Angle/Database players with multiple contenders

[Cesario!]

I was wondering what the players who engage in database/angle handicapping-- based on a successful recognized patten that projects improved performance--do in the situtation where you find multiple longshot contenders.

Do you bet them all if the individual price is appropriate based on your target odds? Or do you combine the prices to insure a single overlay price (e.g. are 14-1 and 6-1 still overlays at 7-1 and 3-1...a la Barry Meadow?) Or, do you dutch them...or perhaps, do you do something else?

I've always thought that these standard solutions were insufficient. For the first method, the presence of multiple bets means that one (or several) will lose, thus reducing the inherent probablilities somewhat -- that seemingly has to affect bet size to some degree. The second, however, seems too strict and treats a race as more important than an overall series of races. The third method (dutching) seems to reduce the potential payout when you are talking about extreme longshots that only pay out over a long series -- why should you lower the return simply b/c it came in a race with multiple qualifiers?

A bit rambling here...I might try to clarify a bit more later.

[njcurveball]

You ask a Masters level question.

If you have a spot play system that generates 10% winners at a very large mutuel. Anything other than flat betting runs the risk of draining the profits.

Obviously if it has a very high winning percentage there will not be many races with multiple qualifiers.

I think the biggest mistake is to bet the lower odds horse for more. I would say the most profitable approach given that a series of bets is going to win is simply to flat bet them all for the same amount.


A more advanced approach can look to see if the horses come in the money often as well. If they do and they are nice longshots. Exactas and Trifectas become profitable and a good way to use them.

Jim

[Cesario!]

You ask a Masters level question.

If you have a spot play system that generates 10% winners at a very large mutuel. Anything other than flat betting runs the risk of draining the profits.

Obviously if it has a very high winning percentage there will not be many races with multiple qualifiers.

I think the biggest mistake is to bet the lower odds horse for more. I would say the most profitable approach given that a series of bets is going to win is simply to flat bet them all for the same amount.


A more advanced approach can look to see if the horses come in the money often as well. If they do and they are nice longshots. Exactas and Trifectas become profitable and a good way to use them.

Jim

Jim,

Thanks for the response. And that's exactly the 'system' I've developed -- I get about 11-12% winners at high mutuels, which gives me a ROI of about 1.15 -- tested at about 600 bets. I haven't quite determined the average number of bets required for "return to the mean" however -- within a series of 100 bets, it's been as high as 1.40 and as low as .87. But, that's outside the topic...sort of...I think. I'm sure there's a statistical way to measure the expected variation that beyond my current expertise.

Anyhow, I have been using the flat bet approach which you suggested. I have, as I mentioned, been concerned about the effect of betting too many contenders in a race. The more I think about it, the only way to tell may be through further data analysis --- checking long-term the effect of betting multiples in a race vs. passing the race altogether. I'll happily admit the breakthrough in my gambling career came on the day I stopped trying to split hairs.

BTW, these horses win about 12% of the time, but also finish 2nd (I haven't tracked third) only about 20% of the time -- something I discovered when I researched the benefits of place betting (which I determined was not beneficial). I'll have to do some backfitting to check out exacta possibilities, though.

Seth

[Dave Schwartz]

My experience is that when you find a spot play system that generates an occasional set of multiple plays in one race, one usually finds that the races with more than one play are less profitable than those with a single play.

That being said, I would probably still try to bet them all.

Dave

[DaveP]

I have a strike rate of around 15% and this comes up everyday for me .. I bet everything to level stakes, no matter how many qualifiers there are in a race. It works out the same, regarding strike rate and returns and reduces the long losing run.

[Tom]

I have no problem playing multples in the same race. I am playing the spot, not the race. If I chose between them, you just know I would alwys choose the wrong one to play - Newton's Fouth law.

[delayjf]

I'm inclinded to go for the kill. I would play more on the longer odds horses and might use exactas on the lower price horses to break even or show a slight profit. Given the odds you quote, betting 200 on this race, I'd go 150 on the 14-1 and 50 on the 6-1.

[singunner]

Wouldn't you bet more on the one whose odds are more underrated than the other? And wouldn't you do so in a statistically derived manner? I suppose you may not have access to that data, but otherwise that should be the appropriate move.

