Monte Carlo simulation takes your strategy's historical trade results, shuffles the order randomly thousands of times, and shows you the full range of possible outcomes — not just what happened, but what COULD have happened if the same trades arrived in a different sequence. This is critical because a backtest shows only one path through history. Monte Carlo reveals that the same strategy with the same trades could have produced a maximum drawdown ranging from -8% (lucky sequence) to -25% (unlucky sequence). The 95th percentile worst-case drawdown is the number you should plan for.
A backtest that shows 15% max drawdown might have a Monte Carlo 95th percentile of -22%. This means there is a 5% chance that your live trading — using the exact same strategy with the exact same edge — will experience a -22% drawdown simply from the random ordering of wins and losses. The drawdown management guide and the Drawdown Calculator should use this Monte Carlo worst-case — not the single-path backtest result — for protocol thresholds and position sizing.
Why a Single Backtest Is Not Enough
A backtest tells you what happened. Monte Carlo tells you what COULD happen. The distinction is the difference between false confidence and genuine risk awareness.
Consider a strategy backtested over 5 years with 200 trades. The backtest shows: 55% win rate, 2:1 reward-to-risk, 25% annual return, 12% maximum drawdown. These results look excellent. But they represent ONE ordering of those 200 trades — the exact historical sequence.
Now imagine shuffling those 200 trades randomly. The wins and losses are the same. The total return is the same. But the order changes. In one random shuffle, the first 8 trades might all be losers — producing a -12% drawdown before any recovery. In another shuffle, winners and losers alternate evenly — producing a maximum drawdown of only -4%. Same strategy, same edge, dramatically different drawdown experiences.
Monte Carlo simulation runs this shuffle 1,000-10,000 times and shows you the full distribution of outcomes. You see the median drawdown (the typical case), the 5th percentile (the best reasonable case), and the 95th percentile (the worst reasonable case). The 95th percentile is the number that matters for risk planning — it answers: 'What is the worst drawdown I should realistically expect from this strategy?'
Without Monte Carlo, traders plan for the backtest drawdown (12% in the example above). When live trading produces a 20% drawdown (which Monte Carlo would have shown as the 90th percentile), they panic — abandoning a strategy that is performing within its expected range. Monte Carlo prevents this by establishing realistic expectations before going live.
The Backtesting Simulator provides the trade data that feeds Monte Carlo analysis. Chapter 17 of the free trading book covers statistical validation methods including the principles behind simulation-based stress testing.
How Monte Carlo Simulation Works
The process is conceptually simple, even though the mathematics behind it are sophisticated.
Step 1: Collect your trade results. You need a list of individual trade outcomes (in R-multiples or percentage returns). Example: +2.5R, -1R, +1.8R, -1R, +3.2R, -1R, -1R, +2.1R... at least 50 trades, ideally 100+.
Step 2: Shuffle and simulate. The simulator randomly reorders these trades and calculates the equity curve for each random ordering. One simulation might put the three -1R losses at the beginning (producing a deeper early drawdown). Another might spread them evenly. Each simulation uses the SAME set of trade results — only the sequence changes.
Step 3: Repeat 1,000-10,000 times. Each repetition produces a different equity curve with different drawdown and return characteristics. After 1,000 simulations, you have 1,000 possible equity curves — all statistically valid given your strategy's edge.
Step 4: Analyse the distribution. Sort the 1,000 maximum drawdowns from smallest to largest. The 50th (median) represents the typical experience. The 950th (95th percentile) represents the worst case you should plan for. The range between the 5th and 95th percentile is the confidence interval — your strategy's outcomes will fall within this band 90% of the time.
The critical outputs are the 95th percentile maximum drawdown (for setting your drawdown protocol thresholds), the median annualised return (for setting realistic profit expectations), and the probability of ruin (the percentage of simulations that breach your maximum tolerable drawdown).
| Percentile | Drawdown | Meaning | Use For |
|---|---|---|---|
| 5th | -6% | Best reasonable case | Not useful for planning (too optimistic) |
| 25th | -9% | Good outcome | Minimum drawdown to prepare for |
| 50th (median) | -13% | Typical outcome | Set baseline expectations |
| 75th | -17% | Below-average luck | Drawdown protocol Tier 2 threshold |
| 95th | -22% | Worst reasonable case | Drawdown protocol Tier 3 threshold |
| 99th | -28% | Extreme worst case | Maximum position sizing constraint |
Interpreting Monte Carlo Results
Three outputs from Monte Carlo analysis should inform every trading decision.
