Read on Twitter
1. Quick THREAD on using the Kelly fraction to optimally size positions in your portfolio, just like Ed Thorpe did in the 1970's (1/N):
Don't do it, you colossal idiot. It's not the 1970's, and you're not Ed Thorpe. (N/N)
Since this joke got popular, the actual explanation: The style of trades Thorp specialized in, convert arbs, have an ex-ante quantifiable E[R] that is reasonably accurate, which means your Kelly estimates are ~ right. Other asset classes and strategies do not have this property.
Since Kelly depends linearly on E[R], all you are doing is transforming your highly uncertain estimate of E[R] into a highly uncertain estimate of w *that you now think is optimal*; but for realistic error bars on E[R], your CI for w almost surely includes all of [-100%,100%]!
The only case where this is not true, and your calculated Kelly bet size is actually useful in trading, is if you can routinely generate trades with an information ratio >1. On the other hand, if that's the case you're shopping for your next Basquiat, not reading my Twitter feed.
