3 ms·
Is it unstated when they give the numbers? And the type of expected return they're calculating is neutral.
by Dylan16807 4y ago
Is it unstated when they give the numbers? And the type of expected return they're calculating is neutral.
- s1artibartfast 4y agoHow is it neutral if you offer a coinflip that pays 50% for heads and costs 33% for tails? >To illustrate this concept, you could consider a simple coin-flip game where you make a 50% return on your money if the coin lands on heads, or lose 33.3% of your money if the coin lands on tails. This simple game has a expected long-term return of zero.
- s1artibartfast 4y agoI spent too much time thinking about this today. The slight of hand is that they confuse % change of portfolio or market and the % profit on the bet. The game is positive EV and the only reason that figure 1 doesn't make money is they wager less every time they will win and more every time they will lose. Having a reserve/bankroll to recapitalize is key to winning any +EV game. This is what is demonstrated in their figures 3 and 4. If the bet was truly neutral, +/- 50% each time, there would be no gain. It is obvious if you look at figure 3. Each $100 bet will make $50 or lose $33. There is no strategy to make money on a neutral 50/50 bet, balance all you want. The same is true for a stock with a 50/50 chance of going up or down equal amounts. Otoh, if you knew it was oscillating around a mean, and not doing a random walk, you could always sell high and buy low
- Dylan16807 4y ago> The game is positive EV and the only reason that figure 1 doesn't make money is they wager less every time they will win and more every time they will lose. But that's how investing in a stock works. It's not a weird anti-strategy they came up with. If you put money in and let it ride, it bounces around and you gain nothing. And you gain nothing in part because they do focus on that single asset, with no diversification. If it was +50% and -40% you'd lose money if you tried to do a normal investment, but you'd gain money if you rebalanced properly. So I still think they're discussing some useful properties and scenarios, even if they don't make it clear that there are different types of expected outcome. > Having a reserve/bankroll to recapitalize is key to winning any +EV game. Not really. If it was +10% and -5% you wouldn't need to recapitalize ever. > Otoh, if you knew it was oscillating around a mean, and not doing a random walk, you could always sell high and buy low Notably, the strategy in the post does work for a random walk.
- s1artibartfast 4y ago>Notably, the strategy in the post does work for a random walk. I don't think this is true. Rebalancing doesn't work in the oscillating example provided if you change the rewards from +50%/-33% of bet to +/-50%. I built their model in excel and tested it, and you can too. The Kelly Criterion defines how much of your bankroll to invest in a asset with known upside and downside. For the plus EV game of +50%/-33% you should bet 50% of your bankroll to optimize long term gains. for the neutral EV game of +/-50%, the criterion yields a bet of 0% of bankroll because there is no edge. https://en.wikipedia.org/wiki/Kelly_criterion https://en.wikipedia.org/wiki/Kelly_criterion
- Dylan16807 4y agoDoesn't "random walk" just mean that it's random whether the coin lands heads or tails each day? The contrast you made was random walk as opposed to oscillating around a mean. If you try to pick a value to sell over and buy below, you'll do fine with oscillation but a random walk will very often leave that range and never return, and you'll make almost no money. The payout percents are a different issue. I wasn't saying it works with arbitrary payouts. Note that the +50%/-50% model is not very realistic either when it comes to stocks or assets. The value over time rapidly spirals to zero.