Output 1: 95th Percentile Maximum Drawdown. This is the single most important Monte Carlo output. It tells you: 'In 95 out of 100 possible futures, your maximum drawdown will not exceed this number.' If the 95th percentile drawdown is -22%, your drawdown protocol should have its Red Zone (Tier 3: close all positions) set at -22% — not at the backtest's -12%. The backtest drawdown was one lucky path; Monte Carlo shows the range.
Output 2: Probability of Ruin. What percentage of simulations breach your maximum tolerable drawdown (typically -50%)? If 2% of 1,000 simulations breach -50%, your risk of ruin is approximately 2%. The risk of ruin guide covers why keeping this below 1% is essential for long-term survival. If Monte Carlo shows ruin probability above 5%, your position sizing is too aggressive — reduce per-trade risk until the probability drops below 1%.
Output 3: Return Distribution. Monte Carlo shows not just one return number but a range. The median annual return might be 18%, but the 25th percentile might be 8% and the 75th percentile might be 28%. This range sets realistic expectations: in any given year, your return will likely fall between 8% and 28%, with 18% being the most probable. Traders who expect exactly the backtest return (say, 22%) become frustrated and abandon the strategy if they achieve 'only' 12% — which Monte Carlo would show is within the normal range.
The practical rule: always plan for the 75th-95th percentile drawdown (the bad scenarios) while expecting the 25th-50th percentile return (conservative estimates). This asymmetric planning — prepare for the worst, expect the moderate — produces the psychological resilience that sustains long-term trading.
Running Your Own Monte Carlo Analysis
You do not need expensive software or programming skills to run Monte Carlo analysis. Here is the accessible approach.
Method 1: Spreadsheet (Excel/Google Sheets). Export your trade results (R-multiples) from the Trade Journal. In a new column, use the RAND() function to assign a random number to each trade. Sort by the random number (shuffling the trade order). Calculate the cumulative return and maximum drawdown for this shuffled sequence. Copy the sheet 100 times (each with a different random shuffle) and record the max drawdown from each. After 100 iterations, you have a drawdown distribution. This method is manual but requires zero coding.
Method 2: Online Monte Carlo calculators. Several free tools accept a list of trade returns and run 1,000-10,000 simulations automatically, outputting drawdown distributions and confidence intervals. Search for 'Monte Carlo trading simulator' and input your trade history.
Method 3: Python (for quantitative traders). A Monte Carlo simulation in Python is approximately 20 lines of code using NumPy. Generate random permutations of your trade array, compute the equity curve for each, and extract drawdown statistics. This method provides the most flexibility and allows 10,000+ simulations.
Regardless of method, the minimum input is 50 trade results (ideally 100+). With fewer than 50 trades, the Monte Carlo distribution is not statistically meaningful — the range of outcomes is too wide to be actionable. The Backtesting Simulator produces 100+ trades across 20 years of data, providing the ideal input dataset.
Chapter 5 of the free trading book covers position sizing principles that Monte Carlo analysis informs — connecting the simulation output to practical allocation decisions.
Monte Carlo and the Grade A-E System
Monte Carlo analysis provides specific insights when applied to the Grade A-E conviction system.
Test 1: Grade A-only versus mixed-grade Monte Carlo. Run separate Monte Carlo simulations for Grade A trades only, Grade A+B combined, and all Grades A-C. The Grade A-only simulation typically shows a tighter drawdown distribution (less variance between best and worst case) because Grade A trades have higher win rates and higher payoff ratios. The all-grades simulation shows a wider distribution because Grade C trades introduce more variance. This confirms the Grade A selectivity advantage in risk-adjusted terms.
Test 2: Regime-filtered versus unfiltered. Run Monte Carlo on trades taken WITH the macro regime filter versus trades taken WITHOUT it. The filtered simulation shows lower 95th percentile drawdowns because the regime filter eliminates the most damaging trades (longs during Stagflation, shorts during Goldilocks). The unfiltered simulation includes these regime-contrary trades, which cluster into losing streaks during transitions.
Test 3: Position sizing sensitivity. Run Monte Carlo at your current sizing (e.g., 2% risk per trade) and at 50% of current sizing (1% risk). The 95th percentile drawdown drops roughly proportionally — from -22% to approximately -12%. If the lower-sizing drawdown is more psychologically tolerable, the reduced compounding rate may be a worthwhile trade-off for sustainability.
