Long shots or Favorites

[Capper Al]

Originally Posted by Overlay

Ideally, if you have a full-field fair-odds line, you should know both how much and why you like each of the horses in the race, and at what odds they become favorable (overlaid) or unfavorable (underlaid) betting propositions (regardless of whether you rate them as the one most likely winner of the race or not).

Also, if the favorite is underlaid, one or more of the other horses in the race must be overlaid. That doesn't mean that you necessarily have to bet against the favorite (depending on your tolerance for risk). You can always pass the race and wait for a situation where the horse that you feel is the best is at odds that are higher than they should be (whether the horse is the favorite or not), and you can account either for the reason why the horse is not being bet, or for the false reasons why the other horses are being bet, or both (provided you're not of the conspiratorial mindset that, just by virtue of the fact that the horse is an overlay, it means that "they" know or have decreed that the horse will not win today).

Not so. Favorites use a typical comprehensive formula like e-Ponies. My long shots use another formula. The two don't mix. It's like adding up the apples for favorites, and adding up the oranges for long shots.

[Dark Horse]

Originally Posted by Capper Al

Not so. Favorites use a typical comprehensive formula like e-Ponies. My long shots use another formula. The two don't mix. It's like adding up the apples for favorites, and adding up the oranges for long shots.

There's your problem. An evaluation process has more merit if it is integrated for the entire field, rather than patched together with bits and pieces. Otherwise there is no true way to recognize value. To me it seems that your underlying question is not so much WHO to bet, but HOW to bet.

Often all I do is toss out a number of horses. If that turns the playing field positive, I'm interested in the race. If not, I pass. It doesn't matter who provides the value. Then the remaining horses are shown with their individual, combined and average overlays. Depending on these factors I then choose an exotic wagering formula, that usually includes backup exotic wagers. For example a trifecta backed up by a couple of exactas. I programmed these formulas, but had to learn them one by one before understanding which one was best for which situation. That's the learning curve, just as handicapping has a learning curve.

So many things can go wrong during a race that it's crucial to learn HOW to bet rather than rely on WHO wins. I'm often completely wrong about the winner, but end up with a nice profit on the race just the same.

Today's example from R7 at SA. Field reduced to 5 horses. Bets: 1-3/1-3-7-8-2 with 1-3-7-8-2/1-3 (two exactas), and 1-3/1-3/7-8-2. Note that all three bets can win. The favorites happen to be the 1 and 3, which lowers payouts for them coming in first and second, but opens the door to all three tickets winning. The result: 8-1-3 paid just one exacta, but at 41/1 was well worth it. Not a big payout by any means, but that's not something I can control. Betting formulas do open the door wide to juicy payouts, but also deliver at lower levels. All I really did was toss out four horses, and three of those had double digit odds. If you had asked me before the race WHO would win I would have had no clue.

[salty]

I'm confused on what your question is? It is in fact so vague I don't know how anyone has even managed to come up with any answer. I've never used computer handicapping software, so maybe that's why?


I look at the probable race shape then decide who has the advantage. If the race looks to set up for the favorite then they will probably win. You know sometimes no matter how much handicapping you do the money ends up being amazingly accurate. If the favorites running style can be compromised by the way the race sets up then it's a good idea to go with the long shot. Best answer I have for what I think you are asking.

Base the decision on the race shape.

[salty]

Dark horse,

That example makes absolutely no sense. Your are risking what? $22 on exotics that in many cases return negative to what you bet?

The exacta paid $41 which is $3 shy of getting even money on your bet.

[Dark Horse]

Originally Posted by salty

Dark horse,

That example makes absolutely no sense. Your are risking what? $22 on exotics that in many cases return negative to what you bet?

The exacta paid $41 which is $3 shy of getting even money on your bet.

Quite a typical reaction for this board. I'm just as happy to suggest a better way of betting as to let folks like you gloat in their own ignorance. It makes no difference to me. The ROI was 87%, if you must know. When the race is in clearer focus, there are leaner ways of betting; but this race fit this standard and proven betting formula quite nicely.

[salty]

I'm ignorant of what? The fact that you have to get lucky more than half the times you make that bet to turn a profit?

That exacta only returned so much because it was a total question about what the 8 would do when it tried grass for the first time. Also the 10 took a lot of money for some strange reason.

[Dark Horse]

Originally Posted by salty

I'm ignorant of what? The fact that you have to get lucky more than half the times you make that bet to turn a profit?

That exacta only returned so much because it was a total question about what the 8 would do when it tried grass for the first time. Also the 10 took a lot of money for some strange reason.

Pointless debate. Why? Because you have no knowledge of exotic wagering formulas. It would take, at least, a few months of serious study to become proficient with those. Your criticism is on par with telling me that I couldn't reduce a 9 horse field to 5 contenders. Your more than welcome to that opinion.

Oh, how did you do with this race?

[Dark Horse]

As to ROI. The 'goal' was to win the trifecta. The two exactas were backups. One could play the same formula without the trifecta, and increase ROI. The prices for the two favorites were 5/2 (240) and 3/1 (310), so if they finished first and second the payout would be high enough (paid twice! -with the trifecta on the table ... ) to warrant a ticket with five contenders. Other options, such as the winning exacta, would involve higher payouts. So the only condition for that, aside from the correct reduction to five contenders, was that one of the two favorites finished first or second. Luck had nothing to do with it. It's the art of cashing tickets. And the part of horse racing most widely ignored.

