What this game teaches
You are playing with an edge: the coin lands heads 60% of the time, and you can bet as much or as little as you like on each flip. It is hard to imagine a better opportunity. Yet in a 2016 experiment by Victor Haghani and Richard Dewey, 61 finance students and young professionals were given exactly this game with $25 and a $250 cap. About a quarter went bust, and only about one in five reached the cap. Many bet far too much, or bet on tails after a run of heads. (Haghani and Dewey, "Rational Decision-Making Under Uncertainty: Observed Betting Patterns on a Biased Coin", 2016.)
The lesson carries straight into trading: having an edge is not enough. How much you bet decides whether the edge makes you rich or ruins you.
The maths: why betting more can make less
If you bet a fixed fraction f of your bankroll each time, a win multiplies it by (1 + f) and a loss by (1 − f). Over many flips, what matters is the average growth per flip, measured with logarithms:
| Bet each flip | Typical growth per flip | Typical bankroll after 100 flips (from ₹10,000) |
|---|---|---|
| 5% of bankroll | 0.88% | ₹24,005 |
| 10% of bankroll | 1.50% | ₹45,005 |
| 20% of bankroll | 2.01% | ₹74,899 |
| 30% of bankroll | 1.47% | ₹43,704 |
| 40% of bankroll | -0.24% | ₹7,829 |
| 50% of bankroll | -3.40% | ₹334 |
| 75% of bankroll | -21.87% | ₹0 |
Growth peaks at 20%, the Kelly fraction: for an even-money bet, Kelly = 2p − 1 = 2 × 0.6 − 1 = 20%. Bet 40% and the typical result is a small loss, despite winning 60% of the flips. Bet more and you are almost certain to be ruined, because a few losses in a row wipe out most of the bankroll and the wins that follow are on a much smaller base.
Half-Kelly (10%) gives up a little growth for a much smoother ride, which is why professionals who use Kelly usually bet a fraction of it. In trading, where your real edge is uncertain, the safe choice is even smaller, typically 0.5% to 2% of capital per trade.
Mistakes players commonly make
- Betting tails after a run of heads. The coin has no memory. This is the gambler's fallacy.
- Raising the bet after losses to "win it back", which is how most players go bust.
- Going all in after a few wins. One tail ends the game.
Frequently asked questions
Is the coin really biased 60/40?
Yes. Each game draws 100 flips in advance with a 60% probability of heads, using your browser's random number generator. The comparison at the end uses the same flips.
What is the best strategy?
Always bet on heads, and bet a steady share of your bankroll: around 10% to 20%. Twenty percent maximises long-run growth (the Kelly fraction); ten percent is smoother.
How does this relate to trading?
A trading strategy with a positive expectation is like the biased coin, except that its true win rate is unknown and changes over time. That uncertainty is why traders risk only a small fraction of capital on each trade.
Related reading
This game uses simulated prices and random outcomes for learning. It is not a trading platform, uses no real money and does not predict real markets. Everything runs in your browser; nothing you enter is sent to us. Read our disclaimer. Built and checked by Pradeep Rawal.