Kelly Horse Race or Ziemba Roulette

[jfdinneen]

Originally Posted by Dave Schwartz

John,
Correct me if I am wrong, but isn't one of the concepts here based upon the fact that the worse your advantage (to the point of being a disadvantage) the better "bold strategy" is for your game plan?
Dave

Dave,

You are correct in concluding that the "Bold Play" strategy is predicated on facing a negative expectation event. To counter the negative impact of expectation, bold play deliberately increases the standard deviation of final outcomes so that, with equivalent starting bankrolls, some bettors will become profitable and others will go bankrupt more quickly! Note that most handicappers overestimate how quickly expectation will dominate volatility.

John

[jfdinneen]

Originally Posted by TrifectaMike

Let me make it clear. I am not an advocate for playing negative expectation games. I am merely presenting information.
With that said, the critical item is (d). Although playing a negative expectation game at a level stake (without strategy) will lead to ruin, the number of events (races) can be approximated with an associated probability of 1 for ruin.
However, the more aggressive Bold Play which increases the the probability of winning under volatility events (races played) are greatly reduced to maximize the winning probability.
To put it more simply. The expected loss of Bold Play is much less for each session and that without Bold Play (Strategy) the expected value is about 0. However, with repeated use of Bold Play one will lose much, much more by the time the player without strategy loses their bankroll.
Determining (d) is not a simple matter. It is not obvious to me. Increasing the number of events reduces the probability of winning the session, but decreases losses. Decreasing the number of events increases the probability of winning the session, but exposes one to greater losses.
Mike

Mike,

Apologies, if you believed that I was suggesting you recommended playing negative expectation games.
As you know, "Bold Play" falls under the heading "How to Gamble If You Must?" when faced with negative expectation games. What I was trying to get across (unsuccessfully - my fault) was that if the majority of handicappers do not have an exploitable edge over the "Crowd" (Public Odds) then is it wise to continue advising them to use edge-based staking (unless someone repeals the laws of mathematics, they will go bankrupt).

John

[TrifectaMike]

Originally Posted by Dave Schwartz

The strength of my own handicapping is that it picks great contenders. Not good, but great. So great, in fact, that I can wager 3-4 of them per race flat and come within 4.5% of even. When all is said and done, one finds that profit begins with the 3.5th public choice.

Of course, that is all hindsight - as one cannot truly discern which horses will be what odds with any degree of certainty before the race is run.

So, your question is

Is it possible to develop a predictor for the 3.5th public choice plus some acceptable error?

Does an adequate predictor exist? Maybe or maybe not. I actually don't know. It is an interesting exercise. However, I can suggest a starting point for you.

The odds distribution follows a power law.

Mike

P.S. Dave, you are an interesting fellow. You're always thinking outside the box. Always exploring, thinking, adapting concepts, and quite often right.

[ReplayRandall]

I'm BUMPING this nearly 6 yr. old thread because it truly is worth a read, from start to finish.....Trust me on this one, you'll find some real nuggets here.

[overthehill]

yes this is a fascinating discussion. trying to predict final odds based on information that you have say with a the toteboard reading 1 minute to post time.

That could be the subject of a dissertation. In general, i would say that final odds of horse over 10-1 are pretty impossible to predict. as a general rule . if a horse is an obvious favorite the odds are going to be from 4/5-6/5 and not usually bettable. in races where there is an obvious odds on favorite everyone looks like an overlay. so the key question becomes how likely is this odds on favorite to win. and how easy is that to evaluate. in general I believe that if you arent keyed on a horse that isnt among the first two morning line favorites and isnt the actual favorite, it is likely that the odds you see are goiing to be about what you get or better.

[lansdale]

This thread is a discussion on the merits of whether 'bold play' can be a profitable strategy for the 99+% of horseplayers who are effectively playing a negative expectation game. For those who have followed his posts, this seems to be the strategy suggested by Delta Lover, who, like jd, doesn't believe it's possible to achieve a long-term edge over the betting public. jd is a very smart man, but I think his conjecture was effectively refuted here by Overlay, TM, and sjk.

[Magister Ludi]

Originally Posted by overthehill

yes this is a fascinating discussion. trying to predict final odds based on information that you have say with a the toteboard reading 1 minute to post time.

That could be the subject of a dissertation. In general, i would say that final odds of horse over 10-1 are pretty impossible to predict. as a general rule . if a horse is an obvious favorite the odds are going to be from 4/5-6/5 and not usually bettable. in races where there is an obvious odds on favorite everyone looks like an overlay. so the key question becomes how likely is this odds on favorite to win. and how easy is that to evaluate. in general I believe that if you arent keyed on a horse that isnt among the first two morning line favorites and isnt the actual favorite, it is likely that the odds you see are goiing to be about what you get or better.

