You have found a bet with an edge. Excellent. Now the harder question, the one that ruins far more bettors than bad picks: how much do you put on it?
The problem, stated plainly
Stake too much and a losing run — which is certain to happen — takes your bankroll to zero. Stake too little and your edge never turns into money. Between the two lies an optimal amount, and a physicist named John Kelly worked it out in 1956, for telephone lines rather than for betting.
The formula, without the algebra
Kelly says: stake a fraction of your bankroll proportional to your edge, divided by the odds. Concretely, with a probability you rate at 55% and odds of 2.10, the formula suggests roughly 8% of your bankroll. With a thinner edge, it suggests less. With no edge, it suggests nothing at all — and that answer is the most valuable of them.
The key property: Kelly maximises the long-run growth of your bankroll. Not your gain on the next bet, which is a different objective entirely — and a dangerous one.
Why nobody uses full Kelly
Because the formula assumes your probability is exact. It never is. If you rate 55% what is really 50%, full Kelly makes you stake far too much, and the variance becomes brutal: drops of 50% of the bankroll are normal, not exceptional.
Professionals therefore use fractional Kelly — a half or a quarter of what the formula says. You give up a little growth and you buy a great deal of peace. PROLIFICK applies a fractional, capped Kelly by default, and shows the three levels (full, half, quarter) so you can see the trade-off rather than take it on trust.
What it protects you from
Kelly is not a way to win more. It is a way not to lose everything while your edge does its work. It mechanically rules out the two behaviours that empty accounts: putting everything on one match, and doubling up after a loss to get even.
The takeaway: the edge tells you whether to bet; Kelly tells you how much. Without the second, the first is worth nothing — see value betting for the other half of the pair.