The problem isn't the estimate of the expected payoff, it's the assumption of normalcy. Financial return series tend to have far heavier tails than the normal distribution implies. While "half" isn't a hard and fast rule, you should treat Kelly as a ceiling on your sizing, not the target.
Unless of course you have a good estimate of the shape of the distribution of your strategy, and expect it to not change in the future....
The following user says Thank You to imPairsonator for this post:
This is correct. The common Kelly derivation maximizes log growth but says nothing of the drawdowns that you will experience. And if you in fact do assume the moments are constant in time and satisfy a normal distribution and simulate using Kelly leverage on every trade, the drawdowns will clearly not stand the test of real-life constraints although it keeps to the promise of keeping your ROI above -1. I want to reiterate that most people complain about the normal distribution but this is a minor issue.
As promised, I've given away my copy to someone else as I don't think it's in the spirit of this giveaway that I keep the book. I found a college sophomore at one of the Ivies who wants to work at the prop desk of an investment bank, but the odds are stacked against her because she has a speech disability.
The following user says Thank You to artemiso for this post:
I realize that I am answering an old post but I think I'd try to contribute, if not for the OP's benefit, then for the other members reading this "sticky" thread.
First of, for the record, the half-Kelly bet is not equal half the size of the full Kelly bet.
That aside, the main reason to bet half-Kelly is that there is a big difference between underbetting the full Kelly and overbetting it. While EV>0 (expectancy is positive), the former ensures that the Expected Growth (EG) is positive while the latter ensures that the EG is negative! This mathematical fact is counterintuitive for the Kelly novices because the common sense tells us that slightly overbetting full Kelly should be a bit less profitable than betting full Kelly. In fact, it's a money loosing proposition!
So, once you plug in the fact that one can only estimate his trading edge, it is obvious that even for a risk neutral trader it is better to err on a safe side.
The following user says Thank You to igoeesd for this post:
You may be aware that @Big Mike is away, dealing with medical issues. As of now, we have no ETA for his return, and he is the only one who can address the issues with the older webinars.
I know that converting the older webinars is an issue that members sometimes ask about, and I wish I didn't have to keep making the same reply, but I do.
When Mike is back, he will have the webinar conversion issue as a known problem to be addressed (along with others). But as of now, it cannot be.
Sorry that this is the answer, but it's what we have right now. I hope to have a different one, but I don't know when it will be.
Bob.
When one door closes, another opens.
-- Cervantes, Don Quixote
The following user says Thank You to bobwest for this post:
