Odds lines and entries

[banacek]

I have a home grown software that gives an odds line for each horse. I am analyzing the data from last year and am not sure how to deal with entries.

Say I have #1 at odds of 9-1 (10%) probability and 1A at odds of 4-1 (20%) probability. It seems logical to pool them and call the entry 2.33-1 (30% probability) , but in real life I am not real comfortable with it.

I generally don't pool the probabilities when I make wagers - I think each horse should stand on his own. But when analyzing the effectiveness of my odds line it seems logical to pool.

How do others who make odds lines deal with this?

[Robert Fischer]

I pool them for a total of 30% if I believe they have a .1 and .2 individual probability. It comes down to how your are assigning the line.
A lot of people err on the safe side, and thats why your 7% horse and your 16.8% horse don't feel like a 30% combination. Nothing wrong with that.

[sjk]

As long as you think their results are independent (and I would expect them to be) the probabilites are additive.

[Dave Schwartz]

Banacek,

In HSH we have several options:

Best - Take the best of the entries
Worst - Take the worst of the entries
Average - Average the entries together
Both - Sum all the entries together


My experience indicates that, with the exception of high class races, the worst half of the entry usually wins the race. However, a recent study (done by someone else) indicated that, although the weaker half of an entry is the more likely winner, he wins a percentage which is closer to that of the best half!

Therefore, I use "best."

Under no circumstances could I recommend "Both."


As you begin to write code, you will see how much difficulty it is to treat an entry as "one horse." It forces one to do all the calculations for each horse and then pour those calculations into an entry engine of sorts.

In addition, it forces you to see a race with 8 horses but 7 betting interests as "different" than an 8-horse field. (This was a huge re-write for me.)

There is also the difficulty of stuydying past data. For one thing, you will find that as you study past data, there are places where the entries are studied individually. In so doing, I can see now other way that to award the full percentage of the pool wagered on both halves of the entry. This causes all of the P.I.V. (Pool Impact Values) to be skewed.


Dave

[banacek]

Thanks for the responses. I also wonder how horses that you have no rating for (First-timers, foreign, etc) are dealt with. Last year I gave them an equivalent percentage to their morning line (but adjusting it down to a 100% line) just to give them something and not overrate the other entries. It seemed to work out ok.

[BMeadow]

This is a simple math problem:

Imagine a three-horse race (two betting interests) where:

#1 has a 40% chance to win
#1A (his entrymate) has a 15% chance to win
#2 has a 45% chance to win

It should be obvious that the entry has a 55% chance to win. When making an odds line, simply add the probabilities of each member of the entry.

[singunner]

Of course, keep in mind that it costs double to bet on them both... If you bet on both of them, you need to be sure that you're not undercutting your necessary, umm, "underlay/overlay"? I don't have the terminology down on that one yet. You want to ensure that you look at not just "return*probability", but include "/investment"... I think.

That is, from your original, it's not 2.33-1 because you're spending extra money.

[PriceAnProbability]

I have a home grown software that gives an odds line for each horse. I am analyzing the data from last year and am not sure how to deal with entries.

Say I have #1 at odds of 9-1 (10%) probability and 1A at odds of 4-1 (20%) probability. It seems logical to pool them and call the entry 2.33-1 (30% probability) , but in real life I am not real comfortable with it.

I generally don't pool the probabilities when I make wagers - I think each horse should stand on his own. But when analyzing the effectiveness of my odds line it seems logical to pool.

How do others who make odds lines deal with this?

Rate the entry separately and don't consider it "value" unless one of the two horses offers it.

[PriceAnProbability]

This is a simple math problem:

Imagine a three-horse race (two betting interests) where:

#1 has a 40% chance to win
#1A (his entrymate) has a 15% chance to win
#2 has a 45% chance to win

It should be obvious that the entry has a 55% chance to win. When making an odds line, simply add the probabilities of each member of the entry.

But when measuring "value" only use the 40 percent number, not the 55 percent.

[PriceAnProbability]

Thanks for the responses. I also wonder how horses that you have no rating for (First-timers, foreign, etc) are dealt with. Last year I gave them an equivalent percentage to their morning line (but adjusting it down to a 100% line) just to give them something and not overrate the other entries. It seemed to work out ok.

Usually a class system (class 1, class 2, etc.) would work, with class 1 being on a par with the favorite, class 2 a moderate contender, etc.

[banacek]

That is, from your original, it's not 2.33-1 because you're spending extra money.

I'm not sure what you mean by "extra money". They are an entry.

[banacek]

Both - Sum all the entries together


Under no circumstances could I recommend "Both."





When making an odds line, simply add the probabilities of each member of the entry.

Dave Schwartz and Barry Meadow - pretty high level responders and they treat it differently!

This is why I can't make up my mind how to deal with it. In practice I ignore the other part of the entry, but adding them doesn't seem like it works in practice. But mathematically you should.

[PriceAnProbability]

Dave Schwartz and Barry Meadow - pretty high level responders and they treat it differently!

This is why I can't make up my mind how to deal with it. In practice I ignore the other part of the entry, but adding them doesn't seem like it works in practice. But mathematically you should.

It doesn't work in practice because rarely does a trainer try with both halves of the entry.

Don't mix the values except when calculating the odds line.

