belief in that horse. Once I got over that hurdle, then much progress was made.

[Capper Al]

As a sinlgle bet, I would never bet a 4-5 shot. Pass or find something else. 4-5 is like playing a change machine. :rolleyes:

Quirin's and others show that in general horses going at 1/2 odds or lower are good to bet and have a profitable return over the long run. Not that I don't share your sentiment on this, and I may skip the race also. It is a repeatable stat that seems to have stood the test of time.

[GaryG]

Quirin's and others show that in general horses going at 1/2 odds or lower are good to bet and have a profitable return over the long run. Not that I don't share your sentiment on this, and I may skip the race also. It is a repeatable stat that seems to have stood the test of time.You must mean to place, certainly not to win.

[Overlay]

You must mean to place, certainly not to win.

The statistic that CapperAl is citing can be found on page 298 of Winning at the Races, where Quirin was presenting an analysis of the 2189 races run at the three NYRA tracks (Aqueduct, Belmont, and Saratoga) in 1970, to test the validity of the proposition that horses win at an overall rate consistent with their toteboard odds. Of the 19,075 horses that competed in those races, 15 went off at odds of 1-5, of which 13 won; 8 went off at 2-5, of which 6 won; and 18 went off at 1-2, of which 12 won. The $NET for the 15 1-5 horses was $2.20. The cumulative $NET for the 23 horses at 2-5 or lower was $2.17. The cumulative $NET for the 41 horses at 1-2 or lower was $2.09. However, when the 86 horses that went off at 3-5 (of which 43 won) were added to the mix, the cumulative $NET dropped to $1.79, and kept on dropping as the odds got higher, reaching a final value for all 2189 races of $1.56. (Again, this is 1970 data, but the results of the statistical tests that Quirin applied to it were presented as indicating that the profitability he found for those odds ranges would continue into the future.)

[HUSKER55]

If odds are determined by how many times would this horse win if this race were run 100 times then that super imposes that the SAME horses, (ie., field), track, distance, surface, weather and etc are identical. After today it will never happen again. There are no multiple events.

I AM NOT A MATH MAJOR. IF YOU ARE AND I AM WRONG PLEASE ADVISE. :)

I try to make the most money I can each race. If I say #1 and #2 will win and the track says #4 and #5,.....one of us is going into the next race with a smile on our face.

If I say #1 and #2 and the track says #1 and #2,..if I can not find a way to at least double my money then I add my alotment for that race to the next and move on.

Why do we handicap except to seperate the contenders from those who are not. We have assigned odds when we make our picks. Next step is to make money.

SUPPOSE, you have narrowed down a ten horse field to two contenders and one maybe and the two contenders are close.

Seems easy. 2:1, 2 1/2 : 1 & 3:1 and everything else is 20:1 (I am not interested in the rest)

If I have missed the boat will someone throw me a life jacket? :) I do not understand the notion of making your own "line".

[cj's dad]

The odds are that tracks will pay 33% to 38% on post time favorites. So, if you factor out the 1/5 < 4/5 shots the odds are skewed. I'm not sure of the increase/decrease when the 1/5< 4/5 shots run 2nd or worse, but it would seem that betting against these odds on runners would be profitable. Interested to hear comments regarding this approach.

[Robert Fischer]

when i handicapp i dont give a care what the public or the morning line thinks.

When I wager, its a different story :D.

In general I don't bet any horse to win, that I don't rate as one of the top 2 or 3 win candidates regardless of odds.

[BetHorses!]

I do not understand the notion of making your own "line".


All I can say is unless you start understanding you will be a losing player at the end of the year...every year.

[Cratos]

The statistic that CapperAl is citing can be found on page 298 of Winning at the Races, where Quirin was presenting an analysis of the 2189 races run at the three NYRA tracks (Aqueduct, Belmont, and Saratoga) in 1970, to test the validity of the proposition that horses win at an overall rate consistent with their toteboard odds. Of the 19,075 horses that competed in those races, 15 went off at odds of 1-5, of which 13 won; 8 went off at 2-5, of which 6 won; and 18 went off at 1-2, of which 12 won. The $NET for the 15 1-5 horses was $2.20. The cumulative $NET for the 23 horses at 2-5 or lower was $2.17. The cumulative $NET for the 41 horses at 1-2 or lower was $2.09. However, when the 86 horses that went off at 3-5 (of which 43 won) were added to the mix, the cumulative $NET dropped to $1.79, and kept on dropping as the odds got higher, reaching a final value for all 2189 races of $1.56. (Again, this is 1970 data, but the results of the statistical tests that Quirin applied to it were presented as indicating that the profitability he found for those odds ranges would continue into the future.)

If you do the math you will find that if a $2 bettor betting win only had won with horses of which 15 went off at odds of 1-5, of which 13 won; 8 went off at 2-5, of which 6 won; and 18 went off at 1-2, of which 12 won. The bettor would show a profit of $.0437 on the dollar or less than the track’s guaranteed $.05 on the dollar. Although the bettor's actual profit wasn’t negative, the profit against the track’s guarantee was and the track achieved a $.0065 profit on the dollar from the bettor. Very bad wagers.

