VolumeLeaders // Kelly AP
Asymmetric payout · bet sizing simulator

Kelly Criterion Simulator

The coin lands heads 60.0% of the time. Choose a side, size your bet, and see what over- or under-betting the edge does to a bankroll — including asymmetric reward:risk payouts.

01 Configure
Coin
Your edge per trade, betting heads +20.0%
Kelly optimal bet20.0%
Bet sizing
Bet$ this flip$2,000.00
Pace & limits
Bounds are clamped to $0–$1,000,000,000,000 and can't cross each other.
02 Run
Heads
Tails
Space flip the coin
Enter start / stop auto-spin
03 Results
Balance
$10,000.00
flat since start
Win rate0.0%
Trades0
Elapsed0d
Best streak0
Worst streak0
Balance hit $0 — betting stopped. Reset to try again.
Balance hit $1,000,000,000,000 — simulation complete. Reset to run it again.
FIXED · log +0.00% · stake 0.0% · vol 0.00%
Bankroll over time
0 trades · through
Study context

This setup echoes a well-known 2016 experiment by Haghani and Dewey, in which participants were given a real bankroll and a coin they were told would land heads 60% of the time — a genuine, sizeable edge. Despite that edge, a large share of players still went bust, mostly by betting far more than the edge justified, switching sides after a few losses, or chasing losses with bigger stakes. The lesson isn't just "the math has an optimum" — it's how easy it is to abandon that optimum under pressure, even when the odds are known and in your favor.

This version adds an adjustable reward:risk payout so you can model asymmetric edges — low hit-rate / high payoff strategies, not just even-money coins. A 30% win rate at 3:1 can still be +EV; Kelly sizing changes with both probability and payout.

A 60% coin is generous, though — it's a teaching device. Most real trading edges are nowhere near that clean. Discipline means sizing every trade the same principled way without inflating after wins or abandoning after losing streaks — both ways of quietly giving up the edge you're trying to harvest.

The hardest skill isn't finding an edge — it's telling the difference between "this losing streak is just variance" and "my edge is actually gone," and behaving correctly under each.