Broker: Interactive Brokers, MB Trading, Tradestation for data
Trading: FX, Equities
Posts: 39 since Oct 2012
Thanks: 44 given,
41
received
It's great to hear from experienced professionals, especially when they are practical and modest like Ernie.
Thank you Mike once more for giving people from all around the world access to serious education forums.
I think Ernie did a good job for his choice of topic, the diversity of the audience and the content he had to cover. I loved his answers, didn't agree with parts but didn't want to interrupt. He has his critics but I think it takes a lot of generosity to share as openly as he does. Would be great to have him back again, maybe to talk about backtesting.
The following user says Thank You to artemiso for this post:
I enjoyed the webinar, and I look forward to watching it again once you post the recording. While I would not presume to be as advanced as Mr. Chan in my own understanding an application of Kelly, I do wonder why he would advise a retail trader use 1/2 Kelly. If you have an edge, a known, quantified edge, and not a " just pulled this number out of my you know where" edge, then I see no reason not to use Kelly's original formula.
The following user says Thank You to cygnetnoir for this post:
Broker: Interactive Brokers, MB Trading, Tradestation for data
Trading: FX, Equities
Posts: 39 since Oct 2012
Thanks: 44 given,
41
received
I'd assume the main reason for suggesting half-kelly is because the average person would prefer sub-optimal returns with a benefit of avoiding ulcer from the stress of a lot of leverage!
The average person speaks of having an edge, but he truly has no idea whether or not he has one; and if he has no idea what, if any, his edge is, then I agree, he should probably not be betting at all, but if he must bet, by all means he should bet small.
But if you have an edge - a quantified, calculable edge - based on you yourself in the pilot seat, then why not press your edge for all it is worth?
Broker: Interactive Brokers, MB Trading, Tradestation for data
Trading: FX, Equities
Posts: 39 since Oct 2012
Thanks: 44 given,
41
received
And because the trader with a proven edge can still get ulcers.
Also, as I understand, kelly shows you how to maximise returns, but it also increases risk.
It's a matter of risk appetite it seems to me, you can try to make a lot of money with the risk of larger drawdowns or even ruin, or you can adjust for lower profits and expect to sleep easier.
My small experience with trading definitely puts me in the second category!
I get what you are saying, but I still do not see that halving the bet size should be necessary. I missed the first part of Chan's webinar, and so maybe he is using a derivation or variation of Kelly. As I understand Kelly, the formula is asKelly described it: Max Gain = edge/odds, or (P*W-L)/P, where P = expected payoff, W = probability of winning your bet and L = probability of losing your bet. These variables, while certainly subject to a standard deviation that likely does not follow a normal distribution, nonetheless are not (or at least should not) be randomly selected.
Again, my thoughts are predicated upon having a fairly certain and accurately calculated edge, which, by definition, requires an accurately calculated expected payoff. If you have these two numbers, and they are based upon actual trading from the cockpit and not some "pie in the sky" assumptions, then you should not need so large a "margin or error" safety net as a halving of bet size, which will have a much more deleterious efect on growth potential than merely halving it.
Kelly's formula actually describes a method of bankroll management that allows you to maximize capital growth while at the same time minimizing the risk of ruin. Of course, if you start with a small stake, then Kelly may not prevent your capital from experiencing a drawdown sufficient to cause you no longer to meet minimum margin requirements to trade, effectively taking you out of the game, which is, of course, a form of ruin, even if you are not, technically, "broke."