[salty]

Right the art of cashing tickets. So you have 0 opinion on the race and you pick your nose and hope to find gold. Good plan.

Yeah I have no knowledge of exotic wagering formulas, I must be a caveman.

If the goal was to hit the trI shouldn't the bet be

1,3//1,2,3,7,8//1,2,3,7,8

1,2,3,7,8 // 1,3 // 1,2,3,7,8

1,2,3,7,8//1,2,3,7,8//1,3

??????

[Dark Horse]

Originally Posted by salty

Right the art of cashing tickets. So you have 0 opinion on the race and you pick your nose and hope to find gold. Good plan.

Yeah I have no knowledge of exotic wagering formulas, I must be a caveman.

If the goal was to hit the trI shouldn't the bet be

1,3//1,2,3,7,8//1,2,3,7,8

1,2,3,7,8 // 1,3 // 1,2,3,7,8

1,2,3,7,8//1,2,3,7,8//1,3

??????

Wow, you're able to fill out a winning ticket after the race.

Your play is one-dimensional and only wins if you win the trifecta. My method is much leaner. The trifecta formula is backed up by exactas. Just as a superfecta formula would be backed up by trifectas.

It's over your head, at this point, but you could always learn. Problem is that you think you already know it all. That pretty much closes the door. On yourself.

[traynor]

Originally Posted by Capper Al

I can pick favorites and I can pick long shots, but I can't tell who's most likely to win. Value odds makes it easier to pick long shots. Yet, this is no indication that the favorite is not going to win. I'm guessing that perfecting the art of the false favorite would help. My method for picking favorites and picking long shots are different. This means they are not on the same scale for comparison. Any ideas on how to handle this?

I think Dave Schwartz alluded to (or expressed) the inherent difficulties involved in combining attributes. That is, finding a probable win frequency for your favorite is fairly simple--but only when it is viewed in isolation.

A more complex approach might be to isolate (or "layer") different circumstances with that favorite in combination with other attributes of other entries.

Simplistic example: Favorite wins X% of the time if (and only if) it is the lone early speed in the race. It may win so often under such circumstances that looking at overall figures is deceptive. Like Pareto, 80% of the profit is often in 20% of the races. Find that 20% and marginal methods can often turn into quite profitable methods.

[traynor]

Originally Posted by Overlay

Ideally, if you have a full-field fair-odds line, you should know both how much and why you like each of the horses in the race, and at what odds they become favorable (overlaid) or unfavorable (underlaid) betting propositions (regardless of whether you rate them as the one most likely winner of the race or not).

Also, if the favorite is underlaid, one or more of the other horses in the race must be overlaid. That doesn't mean that you necessarily have to bet against the favorite (depending on your tolerance for risk). You can always pass the race and wait for a situation where the horse that you feel is the best is at odds that are higher than they should be (whether the horse is the favorite or not), and you can account either for the reason why the horse is not being bet, or for the false reasons why the other horses are being bet, or both (provided you're not of the conspiratorial mindset that, just by virtue of the fact that the horse is an overlay, it means that "they" know or have decreed that the horse will not win today).

Easier said than done. To be accurate (and predictive) the relative probability of each horse's performance in a specific race should be made as a factor of the entire field as a group, relative to its inclusion in that specific group (and the individual and collective attributes of each of the other entries in that group). Not an easy task.

Probability values calculated as if each entry is the only entry in the race (ignoring the "confounding variables" present in the other entries) are much less predictive. Simple combining of values so calculated as if they represent composite reality is even less so.

[thaskalos]

Originally Posted by traynor

Easier said than done. To be accurate (and predictive) the relative probability of each horse's performance in a specific race should be made as a factor of the entire field as a group, relative to its inclusion in that specific group (and the individual and collective attributes of each of the other entries in that group). Not an easy task.

Probability values calculated as if each entry is the only entry in the race (ignoring the "confounding variables" present in the other entries) are much less predictive. Simple combining of values so calculated as if they represent composite reality is even less so.

EXCELLENT!

[Robert Fischer]

Originally Posted by Capper Al

I can pick favorites and I can pick long shots, but I can't tell who's most likely to win. Value odds makes it easier to pick long shots. Yet, this is no indication that the favorite is not going to win. I'm guessing that perfecting the art of the false favorite would help. My method for picking favorites and picking long shots are different. This means they are not on the same scale for comparison. Any ideas on how to handle this?

Start with the favorite.
By definition, the favorite has the largest market share of each pool.

[Capper Al]

Originally Posted by traynor

Easier said than done. To be accurate (and predictive) the relative probability of each horse's performance in a specific race should be made as a factor of the entire field as a group, relative to its inclusion in that specific group (and the individual and collective attributes of each of the other entries in that group). Not an easy task.

Probability values calculated as if each entry is the only entry in the race (ignoring the "confounding variables" present in the other entries) are much less predictive. Simple combining of values so calculated as if they represent composite reality is even less so.

A point in the thread 'The Morning Line is Just Wrong' that I was trying to make was that the factors aren't really homogenous in a way that lends itself into a total view of the race. When we look at speed or class or whatever, we are only looking at the tips of the icebergs.