Bankroll risk of ruin: watch a room of bankrolls thin out
Predict what playing twice as long does to the chance of going broke, then watch 300 simulated players with the same bankroll. You will see how bet size and time shape the risk of ruin while the expected loss stays tied to the edge.
How it works
Players stake a flat even-money bet that wins with probability (1 − edge) ÷ 2. A run lasts bankroll ÷ bet × 4 bets, so planned stakes total four times the bankroll and the expected loss is at most 4 × bankroll × edge: $21.60 for $200 on European roulette, whose 37-pocket single-zero wheel puts the edge at about 2.7%. Bet size changes the swing instead. A $200 bankroll is 40 units at $5 but 4 units at $50, and a player is out once the balance cannot cover a bet, which a short losing run can cause at 4 units but rarely at 40.
After your prediction, 300 players start with the same bankroll, and the chart traces 100 of them, turning a line red once it can no longer cover a bet, with dashed lines at the start and at zero. Set the bankroll from $50 to $1,000, the bet from $1 to $50 (capped at a quarter of the bankroll) and the game, and press 'Replay' for new random numbers. Readouts give the share still solvent, the share broke and the median finishing balance. At the defaults, $5 bets from $200 on European roulette over 160 bets, almost nobody goes broke and the median ends at $180.
The prediction asks what happens if play lasts twice as long, and the chart answers it from left to right: a player who can no longer cover a bet stays out, so the share broke can only rise as bets accumulate. With $50 bets from $200 on European roulette, about a fifth of players are out by the halfway point and close to two in five by the end of the 16-bet run. No bankroll or stake size turns a negative edge positive: in Kelly's 1956 analysis of bet sizing, a gambler seeking the fastest long-run growth places no bets when every available bet has a negative expected value.
What it leaves out: Each game becomes a flat even-money bet at its edge, with no slot prize tiers, blackjack doubles or changes of stake, and run length is fixed at four bankrolls of stakes, so session length cannot be set directly. Meant for adults (18+), the simulator is educational and has no way to improve the odds of any game.
Sources
Checked on 2026-11-05. Independent, not affiliated. 18+.
- Pennsylvania Code: 58 Pa. Code § 617a.1, Roulette ball; Roulette wheel and table; physical characteristics (current through 56 Pa.B. 6006, 5 September 2026)pacodeandbulletin.gov
- Pennsylvania Code: 58 Pa. Code § 617a.4, Payout odds (Roulette) (current through 56 Pa.B. 6006, 5 September 2026)pacodeandbulletin.gov
- Kelly JL Jr. A New Interpretation of Information Rate. Bell System Technical Journal 35(4):917-926 (July 1956), DOI 10.1002/j.1538-7305.1956.tb03809.x; scan of the journal pages, see p. 925archive.org