I use 1, 1a, and 1e to differentiate, and only occasionally will I use the "1e," like if you have two 4-1s coupled in a wide-open race and no other clear value.

[ezrabrooks]

Of course, keep in mind that it costs double to bet on them both... If you bet on both of them, you need to be sure that you're not undercutting your necessary, umm, "underlay/overlay"? I don't have the terminology down on that one yet. You want to ensure that you look at not just "return*probability", but include "/investment"... I think.

That is, from your original, it's not 2.33-1 because you're spending extra money.

I am lost here... it cost double to bet on what?

Ez

[banacek]

Don't mix the values except when calculating the odds line.

:confused: Now I'm even more confused.

[BetHorses!]

Dave Schwartz and Barry Meadow - pretty high level responders and they treat it differently!



Definitely not a knock against Dave but I add like Barry says, I feel thats the only way

[PriceAnProbability]

:confused: Now I'm even more confused.

Say you make half of the entry 5-2 (28 points) and 8-1 (11 points). The total is 39 points, or 8-5, but for "value" purposes the #1 should be 5-2 and #1a should be 8-1. Only the "#1e" (combined entry) is 8-5, and only in rare situations will that be legitimate (separate trainers or owners helps a bit).

The same also applies to uncoupled entries, but not to the same degree. When establishing which horses offer value, the odds on any entrymates should be completely discarded, although you should be willing to include the #1e in the second tier, just not on top.

[Robert Fischer]

It doesn't work in practice because rarely does a trainer try with both halves of the entry.
...


I think this statement by PnP represents the crux of the matter.



Those who suggest adding the probabilities of the entries(myself B.Meadows…) have already considered intent when first assigning the individual probabilities.

I can't speak for anyone else but I can say that if I assign a .1 probability and a .2 probability, then those are my probabilities. If one of the horses is a "rabbit" or any signs that the intent is anything other than to win, then of course I will not assign that part of the entry a 10% probability in the first place.

[PriceAnProbability]

I think this statement by PnP represents the crux of the matter.



Those who suggest adding the probabilities of the entries(myself B.Meadows…) have already considered intent when first assigning the individual probabilities.

I can't speak for anyone else but I can say that if I assign a .1 probability and a .2 probability, then those are my probabilities. If one of the horses is a "rabbit" or any signs that the intent is anything other than to win, then of course I will not assign that part of the entry a 10% probability in the first place.

How does one quantify "intent?"

My power ratings are objective so they wouldn't cover intent.

[Robert Fischer]

handicapping experience

and most of what has been done before will be done today , and tomorrow.



Super Trainer's favorite stakes-level closer attempts to run down todays stakes field over 9f today....
The entry(1a) is a decent 6f sprinter who has trouble staying....

probabilities 1- x%(lets say 25%) 1a- maybe 2-3%

working from the easy examples and getting a little finer when possible...

[singunner]

I am lost here... it cost double to bet on what?

Ez

If you're betting two horses, the bet costs twice as much. It's nothing at all like having a single horse at 30% and shouldn't be played that way. The odds may be cumulative form a mathematical standpoint, but you've still just got two horses with their own odds, and they should likely be played that way.

[ezrabrooks]

If you're betting two horses, the bet costs twice as much. It's nothing at all like having a single horse at 30% and shouldn't be played that way. The odds may be cumulative form a mathematical standpoint, but you've still just got two horses with their own odds, and they should likely be played that way.

I thought we were discussing coupled entries..

RP

[singunner]

Ahh. Guess I didn't read the first post well enough.

[PriceAnProbability]

I thought we were discussing coupled entries..

RP

He was.

[Dave Schwartz]

It should be obvious that the entry has a 55% chance to win. When making an odds line, simply add the probabilities of each member of the entry.

Ah yes. Horse racing is such a simple game. You just add up the columns and the best horse wins.

LOL - Yes, it should be obvious. And it should work that way.

Too bad it doesn't seem to work that way.


Regards,
Dave Schwartz

[Kelso]

[QUOTE=PriceAnProbability]

Say you make half of the entry 5-2 (28 points) and 8-1 (11 points).

The total is 39 points, or 8-5, but for "value" purposes the #1 should be 5-2 and #1a should be 8-1.

QUOTE]


In your example, do you consider a coupled betting overlay to begin at 3-1 or at 9-1? Thank you.

[PriceAnProbability]

[QUOTE=PriceAnProbability]

Say you make half of the entry 5-2 (28 points) and 8-1 (11 points).

The total is 39 points, or 8-5, but for "value" purposes the #1 should be 5-2 and #1a should be 8-1.

QUOTE]


In your example, do you consider a coupled betting overlay to begin at 3-1 or at 9-1? Thank you.

3-1 if #1 or both go, and 9-1 if only #1a goes. If the weaker half of the entry is intended to win, the stronger half will usually be stiffed anyway.

At 2-1 this entry would be a sucker bet.

The lone exception is when the entry has the top two or three power ratings, such as occurred in the 1986 Lexington at Belmont, when the three-horse entry of Glow, Manila (Eclipse winner that year), and Pillaster ran 1-2-3, paying $3.40, $3.80, and $2.80. The accountant I was sitting next to had $750 to place and show on the entry.

In that case, the entry is #1e and the combined value is used, but only underneath any other value horses in my top tier.