Monte Carlo Simulation vs. Traditional Backtesting: What Traders Miss

The Single Path Problem

Traditional backtesting runs a strategy against historical price data and produces one equity curve — one specific sequence of trades in one specific market environment. As David Aronson wrote in *Evidence-Based Technical Analysis* (Wiley, 2006), "a single backtest result has a confidence interval so wide that it is statistically indistinguishable from random chance in most cases."

The problem: that sequence will never repeat exactly. The next 200 trades will come in a different order, and order matters enormously for drawdowns and survival.

What Backtesting Tells You

A well-constructed backtest answers:

• Did this strategy have an edge in this specific historical period?
• What was the maximum drawdown in this particular sequence?
• What was the win rate and average R-multiple?

These are useful inputs. But they're not predictions. According to a study by Bailey, Borwein, Lopez de Prado & Zhu (*Notices of the AMS*, 2014), the probability of finding a backtested strategy that appears profitable by chance alone exceeds 45% when testing fewer than 7 configurations.

What Monte Carlo Tells You

Monte Carlo simulation takes the trade statistics from your backtest (win rate, R-multiples) and runs thousands of randomized sequences. Instead of one equity curve, you get a distribution.

This answers:

• What is the range of possible outcomes with these trade statistics?
• What is the probability of a 30%, 50%, or total drawdown?
• What does the worst 5% of scenarios look like (P5)?
• What does the median outcome look like (P50)?

Key Differences

| Factor | Traditional Backtest | Monte Carlo Simulation |

|--------|---------------------|----------------------|

| Outputs | One equity curve | Thousands of equity curves |

| Drawdown info | One max drawdown number | Distribution of max drawdowns |

| Sequence dependency | Locked to historical order | Randomized across all possible orders |

| Survival analysis | No | Yes (probability of ruin) |

| Overfitting risk | High (curve-fitting to history) | Lower (tests robustness across paths) |

| Confidence level | Single point estimate | Percentile ranges (P5/P50/P95) |

The Combined Workflow

1. Backtest your strategy to get trade statistics

2. Feed those statistics into Monte Carlo simulation

3. Examine the P5 (worst case) and P50 (median) outcomes

4. Adjust position sizing until P5 drawdown is within your tolerance

5. If even small position sizes produce unacceptable P5 outcomes, the strategy's edge is likely too thin

Try Both Approaches

The Risk of Ruin Calculator gives you a quick Monte Carlo analysis from your backtest stats. RangeSight runs 10,000 simulation paths with real historical volatility for a complete picture.

Bottom Line

Backtesting tells you if a strategy worked. Monte Carlo tells you if it will survive. You need both.