Van Tharp R-Multiple, Expectancy & SQN100 Explained: Why a 64% Win Rate Still Goes Broke
Master initial risk (1R), endogenous 1R_habit normalization for unstopped brokerage logs, SQN100 system quality grading, and Monte Carlo risk-of-ruin math.
The Win-Rate Trap: Anatomy of a 64% Winner Losing -14.32R
Amateur traders obsess over win rate because being right feels good to the human ego (Self-Serving Bias). Yet consider our 50-trade forensic benchmark:
- 32 Winning Trades (64.0% Win Rate): Average gain
+0.57R(snatching small profits out of fear of giveback). - 13 Disciplined Stops: Average
-1.00R(🏅 Disciplined Good Losses). - 5 Emotional Bag-Holds: Ranging from
-3.20Rto-4.10R(-18.2Rcombined damage).
Net Result: -14.32R (Expectancy = -0.286R per trade). At 2% risk per 1R, Monte Carlo simulation across 10,000 paths yields a 48.4% probability of losing half your account within 100 trades.
Three Engineering Upgrades Over Textbook Formulas
- Logical Position Grouping: Scaling in 3 times and scaling out 3 times is 1 logical decision (
N = 1), not 3 separate trades. Failing to group fills distorts both win rate and√N. - Endogenous
1R_habitNormalization: Brokerage CSVs rarely store your mental stop. Instead of brittle external ATR guesses, extracting your median routine stop loss (1R_habit) reveals your true baseline risk unit immediately. - Standardized
SQN100Grading:- Below
1.6: Poor / Negative Skew Leak — halt live sizing and cap tail losses first. 1.6 – 2.4: Tradeable baseline with conservative0.5%–1.0%sizing.2.5+: Robust positive-expectancy engine.
- Below
Test your own last 20–100 closed trades instantly in our 100% client-side Account Leak X-Ray.
Scenario: You bought a clean consolidation breakout at $100.00 with a pre-committed stop at $96.00 (1R = $4.00). Six bars later, price hits $114.20 (+3.55R), then prints a 3.1x volume upper wick closing at $109.60 while QQQ shifts to Yellow. What do you commit BEFORE the next candle prints?
Frequently Asked Questions
How can I calculate R-multiples if my historical brokerage CSV has no stop-loss column?
Use endogenous 1R_habit normalization: compute the median of your regular disciplined losses after trimming the worst 15% emotional tail blowups. Our free Browser X-Ray automates this in one click.
Why use SQN100 instead of raw SQN when trade count exceeds 100?
Raw SQN multiplies by sqrt(N), artificially inflating mediocre systems with 500+ split fills. Capping at min(N, 100) and merging split fills into logical positions prevents false confidence.