This video is stupid. Chances are none of these people have ever heard of the Kelly criterion, much less know how to apply it. Obviously his $10-for-$10 bet is a nonstarter, except for degenerate gamblers. His $12-for-$10 bet requires that I have a minimum of $10/(.50-.5/1.2) = $120 of disposable income to make risking $10 worthwhile. His $20-for-$10 bet requires a minimum of $10/(.50-.5/2) = $40 of disposable income to make risking $10 worthwhile. Given that the average American is innumerate, no surprise that random-interview guy couldn't find any takers. Next.
Not that I am a believer in Kelly being the most optimal bet sizing approach but I was pretty spot on with my statement that 30-40 dollars in the bank/pocket would result in a highly leveraged bet when talking about losing $10 with probability 50%. So, yes, the expected payoff would have to be above a certain threshold to make this bet make sense even with repeated tosses. And I agree the experiment how it was conducted was stupid though the idea and peoples' reaction was interesting to watch. And I do not think you would garner a different response in Japan, Germany, or the UK.
I'd like to see the response in NYC. "let me see the 10 bucks", then take it and run while you yell "sucker".
Kelly does that implicitly. The problem with Kelly is that the public formulae were designed for casino games, not for trading. Trading is not a casino game: there are no fixed probabilities and no fixed payouts. So the problem is more complicated but it can be expressed explicitly and even 'solved' to a large extent, but I don't expect to ever see a trading-worthy Kelly formula on the internet. Too much casino baggage and misapplication thereof dragging it down. Calling it something different (e.g., geometric mean maximization) doesn't seem to have advanced the science much.
Kut2k2, what is the kelly fraction for this bet? The 12 for 10 one. I calculate it at 50%, which doesn't sound right.