3 ms·
I went down the rabbit hole with the Kelly Criterion recently, last time I thought about venture capitalism. Starting with a scenario of a repeated bet where yo
by tunesmith 7y ago
I went down the rabbit hole with the Kelly Criterion recently, last time I thought about venture capitalism. Starting with a scenario of a repeated bet where you have a 10% chance of a 50:1 upside, Kelly says to bet a percentage of your bankroll each time, while someone of a more conservative bent might want to have a 95% chance of not going below a certain drawdown.
With those stats, you have to bet 28-29 times to have a 95% chance of winning your 50x upside. So depending on your maximum drawdown, you have to make a much smaller bet than what Kelly Criterion counsels.
In fact that seems to be a general thing - Kelly Criterion is generally about maximizing geometric returns, but the volatility will kill you - it takes lot of betting to have high odds of being close to those theoretical returns.
I'm not really experienced enough with the math to find strategies that maximize your expected growth when subjected to a certain drawdown restriction, but that's what people need when they are at lower net worth levels.
- devoply 7y agoSo the Matthew principle strikes again. Rich get richer, poor get poorer.
- disjhshava 7y agoThe math works out a lot better if your society has a strong social safety net, and friendly bankruptcy and limited liability laws.
- tunesmith 7y agoMan, I have to learn more about that at some point. I own a business but it was never designed to be one that took out a business loan. I never even considered it because I just equated losing it with the awfulness of losing a bunch of personal money. Is there a class one can take that teaches people how to work the bankruptcy/llc system the way it's intended to be worked? Meaning, not abusing it, but being "entrepreneurial" in the way that those laws are apparently supposed to incentivize?
- joncrane 7y agoI think it's more of a thing where you talk to a bankruptcy lawyer. I'm sure some are willing to talk for free for 30 minutes or so. Maybe talk to one or two, then draw up a list of questions and pay one for a full consultation.
- tlb 7y agoThe Kelly criterion works for serial bets, but since most startup bets take 10 years to pay off, you can only make a few in series over your lifetime. You want to have many simultaneous investments. You can model this with Black-Scholes, except you don't know any of the parameters. So in practice, just put x% of your available capital into venture bets and put your energy into picking good startups.
- cjlars 7y agoI was typing up the same comment when your hit. Kelly criterion is a great mental framework, but the math obviously doesn't work out for VC placements. The best alternative I've been able to come up with is to brute force / simulate thousands of different portfolio outcomes and optimize from there.