Odds lines are useless

[Overlay]

Originally Posted by Capper Al

Let's look at what the odds lines are good for.

1. Ordering your selections
2. Expressing the selections in an opinionated way numerically.
3. Understanding what the public might think.
4. Indicating that a horse deserves a second look if the tote odds are too high or too low.

That's all I can think of.

Originally Posted by cashmachine

5) Computing whether horse is overlay or not (given probability of win)

6. Determining the optimum size and distribution of win bets or exotic wagers based on the disparity between your assessment and the public's (that is, the degree to which a horse, a group of horses, or a multi-horse exotic combination is overlaid in comparison to your line).

[Sapio]

Originally Posted by Overlay

6. Determining the optimum size and distribution of win bets or exotic wagers based on the disparity between your assessment and the public's (that is, the degree to which a horse, a group of horses, or a multi-horse exotic combination is overlaid in comparison to your line).

7. Allowing the data to speak for itself in a systematic manner.

[thaskalos]

Originally Posted by Sapio

7. Allowing the data to speak for itself in a systematic manner.

8. Instilling discipline into your game by keeping you from chasing your losses, by convincing yourself that wagers are better than they really are. There is a tendency for us to press when we are losing, in an effort to get even in a hurry...and we often look for excuses to play some marginal horses that we would never bet on when we are ahead. With an odds line, there is no ambiguity. A bet is either a sound wager...or it isn't.

[JoeLong]

What really matters is... Your own odds line.

From even money out to about 6/1 do your odds line mimic reality? For instance do your 4/1 shots win about 20% of their races, do the horses you peg at 25% probability win at around 3/1. Do your even money horses win about half their races? If so you have a workable odds-line which can be used to exploit the win pool.

It would be an achievement beyond current computing to solve the handicapping dilemma in a match duel, let alone running against multiple contenders, so that rules out game theory and leaves us with exploitation as the only method to beat the para-mutuals.

Handicapping is a zero-sum game, to win someone else has to lose, that being the case the only way to beat that type of game is with an edge which leads to a dynamic that requires the player to be risk adverse. The only way I know of being risk adverse is to have some notion of the odds being offered. No?

[traynor]

Originally Posted by JoeLong

What really matters is... Your own odds line.

From even money out to about 6/1 do your odds line mimic reality? For instance do your 4/1 shots win about 20% of their races, do the horses you peg at 25% probability win at around 3/1. Do your even money horses win about half their races? If so you have a workable odds-line which can be used to exploit the win pool.

It would be an achievement beyond current computing to solve the handicapping dilemma in a match duel, let alone running against multiple contenders, so that rules out game theory and leaves us with exploitation as the only method to beat the para-mutuals.

Handicapping is a zero-sum game, to win someone else has to lose, that being the case the only way to beat that type of game is with an edge which leads to a dynamic that requires the player to be risk adverse. The only way I know of being risk adverse is to have some notion of the odds being offered. No?

I don't understand. What other than relative probability of winning would be used to construct an odds line?

[AndyC]

Originally Posted by JoeLong

Handicapping is a zero-sum game.......

We should all be so lucky. Horse racing is actually a negative sum game.

[JoeLong]

Originally Posted by traynor

I don't understand. What other than relative probability of winning would be used to construct an odds line?

Well for games which can be solved the best way to play is game theory optimal GTO which of course incorporates probability, but since horse racing isn't solvable the next best thing is to exploit any perceived weakness above your break even expectation EV.

[ultracapper]

Originally Posted by JoeLong

Well for games which can be solved the best way to play is game theory optimal GTO which of course incorporates probability, but since horse racing isn't solvable the next best thing is to exploit any perceived weakness above your break even expectation EV.

Are you implying that no horse race is solvable?

[ultracapper]

Originally Posted by traynor

I don't understand. What other than relative probability of winning would be used to construct an odds line?

ML makers construct odds lines based on where the money is expected to go. Winning probability should not be a factor at all. How will the public bet? That is the ML maker's job to anticipate.

[Sapio]

Originally Posted by ultracapper

ML makers construct odds lines based on where the money is expected to go. Winning probability should not be a factor at all. How will the public bet? That is the ML maker's job to anticipate.

Hi ultracapper

That is what is said and believed by many, but I highly doubt that is what they do. They are not that analytical.

Thomas Sapio

[ultracapper]

I've never really tried, but I don't think it would be all that difficult to make a fairly accurate ML. High Beyers, lots of 1s and 2s in running lines, connections, annual and lifetime records, bullet and other fast works, class, these are the things that handicappers weigh heavily and attract money. You got Jon White telling us about some subtle trouble a horse had in his previous race as his reason for making a horse a ML fav drives me crazy.

[misscashalot]

finding anomalies using track ml in conjunction with place odds work well.
follow the $ ... the board speaks volumes using morning lines.

[traynor]

Originally Posted by ultracapper

ML makers construct odds lines based on where the money is expected to go. Winning probability should not be a factor at all. How will the public bet? That is the ML maker's job to anticipate.

What does a morning line have to do with an odds line for wagering? Two very different concepts.

I understand that some believe a morning line can be tweaked with arcane algorithms to become more predictive. The kindest thing I can say about that is that I don't believe it is especially useful. That is not a spurious comment.

[traynor]

An odds line (used for wagering) "should" be based on empirical data indicating the relative probability of each entry winning (or placing) in that specific race. Pretty simple stuff in theory, but not so easy in practice.

In fact, "empirical data indicating the relative probability of each entry winning (or placing) in that specific race" may well be the most valuable piece of information a serious bettor can have.

[ezpace]

algo = last race finish position(within45 days and3min odds today.. compare