What "risk of ruin" actually measures
Risk of ruin is the probability that a trading strategy with positive expected value blows up the account anyway because of bad variance — too many losses in a row, sized too aggressively. It is the single most under-appreciated concept in retail trading. Most traders who lose money are not losing it because their edge is gone; they are losing it because they sized positions for the average outcome and ignored the geometric reality of compounding losses. A 50%-win-rate strategy that pays 2R when it wins is a brilliant strategy. The same strategy run at 10% risk per trade goes to zero faster than a 30%-win-rate strategy run at 1%.
The math is unintuitive because losses compound differently than gains. A 50% loss requires a 100% gain to recover — a 90% loss requires a 1000% gain. This asymmetry is what the Kelly criterion was invented to handle. Kelly tells you the position size that maximizes the long-run growth rate of your capital, taking the geometric reality of compounding into account rather than the additive average. For binary win-loss bets it has a clean closed form, and that is what the calculator above implements.
Why "1% per trade" is folklore, not math
Trading books and courses universally recommend "risk 1% per trade", and that recommendation is so ubiquitous that traders treat it as a law. It is not a law. It is a heuristic that happens to be roughly correct for traders with mediocre edge and limited self-knowledge — a 50%-win-rate strategy with 1.5R wins, run at 1% per trade, has a vanishingly small risk of ruin over any realistic horizon. But a trader with a measured 60% win rate and 2R wins is leaving enormous compound growth on the table at 1%. Their full Kelly is around 35%, and even quarter-Kelly is 8-9%. They could responsibly run 5% per trade and grow capital at multiples of the rate they get at 1%.
Conversely, a trader with a 45% win rate and 1.2R average wins has a slight positive edge but full Kelly of only 4%, and quarter-Kelly of 1%. For them the folklore is correct. For everyone else, the folklore is either too aggressive or far too conservative — and most traders never measure their actual win rate and average R, so they have no way to know which side of the line they sit on.
Why ¼ Kelly instead of full Kelly
Full Kelly is mathematically optimal under one critical assumption: you know your true win rate and average R exactly. In trading you never do. You estimate them from a sample of past trades, and your estimate has standard error inversely proportional to the square root of sample size. With 100 trades of history, your win-rate estimate is plus-or-minus 5 percentage points. With 50 trades, plus-or-minus 7 points. Plug a win-rate that's 5 points too optimistic into Kelly and you can be running 2x the safe sizing — turning a near-zero risk-of-ruin into a meaningful one.
The conventional fix is fractional Kelly: run at ¼ to ½ of the full Kelly fraction. This sacrifices about 30% of your long-run growth rate in exchange for cutting your drawdown variance roughly in half and your risk-of-ruin to negligible levels for any reasonable estimation error. It is the trade-off all serious quant funds make. The calculator above defaults to ¼-Kelly values because that is what the literature converges on for traders working with empirical-not-theoretical edge.
How to use this with your actual trading data
Plug in your real numbers. Open your journal, count your last 100 trades, compute the percentage that closed positive — that is your win rate. Sum the R-multiples of your winning trades and divide by the count — that is your average win R. Most traders never do this and instead size based on what they remember about their last 5 trades, which is exactly the bias the math is trying to correct against. The "median ending capital" output assumes you take all `N` trades; the "P(end below 50% capital)" output is the closed-form probability that your terminal balance lands below 50% of starting capital — a strict lower bound on the more meaningful "did I touch 50% drawdown at any point along the way" question. The path-dependent number is typically 30-50% higher than the terminal one and matters psychologically far more — measure it by running a Monte Carlo simulation on your real trade distribution, not by reading this card.
Once you start sizing positions by edge rather than by feeling, the boring discipline pays out. Most retail blowups are not from bad strategies; they are from oversizing relative to the edge being measured. The calculator above is the math that makes the trade-off explicit. The journaling discipline that lets you measure win rate and R-multiple accurately is what makes the math actionable — and that is the part PerpLog automates.
Related
- → Position Size Calculator (Bybit Perpetuals) — once you know your risk percent, this gives you the contract size.
- → Liquidation Price Calculator (Isolated + Cross) — pair with this once your size is set.
- → Funding Cost Calculator (over time) — the silent tax on multi-day holds.
- → MFE/MAE for Crypto Perpetuals — improving your average R systematically.
Stop guessing your win rate. Measure it.
PerpLog computes your real win rate, average R, expectancy, and risk-of-ruin from your synced Bybit trade history — automatically and continuously. Free tier, no credit card.