[Overlay]

The track does guarantee a minimum payoff of $2.10, but that's only on winning bets. What Quirin's $NET figure is showing is that for every $2.00 bet that was placed on those horses (winning or losing), the bettor received back an average of $2.20, $2.17, or $2.09 (depending on the odds range), based on the ratio of the proceeds from winning bets to the losses on losing bets. Any $NET figure above $2.00 represents actual profit.

[Overlay]

I do not understand the notion of making your own "line".

The odds-assignment example you gave sounds like that's what you're doing. Granted, each race is unique and will never be repeated. But what makes it valid and/or necessary to assign winning probabilities to each horse (however you choose to go about it) is factoring the odds established by the public into the picture. It all boils down to the question of whether you believe that it's possible to say with 100% certainty before a race is run that any particular horse will win it. If it's not possible to do that (that is, if there is any amount of uncertainty about which horse will in fact win), then the wagering decision changes from, "Which horse will win?" to a question of whether the payoff you would stand to receive on any given horse (as reflected by its odds) will adequately or fairly compensate you for the risk you're taking in betting it. That's the judgment that a fair-odds line helps you to make, by providing a basis for comparison between your assessment of a horse's winning chances versus the public's.

[HUSKER55]

Thanks for the input I appreciate your efforts:)

[Capper Al]

Bet what you like. I won't bet an under even money horse to win, but the facts are that this is cutting across the grain and going against the only verifiable underlay in the business. It makes me think.

[Capper Al]

Is there even an interest to watch the NHC on TV? I will try to watch it, if I can. I just am not expecting much.

[jonnielu]

Bet what you like. I won't bet an under even money horse to win, but the facts are that this is cutting across the grain and going against the only verifiable underlay in the business. It makes me think.

What's to think about? You have a probable winner at odds that are not worth the risk of a bet. Go get a hot dog, and think about how horse races are like buses and women. There is another one coming along in 20 minutes.

jdl

[mrharness]

These are some questions for anyone that makes and uses an oddsline.

Assume you are showing a profit. What do you do to verify that your oddsline is correct (or as correct as possible)? As an example, say you are betting horses you think should be 4/1 - do you check to see if your numbers match the toteboard odds you want, after the final flash? If you feel the horse is 4/1, and you bet only if they are 5/1 (or better) on the tote - do you check through all of your 4/1's to verify that you are right on target? And likewise for each odds category (1/1, 2/1, 3/1, etc)? What if you find you are losing money in several categories - but still showing a profit overall? If you use some mathematical formula for your oddline, how can you adjust it, if it is off only in certain categories? If you do it subjectively, how can you adjust your oddsline...what do you tell yourself...I think this is a 4/1 but since my line is off, I will lower it to 7/2 or raise it to 9/2 everytime you want to bet?

If you do make up a subjective line, what factors are you looking at that suggest to you that you think horse A should be 3/1 and horse B should be 5/1? Beaten Lengths? Finish Time? ??? When I talked to a guy that set up the morning line for a track, he would look at the last quarter (harness) and choose what he thought the favorite should be - then there was a list of predetermined odds that he could assign to the rest of the horses. Actually, there was more than one list, depending on what he assigned to the favorite. It wasn't like he made all the odds up in his head and made a real decision.

How many races do you bet before checking/verifying? 20? 100?
Or as long as you are showing a profit, you just figure everything is OK and do not check anything?

I find it a lot less work to not use an oddsline and basically get enough winners to show a profit without regard if each bet is optimal.

[Overlay]

The test of an ideal odds line would be that it achieves the twin goals of predicting the winning frequency of horses (that is, horses at each fair-odds level would win at an overall rate corresponding to their fair odds), and of maintaining that winning percentage regardless of the final toteboard odds for each horse (for example, horses with fair odds of 3-1 would win 25% of the time, whether they went off at actual odds of 3-5 or 10-1). However, since the toteboard odds reflect the public's processing of information about the winning chances of the horses that any individual bettor would not know or have access to, I think that the best the player can do is to rely to the greatest extent possible on handicapping factors that produce such strong positive or negative results that they qualify as independent variables that don't depend on interaction with other factors for their predictive effectiveness; and also to require that the toteboard odds provide a percentage overlay premium as a margin of error to compensate for what the bettor doesn't know. (That is, for example, if a 50% premium were being required, a horse with fair odds of 2-1 would not qualify for a bet until it was at least 3-1, rather than 5-2.) A premium can also help compensate to a certain degree for odds that drop at the last flash.

I think that an advantage of an objective rating method (as opposed to subjective opinion) is that, if you subdivide the elements in your handicapping into categories (condition, class, speed, pace, etc.), with numerical weights assigned to each category, it allows adjustment of those weights based on which handicapping category is not performing as expected, if results indicate a statistically significant trend away from accuracy/profitability.

