Kelly Criterion and Multiple Place Betting

**Sator** · 02-10-13, 02:29 PM

Welcome

I think that you are looking for that http://www.sbrforum.com/betting-tools/kelly-calculator/ you just need to change to the Exclusive outcomes option.

**brettd** · 02-10-13, 05:07 PM

No it's not that simple. I looked into this at one point, the very fact you can have multiple bets on the same event possibly all win or lose makes this very complicated.

I decided to stick to the win market, as at least that way, I understood the correct actions for optimal expected growth.

**Topo** · 02-17-13, 06:12 PM

I prefer the place market simply because the betting pools are smaller. With a smaller betting pool the market can be assumed to be less efficient, which presents greater opportunities for the bettor. Most of the free information out there is about who can win, not who can place, which is great if you want to put in that extra effort to find those mid/long-shots who can sneak a place.

**Miz** · 02-18-13, 10:49 AM

If you are able to conclude what your payout is with reasonable certainty prior to the race beginning, then some kelly approach might be ok. That is the trickiest part of parimutuel betting... the uncertainty in what the payout will be. It is still profitable, but hard to optimize staking, that's all. good luck, and welcome.

**JoeySoke** · 02-18-13, 02:50 PM

Good info.

**MonkeyF0cker** · 02-19-13, 12:42 AM

It's a combinatorial issue before it's a Kelly issue. You need to calculate the probability of your horse winning or placing along with the odds of all other runners winning or placing. You then need the place pool to derive your odds and expected value for each outcome. Only then can you use Kelly to optimize your EG. However, you'll need to consider your wager size as it may affect the parimutuel payout. There is also a fairly high degree of difficulty in assessing the tote/payout accurately as much of the volume typically comes at post, but there are ways to estimate it. Of course, fixed odds would make this a rather elementary task provided your probabilities are relatively accurate.

**byronbb** · 02-19-13, 02:06 AM

Reminds me of this article about HK computer teams working the races there.

Page Not Found

http://www.wired.com/wired/archive/10.03/betting_pr.html

**Miz** · 02-19-13, 07:51 PM

Thanks for posting that article.

**blackpoolbison** · 05-23-14, 08:49 PM

Sorry to drag up an old thread but does anyone have a solution to this problem? I searched the forum and google and couldnt find an answer anywhere else either.

I'm interested in the World Cup group Stage 'to qualify' market. 4 teams. 2 qualify.

Team/Odds/Chance%
Spain/1.2/82.98755
Netherlands/1.727272/58.65102
Chile/2.05/50.50505050
Australia/13/7.142857

(sorry that doesnt add up to 200% but its a best estimate)

Please give me a formula if possible so I can make myself a spreadsheet for future use!

Ideally I dont want to place bets on the -EV outcomes (spain & especially australia) because it will increase my stake sizes too much but if its a necessary part of the formula...

Thanks for any help!

**mjespoz** · 05-27-14, 05:51 AM

If you're betting into a parimutuel market then I recommend that you read Rufus Isaac's paper "Optimal Horse Race Bets". Kelly doesn't take into consideration the impact of your own betting on final price.
Best of luck. Profitable place betting is even more difficult than win betting.

**blackpoolbison** · 06-02-14, 11:17 PM

Can't speak for the OP, but personally I have no interest in parimutuel betting and it doesn't apply in the case I supplied above.

**lamichaeljames** · 06-03-14, 12:43 PM

**barryt** · 06-09-14, 10:20 AM

I know it's over a year since your post but it was relevant to some thing I was thinking about.
Couple of things.
I'm no mathematician but I do play the horses.
1. Betting 2 horses to place, would not be advisable except under very limited conditions say where there is a huge , but vulnerable favourite who is unlikely to hit the board. Even then which 2 to bet? I think the predictable successful place % would be under compensated by the payoffs. Especially if only 1 places.
Sartin recommended TWO horses to win, which I think is probably a better strategy.

2 DR Z wrote a book on this in the 70's and a later paper on the efficiency of the market, and I believe it pointed out the inefficiency of the place pool. But this was pre simulcasting and 95% of the money was bet at the track and at one minute to post at least50% of the money was in the pool. Nowadays it's certainly less than 50% so very hard to predict final distribution.