Risk-of-Ruin is the textbook name for one question: with a given edge and risk-per-trade, what's the probability of catastrophic drawdown over a finite horizon? The closed-form answers (Kelly, Vince's optimal-f) assume infinite horizons, constant R, and constant win-rate. None of those hold in reality.
The Onyx detector runs 10 000 Monte-Carlo paths, each 100 trades long. Every trade in every path is a coin flip weighted by your actual win-rate; winners scale equity by `(1 + avg_R_win × risk%)`, losers scale equity by `(1 - avg_R_loss × risk%)`. The path's max drawdown is recorded. We tally how often each threshold (25 %, 50 %, 75 %) is hit.
The simulation is compounding — your equity changes every trade, not fixed-dollar. That's the right model for the question "how much could I lose" because real accounts compound. Fixed-dollar models understate the tail.
Your current win-rate, avg-R-win, avg-R-loss, and risk-per-trade are read from your closed trades + `UserSettings.risk_per_trade` (defaults to 1 % when unset).
Win-rate 55 %, avg winner +1.6 R, avg loser -1.0 R, risk per trade 1.5 %. Expectancy = 0.55 × 1.6 - 0.45 × 1.0 = +0.43 R per trade — a positive edge.
Severity bands on the 50 % DD probability: - Stable — your edge + sizing combo is sustainable on the simulation horizon. Don't size up unless you have a structural reason. - Risky — edge is real but the sizing is leaving tail-risk. Either reduce risk-per-trade or improve the avg-R-win / avg-R-loss ratio (let winners run a bit longer, cap losers a bit tighter). - Critical — the math says ruin is more likely than not over time. Two roots: * negative expectancy → strategy issue. Don't treat with sizing; treat with strategy. * positive expectancy but oversized → cap risk-per-trade at your declared maximum and re-run the simulation in a week.
The Onyx-Engine assigns your simulation to one of these bands; the cutoff thresholds are TradeOnyx-internal calibration.
Three knobs move the number: - Win-rate up → DD prob drops materially at fixed sizing. - Avg-R-win up → DD prob drops similarly. - Risk-per-trade down by half → DD prob roughly quarters (the cleanest lever).
Day-to-day fluctuations in the number are simulation noise, not real change. Re-look weekly, not daily.
Tier: Pro Plus. The compute is non-trivial and the answer is one of the highest-stakes reads TradeOnyx can give. Free + Pro see the placeholder; Pro Plus sees the real Monte Carlo result.
How to read the card:
1. Hero (left) — the 50 % drawdown probability over 100 trades, with a verdict word (Stable / Risky / Critical). The single number traders should commit to memory. 2. Tail-risk distribution — three MicroStats showing P(DD ≥ 25 %), P(DD ≥ 50 %), P(DD ≥ 75 %). They tell the SHAPE of the tail. A 50 %-DD prob of 12 % with 75 %-DD prob of 0.5 % is a fat-but-bounded tail; same 12 % with 75 %-DD prob of 4 % is a runaway tail (means the strategy is martingale-like in disguise). 3. Provenance footer — "Sim: 10 000 paths × 100 trades · Based on: 55 % win-rate · +1.6 R / -1.0 R · 1.5 % risk/trade". Trust comes from showing what fed the simulation. If the number surprises you, the input row tells you which lever to question first.
The Risk-of-Ruin → Position-Sizing pairing. This card and the Position-Sizing-Consistency detector (Pro tier, lower on the page) are mirror images. Risk-of-Ruin asks "is your edge sustainable at the sizing you declared?" Position-Sizing asks "do you actually trade at the sizing you declared?" If both fire red simultaneously — declared sizing too high AND actual sizing emotional — you've found the priority fix on the page.
Don't treat Risk-of-Ruin as a single-day reading. Re-look weekly. The Monte Carlo error is small but the trade-set the simulation reads from updates on every closed trade.
Tier: Pro Plus.