Kelly System for position sizing

Yes, I think you're both right. I'm going to enlist the help on on of my Uni friends - he's a Maths Phd on symmetry and group theory but he's also very much into maths theories on gambling. I have a feeling he'd be interested in this subject and, therefore, think he's my best bet to try and push forward on this. We'll see...

Good idea to enlist the help of some brainy bods... but a pure math bod?

The thing that makes this problem interesting / difficult is that you can change the rules. Doesn't sit so well with group theory IIRC.
 
A PHD student, never mind graduate is over qualified for this. Quite sure you're all over thinking it.

What are you actually trying to figure out?
 
I was reading about Kelly earlier and need to understand something. If you use the Kelly formula does the Kelly% indicate how much of your capital you should have out on trades, or how much capital to put on just one single trade?
 
Neither!

The Kelly formula gives a % of the MAXIMUM you should put on a single trade. The whole point is that it's the mathematical formula and proven that if you risk more...(with my lack of eloquence I can't find the right words), but if you risk more it's silly, and don't do it :D.
 
No, it's the *optimum*.

That said one of the assumptions of the Kelly criteria is that you're playing one game at a time; it doesn't cover a situation where you have several trades on the go.
 
No, it's the *optimum*.

That said one of the assumptions of the Kelly criteria is that you're playing one game at a time; it doesn't cover a situation where you have several trades on the go.

Its the optimum to maximise your expected long-term growth rate - which might not be most appropriate for all trading styles or traders - Kelly (or half kelly, which is used in practice) isn't the right tool for minimizing drawdowns, for example.
 
No, it's the *optimum*.

That said one of the assumptions of the Kelly criteria is that you're playing one game at a time; it doesn't cover a situation where you have several trades on the go.

So what would you use for optimum growth when having bets/trades simulatenously?
 
You would have to use some common sense; one approach would be to monitor all positions and if your overall loss approaches whatever your loss limit is then cut the loser. I'd probably have tighter stops on each individual position proportional to the expected stop / square root of all trades (assuming no correlation between trades).

Assuming you were still going with the Kelly paradigm of course :)
 
i want to ask something about kelly formula.how can i use kelly formula in order to find leverage??i know kelly formula it finds optimum position size.but what about leverage??
if ı have a system which has positive expectancy is it good to use kelly formula with any leverage??i mean does it matter if ı use 1:1 leverage or 1:10 or 1:50 with my kelly formula??

for example my kelly formula equals to 0,30 and my initial capital 10000
so i will invest 3000.but which leverage will i use?
1:1 1:10 1:50 which one??
how can i determine my leverage??
 
i want to ask something about kelly formula.how can i use kelly formula in order to find leverage??i know kelly formula it finds optimum position size.but what about leverage??

Read through this entire thread. There is a fair amount of good explanation of what the Kelly Criterion is and how to put it to practical use.

Once you have absorbed the knowledge here you will be in a much better position to ask your question or, more likely, you will answer it for yourself.
 
Kelly sizing will only work of you keep running and studying all the statistics on each trade you place.
It's math not magic !!! you'd probably be better off comparing Kelly's ratios against some other quantitative sizing tools
 
Kelly sizing will only work of you keep running and studying all the statistics on each trade you place.
It's math not magic !!! you'd probably be better off comparing Kelly's ratios against some other quantitative sizing tools

Who says it will work if you keep adjusting the stats on each trade you place? I don't think that's what Kelly said.

For the Kelly criterion, you need the probability of a win, and the size of a win relative to a loss to be known. None of these is known for sure on every trade.

You also need for those probabilities to not be affected by the size of your bet. Do you trade the same with a 2% trade as with a 30% trade?

Kelly Criterion also doesn't care about the path taken by your account. But in trading it does matter if you go into a serious drawdown so that you can no longer take the trade. It may also assume independence of results.


If you want to apply Kelly criterion in the hope that it is nearly optimal, then at least be aware that the maths you're relying on doesn't apply to what you're doing.
 
Who says it will work if you keep adjusting the stats on each trade you place? I don't think that's what Kelly said.

For the Kelly criterion, you need the probability of a win, and the size of a win relative to a loss to be known. None of these is known for sure on every trade.

You also need for those probabilities to not be affected by the size of your bet. Do you trade the same with a 2% trade as with a 30% trade?

Kelly Criterion also doesn't care about the path taken by your account. But in trading it does matter if you go into a serious drawdown so that you can no longer take the trade. It may also assume independence of results.


If you want to apply Kelly criterion in the hope that it is nearly optimal, then at least be aware that the maths you're relying on doesn't apply to what you're doing.


I have been using these techniques for quite a while and I can tell you that the sizing of each trade I take is given by an average of 3-4 different models that I use to calibrate the appropriate amount to trade.
Of course, it's not an optimal tool.....there is no certainty in trading and quantitative models can grasp the 25%-30% of the overall movement.
When i said that the Kelly criterion will work if you study the statistics is because trough that (in combination with other calculations) you can get an idea of the probability to place a successful transaction but you will never be 100% sure.
Relying on 1 model is really naive and just an unexperienced trader would do that.
Simply,I did not want to write a long post that's why i limited my answer to a few lines
 
Oiltrader, I don't judge what size you should trade. I simply want to make the point that if you're going to use mathematics, you should make sure the result your using applies to what you're using it for. And if it doesn't, be cautious. Otherwise you're giving yourself a false sense of security in bad maths.
 
Oiltrader, I don't judge what size you should trade. I simply want to make the point that if you're going to use mathematics, you should make sure the result your using applies to what you're using it for. And if it doesn't, be cautious. Otherwise you're giving yourself a false sense of security in bad maths.

If that is the intended meaning of your initial post I TOTALLY agree with you. Thanks for the comments though !!
 
for example my kelly formula equals to 0,30 and my initial capital 10000
so i will invest 3000.but which leverage will i use?
1:1 1:10 1:50 which one??
how can i determine my leverage??

3000 is the amount you risk, not invest. This is how much Kelly is telling you to risk on the next trade. The amount you invest depends on price of asset.

If you are talking about forex, you should look at your trading system and see what is the stop loss for the next trade per contract. If it is X for example, 3000/X is the number of contracts you can buy or sell.

The leverage is built in the value of X. If it is $10./pip you use 1:100, $1/pip use 1:10, and so on.

This is a very good paper on Kelly

http://www.priceactionlab.com/Literature/Kelly.pdf

This is an article by the same author on leverage:

http://www.priceactionlab.com/Literature/LeverageMarginEquity.pdf

Read the papers carefully. Also, he has a blog with some very interesting postings. Here is one about starting capital

http://www.priceactionlab.com/Blog/...t-capitalization-is-primary-cause-of-failure/

He is a good one to follow. Very down to earth and practical. Not specific to forex but good insight in general. Do not get the books unless you are interested in automated pattern discovery. Most material in the books other than that is covered in the free papers.
 
Top