OK forget "Kelly". Too much unnecessary controversy tied to that name. Focus on "geometric mean maximization". For trading, not for gambling. GMM is undeniably based on the trade returns.
Would You Take This Bet?
- Thread starter rmorse
- Start date
lol, call it whatever you like, but your bet sizing formula you applied earlier certainly does not take into account price volatility and hence dynamics in expected return. It just happens to be called Kelly Criterion even if you do not like it.
Why would the bet sizing formula I applied to a gambling situation have anything to do with price volatility? I made a thread called "Kelly for traders". Read the first post so you can see what I'm talking about.
I can answer that:
Because
a) you said your formula is essentially the Kelly Criterion and recommended using it.
b) Then you word for word asked me "So how do you size your bets? Or trades, as the case may be."
c) To which I replied that I take into account price volatility and/or implied/expected volatility.
d) To which you replied that Kelly already accounts for that
e) To which I replied that it is factually incorrect
f) After which you said "OK forget "Kelly""
See the relationship now? ;-)
Kelly is not a bad approach to sizing fixed bets and not even a bad approach to size positions in financial trading, but it is certainly not the most optimal approach, something I exactly said earlier.
Because
a) you said your formula is essentially the Kelly Criterion and recommended using it.
b) Then you word for word asked me "So how do you size your bets? Or trades, as the case may be."
c) To which I replied that I take into account price volatility and/or implied/expected volatility.
d) To which you replied that Kelly already accounts for that
e) To which I replied that it is factually incorrect
f) After which you said "OK forget "Kelly""
See the relationship now? ;-)
Kelly is not a bad approach to sizing fixed bets and not even a bad approach to size positions in financial trading, but it is certainly not the most optimal approach, something I exactly said earlier.
You didn't say (c). You said "...by certainly taking into account implied/expected price/return volatility..."
Certainly Kelly has nothing to do with price volatility but it absolutely has to do with return volatility when the equation to be solved is
0 = sum[ Ri/(1+k*Ri) ]_i=1toN
The Kelly fraction is directly dependent on the trade return values, which includes their volatilities implicitly. You can nitpick this issue to death but I'm done here. You've said nothing to reveal a better optimization method than Kelly/GMM.
you misunderstood what I meant with "price/return volatility". I of course referred to the price volatility AND volatility of log price changes (aka price return volatility). It is not reflected in Kelly and that was my answer to your question.
By the way you are factually still wrong, Kelly does not even take return volatility into account. It takes into account the probability of winning, and your win/loss ratio. But that itself reveals nothing about return volatility because win/loss ratio is simply the avg gain divided by avg loss which says nothing about the distribution of wins and losses. Maybe you wanna refresh your assumed knowledge because what you were saying is not correct.
At least you could have the courage to admit you were wrong about return volatility in the context of the Kelly Criterion. It takes a simple "my wrong",..., and everyone moves on. Its really not that hard.
By the way you are factually still wrong, Kelly does not even take return volatility into account. It takes into account the probability of winning, and your win/loss ratio. But that itself reveals nothing about return volatility because win/loss ratio is simply the avg gain divided by avg loss which says nothing about the distribution of wins and losses. Maybe you wanna refresh your assumed knowledge because what you were saying is not correct.
At least you could have the courage to admit you were wrong about return volatility in the context of the Kelly Criterion. It takes a simple "my wrong",..., and everyone moves on. Its really not that hard.
Last edited:
If you haven't figured out by now that the Kelly formula for betting and the Kelly formula for trading are very different, then there is little hope for you. I will give you a few hints, and leave it to you do the work. Or not.
The best Kelly formula for trading (which is proprietary) has nothing to do with any of the following : winrate, average win, average loss, drawdown, standard deviation.
What is clear most of all is that you are nothing but a top-posting gotcha troll. You go on my ignore list, right along with asiaprop.
because you changed from Kelly Criterion to "geometric mean maximization" and now to some magical "proprietary formula" after you were proven multiple times wrong? I have been long enough in this business to notice a red flag and you have posted several in a row. Enough said. For your own sake I recommend you to inform yourself on then technicalities of betting formulas before you make bold (but wrong) claims.