[badcompany]

If you do the math you will find that if a $2 bettor betting win only had won with horses of which 15 went off at odds of 1-5, of which 13 won; 8 went off at 2-5, of which 6 won; and 18 went off at 1-2, of which 12 won. The bettor would show a profit of $.0437 on the dollar or less than the track’s guaranteed $.05 on the dollar. Although the bettor's actual profit wasn’t negative, the profit against the track’s guarantee was and the track achieved a $.0065 profit on the dollar from the bettor. Very bad wagers.

LOL

You gotta love these "systems" that have you risking after tax dollars to get you what you'd get from the bank without risking your principle.

[Cratos]

The track does guarantee a minimum payoff of $2.10, but that's only on winning bets. What Quirin's $NET figure is showing is that for every $2.00 bet that was placed on those horses (winning or losing), the bettor received back an average of $2.20, $2.17, or $2.09 (depending on the odds range), based on the ratio of the proceeds from winning bets to the losses on losing bets. Any $NET figure above $2.00 represents actual profit.


I am not trying to be cynical or very refine to detail, but your assertion “that's only on winning bets” can only be on winning bets whether they are win, place, or show.

Furthermore I did the same calculations (winning or losing) as Quirin and amortized the winnings over the total wagers and got the resultants. Also 1-5 odds, 2-5 odds, and 1-2, are “giving odds” and in the long run if you are not in the high ninety win percent you will lose money.

[Overlay]

Sorry I wasn't precise in my choice of words. Quirin used $NET in the context of win bets, and it was the return on win bets that I was referring to, although the $2.10 minimum payoff applies to any type of $2.00 wager (win, place, or show). I agree with you about the difficulty of achieving the high winning percentage on wagers that is needed to show an overall profit with very low-odds horses, but, as I interpreted Quirin's data, his figures were indicating that it was possible to do so for those specific win-bet odds ranges.

[BMeadow]

"Again, this is 1970 data, but the results of the statistical tests that Quirin applied to it were presented as indicating that the profitability he found for those odds ranges would continue into the future."

Bad news--Our study of 200,000 favorites reported in the January 2000 issue of Meadow's Racing Monthly, researched by Jim Cramer of Handicappers Data Warehouse, showed there was NO odds range that, when played blindly, proved profitable. The best results were from 1-5 and 2-5 shots, each of which returned 0.86 per $1 bet for a 14% loss. All other odds ranges (from 3-5 through 6-1) returned results of 0.78 to 0.83.

The favorite-longshot bias of decades ago is just about done, although there is a slight bias in favor of the lowest-odds horses (below 1-2) and against the very longest horses (8-1 and up).

[RXB]

If odds are determined by how many times would this horse win if this race were run 100 times then that super imposes that the SAME horses, (ie., field), track, distance, surface, weather and etc are identical. After today it will never happen again. There are no multiple events.

I AM NOT A MATH MAJOR. IF YOU ARE AND I AM WRONG PLEASE ADVISE. :)

I try to make the most money I can each race. If I say #1 and #2 will win and the track says #4 and #5,.....one of us is going into the next race with a smile on our face.

If I say #1 and #2 and the track says #1 and #2,..if I can not find a way to at least double my money then I add my alotment for that race to the next and move on.

Why do we handicap except to seperate the contenders from those who are not. We have assigned odds when we make our picks. Next step is to make money.

SUPPOSE, you have narrowed down a ten horse field to two contenders and one maybe and the two contenders are close.

Seems easy. 2:1, 2 1/2 : 1 & 3:1 and everything else is 20:1 (I am not interested in the rest)

If I have missed the boat will someone throw me a life jacket? :) I do not understand the notion of making your own "line".

This is nonsensical. If there were nothing but "contenders" and "non-contenders" it would be so easy, wouldn't it? Now, exactly where are you drawing this magical line between your "contenders" and your "non-contenders" and how is it that anything on one side of this magical line is 3/1 or less and anything on the other side of it is 20/1?

And what is this "no multiple events" bit? This just in: when you make a bet on a particular possible outcome you had better be able to assign a sound estimate of the probability of said outcome actually occurring. Whether you define it as 40% or 40 times out of 100 is just semantics.

[Greyfox]

"Bad news--Our study of 200,000 favorites reported in the January 2000 issue of Meadow's Racing Monthly, researched by Jim Cramer of Handicappers Data Warehouse, showed there was NO odds range that, when played blindly, proved profitable. The best results were from 1-5 and 2-5 shots, each of which returned 0.86 per $1 bet for a 14% loss. All other odds ranges (from 3-5 through 6-1) returned results of 0.78 to 0.83.

The favorite-longshot bias of decades ago is just about done, although there is a slight bias in favor of the lowest-odds horses (below 1-2) and against the very longest horses (8-1 and up).

:ThmbUp: :ThmbUp: Thankyou for the info.