Originally Posted by
PinkChip
Nice link, but unfortunately no supporting graphics in this "video", it's just merely audio. If I understood the example correctly, the player betting on the more frequent side of the coin had a winning chance of 60 percent and a losing chance of 40 percent in every toss. Since the difference is 20 percent, 20 percent is the edge on betting on the more frequent side of the coin, and thus 1/5 is the proportion of the 25 dollar bankroll which the player should bet at the start of the experiment (in this case 5 dollars). After each coin toss, the bet should be adjusted to one fifth of the current bankroll. for instance, if the player won the first bet, he has now 30 dollars, of which he should bet 6 dollars for the next toin coss. If he lost the first bet, he has now only 20 dollars, of which he should bet 4 dollars for the next toin coss, and so forth.
(It is plausible that theoretically the player could never go bankrupt, since a losing streak would decrease his bankroll asymptotically towards zero. It is an exponential function with base 0.8, the bankroll formula for losing every bet being f(t) = 25 * 0.8^t . Of course he cannot bet less than 1 cent, so there are practical restrictions to this model).
Bookmarks