Re:In a town of gamblers, Mattermark is counting cards
by mynameisme on Thursday August 22, 2013 @02:51PM
(#6259192)
Say there is a game at a casino, that has a 1/100 chance of winning, and if you win, you get a payout of 100.01(original bet). You have $100. How would you play without it being extremely dangerous?
Re:In a town of gamblers, Mattermark is counting cards
by btilly on Thursday August 22, 2013 @03:53PM
(#6259615)
By the Kelly criterion, you should bet edge/odds of your bankroll. Your edge here is 0.01 * 0.0001 = 0.000001. Your odds are 0.01. Therefore you should bet 0.0001 of your net worth. So if you had a million dollars, that $100 bet would be a sensible investment. If you've only got, say, $100k then you should only bet for entertainment value.
The investment rule that underlies Kelly is to maximize the log of your net worth. Why? Bets multiply your net worth by a random factor, take logs and you're adding a random factor. In the long run a sum of independent random factors tends to converge to the expected value, so maximizing the expected log of your net worth will result in a strategy that with 100% odds will, in the long run, beat any other.
Re:In a town of gamblers, Mattermark is counting cards
by mynameisme on Thursday August 22, 2013 @04:31PM
(#6259867)
Sorry, I was being rhetorical, I guess it wasn't obvious. My point to the parent was that no need to bet your entire life savings on one company, you can split it among hundreds and take advantage of your edge.
Speaking of, just because I may know a good investment, doesn't mean I have the resources to actually capitalize on it. I may know a likely way to make 10 Million/year, but if the buy in is 5 million, I'd be out of luck. Might as well sell that information to someone with 10 Million and make a cool 50k/100k for myself in the process.
During the goldrush, selling pickaxes and pans to miners was often a better way to get rich than to actually take the risk of mining yourself. You can see a modern day parallel in the custom bitcoin rigs. The companies selling them could just fire the machines up themselves. In the long run however, its better to sell the machine that could make 10k/yr for a guaranteed 2k in your pocket.
A bird in hand is better than two in the bush.