Kelly Question

If you are only going to stake 1 point maximum doesn't that defeat the whole point of Kelly?

That's what I thought but the 5% edge is over the market as a whole so if the first game told me to stake 52% of my bank I would be in trouble...

Would staking 52% of 1 point for that game etc etc not work the same?

http://www.seriousbet.co.uk/calculators/kelly_staking

I suppose you coul put it in that calculator and it would tell you the % and you would apply the % of 1 point or the % X3 or r something.

Sounds like a lot of work to price up and calculate 100 odd games a weekend :)

You could always start with a smaller 'bank' than your actual available funds. If you genuinely have a 5% edge you should build it up in no time.

Thanks both, that's what I am thinking Knight Rider... I have a computer program to work out the dds and kelly stakes but could do with something to place the actual bets tbh

Good luck. Personally have found Kelly staking to be very dangerous. End up having a couple of massive bets each weekend when your price is way out of line with the market, and your success hinges on those results.

Have you trialled other staking plans to see if they return better than 5% edge? My models do better on level stakes than Kelly (though somewhere in between is optimum).

Thanks. Yes I agree it can be very dangerous particularly when using your whole bank to kelly stake your bets. This is why I am thinking of kelly staking within point limits... thus maximising the value in the odds but limiting the risk as it's spread over 100 bets. I'll let you know how it goes!

Knight Rider 30 Apr 00:22
Good luck. Personally have found Kelly staking to be very dangerous. End up having a couple of massive bets each weekend when your price is way out of line with the market, and your success hinges on those results.

Have you trialled other staking plans to see if they return better than 5% edge? My models do better on level stakes than Kelly (though somewhere in between is optimum).

To me the most likely conclusion to draw from the above is that you do have an edge, but you think it is larger than it actually is.

I don't think level stakes can beat Kelly (or at least a system derived from Kelly) if you can accurately assess your edge.

How about spliting your bets into three categories say odds on, 2 to 5, greater than 5 (or any split that seems sensible to you) and use kelly on the average odds for each category.

Think it all depends on what odds range your looking at / strike rate and then what you feel comfortble with. If I am feeling nervous about the size of a bet then its sign I am betting to much

As has been stated 5% is massive so I wouldnt get bogged down in staking. just take it nice and easy and tweek it around when you get better feel for it.

suppose if Kelly is telling you to bet 52% of you bank they must be skinny odds.

If the edge is over the 120 almost simulatanous markets can you not do Kelly/120, probably half Kelly /120 would be safer to start

The Investor 30 Apr 01:02

To me the most likely conclusion to draw from the above is that you do have an edge, but you think it is larger than it actually is.

I don't think level stakes can beat Kelly (or at least a system derived from Kelly) if you can accurately assess your edge.

----------------------------------------

It's not so much I THINK it is larger, just that my models produce prices which I know aren't 100% accurate and the 'true' price will normally lie somewhere between my price and the market price.

Obviously from results I can see on average what sort of 'margin for error' my prices have, but this doesn't necessarily translate as a consistent difference on each bet. It may well be that most of my prices are spot on, and the occasional one is miles off (to use football as an example, maybe one team is decimated by injuries). This is where Kelly is a killer.

Been doing this for years and tbh I'm still not totally comfortable with staking. It's a work in progress and will probably remain so for years to come :-)

Start with a bank that is significantly lower than the total betting funds available to you, then use Kelly (or a watered down version if you want to be more careful) and if you have an edge you will make money.

Any concurrent matches you need to combine to work out correct Kelly stakes. For example if there were only 2 concurrent matches instead of 2 events both with 3 outcomes, you have 1 event with 9 outcomes where the probability of each is the product of the 2 individual probabilities ...

Obviously this gets quite involved when there are many concurrent matches, but you can split them into groups and assign banks for each group. 10 matches is perfectly achievable to calculate Kelly stakes.