Kelly Criterion

[Kafkaesque]

My question is for betting 20 selections with a bankroll of 1000 E. The selections are at the same time.

If it's betting, my first piece of advice would be to join a forum for betting You're likely to get better and more varied responses. Not that you shouldn't stick around here Second piece of advice. Unless you're in this spot because it's all very short prices, you really shouldn't be going over 100% with 20 selections. You're either wrong about your calculation of chances for your selections, or you're a God among men with picking selections If your "probability of success" is derived from your opinion, take a step back and consider whether you're overestimating your edge over the price. If it's derived from past results/bets, look into p-value to see how many bets it actually takes before results can be considered statistically relevant and sustainable. It's a lot more than you'd expect!

Just my two cents, but as it were, if we assume, your calculations are correct, you've got a few options:

1) Fractional Kelly. Yep Nero, I came along and said it Simply adjust each bet down by a certain percentage to coincide with 100% total.

2) Choose the selections with the highest recommended percentage, and leave the rest out. These should be the ones with the best combination of high edge and low volatility.

3) Choose the selections with the shortest prices, and leave the rest out. Lowest volatility and thus chance of building your bankroll.

Now, here's where I do agree with Nero. I'm not a fan of fractional Kelly. I would suggest option 2, closely followed by option 3. I only mention fractional Kelly, because using full Kelly with option 2 absolutely demands that you know what you're doing. If you're in any doubt, as to whether your basis of calculations is rock solid, then I'd go with option 1.

[Kafkaesque]

I didn't suggest any unit based system. I'm advocating use of full kelly! Using a notional bank. The key is not to change the bank, the bet sizes obviously would change massively because it is full kelly. This is the ideal way of dealing with high repetition of trials, multiple markets and varying levels of liquidity as we are in betting markets. 10 years plus using this myself tells me it is far more optimal than level / unit stakes.. but the implementation for trading isn't obvious.

You're advocating using full Kelly to calculate the stake/units staked within a unit staking system. Not actually using full Kelly. Because, yeah changing the bankroll is absolutely key. The point of Kelly is to grow the bank exponentially, which can only be done by adjusting the bank.

I do agree that the implementation isn't obvious for trading. Said as much in the difference between trading and gambling thread.

[Kafkaesque]

Before coming off as too much of a c***, I should qualify the above by noting that I'm not necessarily saying, you're doing anything wrong with your approach to BR management and staking. I am however adament that your terminology and use is wrong, which by extention made it bad advice to the question at hand. 10 years of doing something doesn't - automatically - make you good at it. I've been making a living for the best part of two decades on a, at various times, combination of poker, betting, day trading, and sports trading. And I'm still very much learning. Every day! More than anything, what I've learned over all those years is to always keep an open mind. Because I've made plenty - plenty! - of mistakes and have held mistaken believes on theories and approaches for any given time before seeing the light. I do hear and acknowledge your approach to BR. I happen to disagree with it as completely ideal, but it's absolutely fine and solid. In the end, all I'm saying is, that I see room for improvement, given that you seem - to me at least - not to have fully come to terms with the workings of Kelly. You're welcome to stick to your steadfast approach of having a full grasp of it, in which case I can only close with the same end that one of your first response to me had. Good luck.

[Derek27]

Hello Everyone,

I would like to use Kelly Criterion but I do not manage to use it at more bets at the same time.
For example: I would like to bet at 20 matches what are playing at the same time.
How do you calculate your stake for more matches with Kelly Criterion?

Thanks

Wikipedia explains how to use KC for multiple bets (by that I don't mean accumulators), but it's quite complex !

https://en.wikipedia.org/wiki/Kelly_criterion

[Kafkaesque]

Wikipedia explains how to use KC for multiple bets (by that I don't mean accumulators), but it's quite complex !

https://en.wikipedia.org/wiki/Kelly_criterion

What you're talking about is multiple bets within the same market, though. Or???

Different from this spot where it's multiple bets in independent markets.

FWIW, a more explanatory approach, even if still sligthly complex for the problem, if it helps anyone:

https://www.pinnacle.com/en/betting-art ... CB3KNYN9CJ

[Derek27]

Wikipedia explains how to use KC for multiple bets (by that I don't mean accumulators), but it's quite complex !

https://en.wikipedia.org/wiki/Kelly_criterion

What you're talking about is multiple bets within the same market, though. Or???

Different from this spot where it's multiple bets in independent markets.

FWIW, a more explanatory approach, even if still sligthly complex for the problem, if it helps anyone:

https://www.pinnacle.com/en/betting-art ... CB3KNYN9CJ

Sorry, I misread this thread. I'll read the Pinnacle article you posted when I've got time.

[Derek27]

Thanks Kafkaesque, that is an interesting article.

[ShaunWhite]

I just stake what feels right for the market, depending on my confidence in it.
I'm not sure a formula would give me a better staking plan.

[Derek27]

The Kelly criterium is mathematically accurate if you know for certain the true probability of a bet winning, but nobody can have real confidence in their estimate, especially for horse racing, so it's best used as a guide, taking a cautious view.