[singunner]

It appears I spent too long reading the thread.

If you want the horseplayer-sounding answer, do what Delayjf said.

If you want the academic-sounding answer, do what I said.

That is, we basically said the same thing.

[Niko]

If there's 2 longshot qualifiers in a race-no problem, I bet both. If there's 3 or more according to my handicapping I used to try to play them all and go for the kill, but I was the one left bleeding. For some reason if there's too many it means I don't understand the race and some other horse always seems to win. So I concur with Dave.

[traynor]

The issue most have with multiple bets in the same race is exemplified by the old horsey saying, "You are betting against yourself." With liberal infusions of Beyer philosophy, dutching seems somehow aberrant.

The key point is that when you bet multiple entries in the same race, you are no longer making a $2 base wager; your base wager is $2 times however many you are betting. The issue of some wagers being "guaranteed to lose" is not relevant; multiple wagers are a single, aggregate wager, and results should be calculated as such.

Example; suppose you have a model of sufficient size to be relatively predictive, that indicates in situations in which your selections meet very specific criteria in conjunction with minimum odds of whatever, your $4 (for two entries) or $6 (for three entries) wagers return $5 and $7.50, respectively.

The situation is one of "framing"; if you view the wagers as a single bet, they are obviously profitable, and multiple entry flat wagering or dutching is both appropriate and profitable. If you view the wagers as either half or two-thirds being guaranteed losers, you miss a LOT of opportunities for generous reward.

There is a related question often posed in graduate business classes; you are given a mass of figures, imbedded in which is the marketing report that says for every $1 you put into a specific ad campaign, $1.40 net revenue is generated. Business majors tend to focus on other areas; liquidity ratios, profitability ratios, earnings per share, etc. when constructing a hypothetical budget. In that scenario, the ad campaign should be allocated the maximum available (with certain minor restrictions and caveats); every other answer is wrong.

The scenario is identical to wagering on multiple entries; if you are motivated by profit, multiple wagers frequently offer the equivalent of a "cash cow" for handicappers and bettors.
Good Luck

[Robert Fischer]

semi-automatic handicapping-
My personal style with spot plays is to take a play that the filters brought into focus, and handicap the race with the intent of evaluating this specific opportunity. Of course, this costs valuable time, and it isn't always extremely helpful (what if your spot play filter included much of what you would be looking for anyway).

quick approach
With multiple qualifiers, my quick approach would be to examine the target odds of each member of the group. Looking at applying the spot wager to the group, I would want even the worst price of qualifiers to fit my standard.
Say I demand 5-1. I have three qualifiers tote odds 20-1 , 15-1 , and 30-1...
- If I then made three $2 win bets the new odds are 6.67-1, 5-1, and 10-1.

If I had time I would handicap the multiple spot plays. I would look at my database , specifically at the sample of multiple qualifier-races...


another consideration is that if you just found multiple live longshots, then take a look at the roi for exotics with that sample.

[Cesario!]

Thanks for all your replies. I'm fascinated with the concept of the "inverted dutch" -- the multiple plays with the higher proportion on the higher odds. Horses with odds from 5-1 to 8-1 are good for me for two reasons:

(1) They tend to keep my bankroll steady and also tend to keep me in the game -- I handicap by hand, so I need to have a positive mindset.

(2) They also allow bankroll churn -- as long as the ROI is positive for a type of bet, I want to get as much money through as I can. The day I realized that it's better to have a system ROI of 1.15 with 4 times as many plays as one with 1.50 was a powerful one. I finally understand what Dave Schwartz has been talking about for all these years.

Anyhow, my bankroll tends to move to new levels, however, with the big hits (the 55-1, the 35-1, etc.).that occur more frequently then the odds suggest....well, obviously.:D My tracking program (BASE) does not track by odds category -- something that I plan to implement in my data collection in the future. I'd be fascinated to see what the ROI is for particular odds ranges. This could be a good way to accomplish my goals and maximize the ROI from this style of playing. The more I think about it, this is information that I need to make these sorts of decisions. I was initially hesitant because it puts weight on the public's assessment for longshot factors.

----

Traynor --

In response to your comment, I don't quite understand why you consider the return from a race as a meaningful statisitic -- isn't the only meaningful statistic your return from a series of races?