There is a simple, yet accurate method of predicting final pari-mutuel odds. My final tote model uses that method plus two other submodels. Average error <1%. No further information forthcoming.

[steveb]

Originally Posted by Magister Ludi

There is a simple, yet accurate method of predicting final pari-mutuel odds. My final tote model uses that method plus two other submodels. Average error <1%. No further information forthcoming.

it always intrigues me when this happens.
may i ask what was the point of your posting anything then?

[thaskalos]

Originally Posted by Magister Ludi

There is a simple, yet accurate method of predicting final pari-mutuel odds. My final tote model uses that method plus two other submodels. Average error <1%. No further information forthcoming.

Please don't think that your display of secrecy isn't appreciated.

[Hoofless_Wonder]

Originally Posted by ReplayRandall

I'm BUMPING this nearly 6 yr. old thread because it truly is worth a read, from start to finish.....Trust me on this one, you'll find some real nuggets here.

I read this entire thread, and came away with these nuggets:

1. If your handicapping stinks, an advanced betting method ain't gonna help....much.
2. Contrary to popular opinion (on this forum), some edges in horse racing do NOT disappear over time.
3. Advanced betting strategies (i.e., Kelly or Bold Play) are nice in theory, and perhaps have their merits, but are not easily adopted by the average player without the computer assisted tools.
4. Almost six years have gone by, and I imagine The Detective still has a headache.
5. I'll agree that predicting final odds is not that difficult, though within 1% is pretty extreme.
6. I'll state that predicting final odds without tote input is very, very difficult.
7. I won't be expecting to find a good book to help with my handicapping (unless Thask comes through) or a good book to help with final odds prediction any time soon.
8. While I may be giving away a couple of percentage points of ROI without using an advanced betting method, until my handicapping improves I've got bigger fish to fry.

BTW - RR I do appreciate the thread bump. Not totally applicable to me, but interesting.

[lansdale]

Originally Posted by Hoofless_Wonder

I read this entire thread, and came away with these nuggets:

1. If your handicapping stinks, an advanced betting method ain't gonna help....much.
2. Contrary to popular opinion (on this forum), some edges in horse racing do NOT disappear over time.
3. Advanced betting strategies (i.e., Kelly or Bold Play) are nice in theory, and perhaps have their merits, but are not easily adopted by the average player without the computer assisted tools.
4. Almost six years have gone by, and I imagine The Detective still has a headache.
5. I'll agree that predicting final odds is not that difficult, though within 1% is pretty extreme.
6. I'll state that predicting final odds without tote input is very, very difficult.
7. I won't be expecting to find a good book to help with my handicapping (unless Thask comes through) or a good book to help with final odds prediction any time soon.
8. While I may be giving away a couple of percentage points of ROI without using an advanced betting method, until my handicapping improves I've got bigger fish to fry.

BTW - RR I do appreciate the thread bump. Not totally applicable to me, but interesting.

Very understandable response to this thread. Just wanted to add, jd was posing Kelly as the opposite of Bold Play here, not something similar. Kelly is for those playing with a known edge-per-race, Bold Play is for those playing a negative expectation game with specific limits. As you say, very few can make use of Kelly (but it does work), and Bold Play doesn't seem to apply to the conditions of horseracing.

[erikeepper]

Originally Posted by lansdale

Bold Play doesn't seem to apply to the conditions of horseracing.

TrifectaMike: "I personally use a form of Bold play"

Well, form of Bold play....

[lansdale]

Originally Posted by erikeepper

TrifectaMike: "I personally use a form of Bold play"

Well, form of Bold play....

The manner in which TM may be using BP has nothing to do with what is being described by jd. As TM made clear, he is playing with a sizeable edge - when last mentioned, somewhere in the mid-20% range. If he's using BP, you can be sure that it's only to enhance an already existent advantage. jd is talking about the average losing (net negative) horseplayer doing this. Difficult to distinguish this in practice from ordinary gambling.

[classhandicapper]

Originally Posted by therussmeister

That's why I consider handicapping writers to be my mortal enemy.

There is another way to look at it.

If all the books are saying the same things and more players are switching their thinking because of it, they might be creating value somewhere else.

[erikeepper]

I was just answering to your post where you said:
"Bold Play is for those playing a negative expectation game with specific limits..." and
"...Bold Play doesn't seem to apply to the conditions of horseracing."

I was just pointing out that you can use a form of bold play even if you have an edge and play horses. Of course your need the expertise of TM to do it and that is another story