paranoid android: applying Kelly to stock market

[Don Schlesinger]

> Something seems wrong with this equation. As
> my average losing trade get bigger, the
> Kelly value also gets bigger, not smaller as
> I would assume.

No, it's correct. If you can forge an advantage despite losers being larger than winners, it means that you win with great frequency and that prolonged losing streaks are rare. That's what makes it good in terms of Kelly.

On the contrary, if your winners are huge, like long shots at the racetrack, but you win very infrequently, it means that you have long losing streaks, and the Kelly fraction has to be reduced.

Don

[Dr. Jekyll]

75 % of probability to win 5 %
25 % of probability to lose 8 %

If I've a bankroll of 100 $ how much should I bet.

Bankroll after a bet:

25 % of the time = 100 -0,08 x
75 % of the time = 100 + 0,05 x

This mean we want to maximise:

(100-0,08x)^0.75*(100+0,05x)^0.25

Theorical value yeld 414

Since "x" cannot be greater than 100, the maximal value is found as 100.

However, I agree totally with Don's advice, the datas results aren't statistically significant.