Seth

[njcurveball]

Well you have certainly provoked much thought in this thread.

I think we all lose something going through school, and that is the ability to "just know the answer".

It is always about proving your work and "looking smart". Many of the answers here would definitely get an A in Math Class.

The problem is that the track is a hybrid of that. The longer priced horses may look like a sweeter proposition, but the crowd is not that far off in the majority of races.

Most of the time the ML 6-1 shot is on the board at 2-1, it is the ML makers fault, not the crowd being smart.

Same thing when the ML 6-1 is 20-1 on the board. The quickest way to ruin a spot play is to try to overthink the bet.

The method I use is simply to advance bet. I didn't put any special criteria on the bets when I figured out it was a winning play.

You do not know if anything other than the spot play is profitable, so you are really taking a leap of faith.

Also, if you advance bet, you wind up with your usual amount on the $100 horse that certainly doesn't look as pretty when you see 50-1 on the board and start seeing all the flaws in the past performances.

As for 2 horses or more in a race, I watched so many Sartin players make money betting 2 or more horses in a race I was able to give up my "horsey training", that it was betting against myself.

Jim

[bigmack]

Knowing the parameters of strong plays that continually yield a healthy roi from a multitude of angles has become my bread & butter. I'll normally single play the spot and dig deeper within that race for the exotics.

Having the technology to sort & pick-out angles that I know work well is a good place to be and REALLY cuts down on the amount of work.

[Nosebob]

While reading this thread, I was reminded of the group overlay method covered in Dick Mitchell’s book "Commonsense Betting".

Mitchell says, "If your estimate of the group probability for your contenders exceeds the public’s, you have a positive-expectation situation that can be exploited."

Your group probability can be calculated as the sum of individual probability of your contenders (if you have established an odds line), or from the fact that you know your top two contenders win 55% of the time and your top three win 72% as an example.

The public’s probabilities can be easily determined from the tote board odds.

I must admit that I have not used this method, primarily because of the calculations required so near post time.

If anyone is interested in more information on the subject, it is covered in Mitchell’s book, starting on page 101.

Nosebob

[Dave Schwartz]

Mitchell says, "If your estimate of the group probability for your contenders exceeds the public’s, you have a positive-expectation situation that can be exploited."

Precisely.

What difference does it make if you get 30% @ 3/1 from one horse, or 30% winners betting across 3 horses per race at 11-1 each?


Dave Schwartz

[Overlay]

While reading this thread, I was reminded of the group overlay method covered in Dick Mitchell’s book "Commonsense Betting".

The thing I like about the group overlay method is the possibility of including a horse that you judge to have a high true probability of winning within the group (even if the high-probability horse is a slight underlay when considered by itself), while actually increasing the expected rate of return from the group wager.

[traynor]

Thanks for all your replies. I'm fascinated with the concept of the "inverted dutch" -- the multiple plays with the higher proportion on the higher odds. Horses with odds from 5-1 to 8-1 are good for me for two reasons:

(1) They tend to keep my bankroll steady and also tend to keep me in the game -- I handicap by hand, so I need to have a positive mindset.

(2) They also allow bankroll churn -- as long as the ROI is positive for a type of bet, I want to get as much money through as I can. The day I realized that it's better to have a system ROI of 1.15 with 4 times as many plays as one with 1.50 was a powerful one. I finally understand what Dave Schwartz has been talking about for all these years.

Anyhow, my bankroll tends to move to new levels, however, with the big hits (the 55-1, the 35-1, etc.).that occur more frequently then the odds suggest....well, obviously.:D My tracking program (BASE) does not track by odds category -- something that I plan to implement in my data collection in the future. I'd be fascinated to see what the ROI is for particular odds ranges. This could be a good way to accomplish my goals and maximize the ROI from this style of playing. The more I think about it, this is information that I need to make these sorts of decisions. I was initially hesitant because it puts weight on the public's assessment for longshot factors.

----

Traynor --

In response to your comment, I don't quite understand why you consider the return from a race as a meaningful statisitic -- isn't the only meaningful statistic your return from a series of races?

Seth

Exactly. If I gave the impression that I was using single races for the criteria, the error was mine. It is essential that the process be effective over time; the results of individual races are almost irrelevant.

