๐Ÿ’ฐ Lesson 3: Kelly Criterion

How much should you bet?

You have an edge. How much of your bankroll do you wager? Bet too much and a bad streak wipes you out. Bet too little and you're not growing.

In 1956 at Bell Labs, John L. Kelly Jr. derived the optimal answer:

f* = (b ร— p โˆ’ q) / b

b = decimal odds โˆ’ 1 ยท p = win prob ยท q = 1 โˆ’ p

This gives the fraction of your bankroll that maximizes long-run growth rate.

๐ŸŽฎ Kelly Calculator

Full Kelly
--%
$0
Half Kelly
--%
$0
Quarter Kelly
--%
$0
Flat 1%
1.0%
$10
Why fractional Kelly: Wharton research showed full Kelly leads to bankruptcy in nearly 100% of simulations when you misestimate win probability by even 5%. Half Kelly: also ruin. Quarter Kelly: profitable. Our default ($10 flat โ‰ˆ 1% of $1000 bankroll) is approximately quarter Kelly for typical Rank-1 picks.

๐Ÿ“Š Bankroll Growth Simulation

100 bets at the win probability + odds you set above. 4 strategies: flat $10/bet, Quarter Kelly, Half Kelly, Full Kelly. Same wins/losses for each strategy (same RNG seed) so you can compare.

Flat $10 Quarter Kelly Half Kelly Full Kelly
Starting bankroll: $1000. 100 bets. Same simulated outcomes across all 4 strategies.