The combined insight: Monte Carlo analysis confirms that the Grade A-E system's selectivity and regime filtering produce a narrower, more predictable range of outcomes than less selective approaches — validating the core methodology on a statistical rather than theoretical basis.
The Position Size Calculator should be calibrated to the Monte Carlo 95th percentile drawdown — ensuring your per-trade sizing keeps the worst-case outcome within your tolerance.
Common Monte Carlo Mistakes
Four mistakes reduce the value of Monte Carlo analysis or produce misleading conclusions.
Mistake 1: Too few trades in the input. Monte Carlo shuffles the trades you give it. If you input only 20 trades, the shuffled distributions are nearly identical — there are not enough data points to produce meaningfully different orderings. Minimum: 50 trades. Ideal: 100+. Below 50, the Monte Carlo output is not actionable.
Mistake 2: Ignoring trade dependencies. Standard Monte Carlo assumes each trade is independent (the outcome of one trade does not affect the next). In reality, losing streaks sometimes cluster because they are caused by the same macro event (regime shift). Advanced Monte Carlo techniques (block bootstrapping) preserve these dependencies by shuffling blocks of consecutive trades rather than individual trades. If your strategy has regime-dependent performance, use block bootstrapping for more realistic results.
Mistake 3: Using Monte Carlo to validate a bad strategy. Monte Carlo stress-tests the RANGE of outcomes for a strategy with a given edge. It does not evaluate whether the edge exists. If your strategy has negative expectancy (as shown by the expectancy calculation), Monte Carlo will show that 100% of simulations lose money — which confirms the obvious. Always verify positive expectancy through backtesting BEFORE running Monte Carlo.
Mistake 4: Planning for the median instead of the tail. The median Monte Carlo outcome (50th percentile) is the most likely single outcome — but you do not experience the median. You experience one random path, which could be anywhere in the distribution. Plan your drawdown protocol for the 95th percentile (the bad tail) and your return expectations for the 25th-50th percentile (conservative). This asymmetric planning is what separates professionals from amateurs.
- 1.Monte Carlo simulation shuffles your trade results thousands of times to show the full range of possible outcomes — not just what happened, but what COULD happen from the same strategy with different luck. The 95th percentile worst-case drawdown (not the single backtest drawdown) should set your drawdown protocol thresholds.
- 2.A backtest showing 12% max drawdown might have a Monte Carlo 95th percentile of -22%. This means there is a 5% chance of experiencing -22% from the same edge purely from unlucky trade sequencing. Planning for 12% and experiencing 22% causes panic and strategy abandonment; planning for 22% and experiencing 15% builds confidence.
- 3.Run Monte Carlo with minimum 50 trades (ideally 100+), use the 95th percentile drawdown for protocol planning and position sizing, and test Grade A-only versus mixed-grade to confirm the selectivity advantage. Always verify positive expectancy through backtesting before running Monte Carlo — simulation stress-tests the range, not the edge.
Monte Carlo simulation takes your historical trade results and randomly reorders them 1,000-10,000 times to show the full range of possible outcomes. Each simulation uses the SAME trades in a DIFFERENT random sequence, producing different equity curves, drawdowns, and return paths. The output shows you not just what happened (the backtest) but what could happen — including the worst-case drawdown you should realistically prepare for.
A backtest shows one path through history — the specific sequence of wins and losses that occurred. Monte Carlo shows thousands of possible paths, revealing that the same strategy could produce a maximum drawdown anywhere from -6% to -28% depending on trade sequencing. Without Monte Carlo, you plan for the backtest drawdown (one outcome). With Monte Carlo, you plan for the 95th percentile worst case (realistic worst outcome), which is typically 1.5-2x the backtest drawdown.
Minimum 50 trades, ideally 100+. With fewer than 50 trades, the Monte Carlo distribution is not statistically meaningful — there are not enough data points to produce significantly different orderings. The Backtesting Simulator produces 100+ trades across 20 years of data, providing an ideal input dataset. Your live Trade Journal can also feed Monte Carlo analysis after accumulating 50+ entries.
Use the 95th percentile for worst-case planning (drawdown protocol thresholds and position sizing). Use the 25th-50th percentile for return expectations (conservative to moderate). This asymmetric planning — prepare for the bad tail, expect the moderate middle — produces psychological resilience. Traders who plan for the median outcome (50th percentile) are unprepared when they experience the 80th or 90th percentile, which happens 20% and 10% of the time respectively.