The primary difference is that in modeling the effect of wagers over time, multiple wagers (in the same race) are considered a single wager, rather than as 3 wagers at odds of 5-1, 6-1, and 7-1 (or whatever). The model suggests that as long as all three are over a specific odds minimum, flat bets to win on all three are profitable over time. That is fundamentally different than layering by odds range, unless you are using a stable selection method that is consistent enough to be meaningful.

For example, if your (relatively consistent) selection method (translation-not tweaked, adjusted, "finetuned," or changed continually) projects 3 entries as first, second, and third AND the available odds on all three of those entries are above a specific value THEN the constraints have been met.

If the selection method is tweaked frequently, the models tend to be flaky; if the selection criteria and projections are relatively consistent, calculating an odds figure that can be used as a minimum is easy. The important thing to remember is that both componenets must fit within the constraints; the selections must be highly rated and the odds must be over some established minimum.

The relevance to multiple wagering is that when the conditions are met, it creates a new "category" of wager that is more than the sum of the components. That is, the sum of the individual returns and probabilities of success for the components may be exceeded by combining them. It is a fairly common concept in business and finance, especially prevalent in hostile takeovers and leveraged buy-outs, defined as "creating synergies."

When individual wagers are regarded as a series of data points in a cash flow, the synergistic effect of certain combinations becomes evident. One of the most productive of those synergies is multiple win bets in the same race.

Caveats; this is not a new idea, nor is it mine. It has been used by wagering professionals for many, many years. The concept of synergies is similarly well-documented, well-researched, and firmly established. Synergies apply as well to exacta wagering as to win wagering. The old-time "dutching" concept popularized by Sartin and others, of adjusting wager amounts based on post-time odds, does not seem to be as productive as using a minimum odds baseline and flat betting selections that fit the criteria.
Good Luck

[traynor]

While reading this thread, I was reminded of the group overlay method covered in Dick Mitchell’s book "Commonsense Betting".

Mitchell says, "If your estimate of the group probability for your contenders exceeds the public’s, you have a positive-expectation situation that can be exploited."

Your group probability can be calculated as the sum of individual probability of your contenders (if you have established an odds line), or from the fact that you know your top two contenders win 55% of the time and your top three win 72% as an example.

The public’s probabilities can be easily determined from the tote board odds.

I must admit that I have not used this method, primarily because of the calculations required so near post time.

If anyone is interested in more information on the subject, it is covered in Mitchell’s book, starting on page 101.

Nosebob



I attended a number of seminars at which Dick Mitchell explained his concepts. He had a lot of very useful insights (including the idea of back-wheeling the favorite in exactas at Santa Anita). The point he missed is that the effect of combining individual components can exceed the sum of the components; the aggregate wager represents a new data item that must be tracked, rather than the simplistic sum of the components.

The "sum of the parts" method applied to handicapping exactas is equally flawed. That is, simplistically calculating the probability of exacta AB as the product of the win probability of A multiplied by the place probability of B looks a LOT better on paper than it does in the real world.

The real question is not the sum of the probabilities; that is only a general guideline. The real question is "What is the probability of Entry B placing in this race if Entry A wins?" Similarly, what is the probability Entry A will win if it fails to get an uncontested early lead in this races--guaranteed by Entry C, a dedicated front runner with tons of early speed?
Good Luck

[DaveP]

The "sum of the parts" method applied to handicapping exactas is equally flawed. That is, simplistically calculating the probability of exacta AB as the product of the win probability of A multiplied by the place probability of B looks a LOT better on paper than it does in the real world.

The real question is not the sum of the probabilities; that is only a general guideline. The real question is "What is the probability of Entry B placing in this race if Entry A wins?" Similarly, what is the probability Entry A will win if it fails to get an uncontested early lead in this races--guaranteed by Entry C, a dedicated front runner with tons of early speed?

Good point traynor, treating dependant outcomes as independant outcomes, can be a big mistake.

[PriceAnProbability]

My experience is that when you find a spot play system that generates an occasional set of multiple plays in one race, one usually finds that the races with more than one play are less profitable than those with a single play.

That being said, I would probably still try to bet them all.

Dave

The races are more profitable in the exotics, since value is being combined with value.

I'm surprised someone who claims substantial winnings every year wouldn't see it the same way.