The simulator runs hundreds of independent price paths, each generated from the same volatility and drift parameters but with different random outcomes. The result is not one answer — it is a distribution of answers. Reading that distribution correctly is the difference between a useful risk assessment and a false sense of confidence.
What Monte Carlo simulation does
Each simulation path generates a synthetic price series using a stochastic model calibrated to your IV and drift inputs. The grid bot runs against that price series exactly as it would in live trading: placing orders, collecting fills, paying fees and funding, and potentially hitting a stop or liquidation. After running 500 paths (the default), the simulator collects all 500 final P&L values and reports the distribution.
The key assumption is that the future will have similar statistical properties to the inputs you provide. If you set IV to 60% but the market moves at 120% vol, the simulation will be wrong. The inputs are your responsibility — the simulation is only as good as they are.
The fan chart
The fan chart is the primary output. It shows P&L over time, with five percentile bands plotted as lines:
| Band | Colour | What it shows |
|---|---|---|
| P95 | Green (dashed) | The 95th percentile — 95% of paths ended below this |
| P75 | Green (solid) | The 75th percentile — a good but not exceptional outcome |
| P50 | Amber (solid) | The median — half of paths ended above, half below |
| P25 | Red (solid) | The 25th percentile — a poor but not worst-case outcome |
| P5 | Red (dashed) | The 5th percentile — 95% of paths did better than this |
The width of the fan tells you how sensitive the strategy is to which price path it encounters. A narrow fan means outcomes are tightly clustered — the strategy performs consistently regardless of randomness. A wide fan means outcomes vary enormously depending on what price does. Wide fans are not automatically bad, but they mean you need to be comfortable with the downside, not just excited about the upside.
The median is not the expected value
The P50 line — the amber median — is the most useful single number for calibrating expectations. It is the outcome you are equally likely to beat or miss. When evaluating a setup, the median should be your baseline assumption, not the mean and certainly not the P75 or P95.
The mean (arithmetic average of all paths) can be skewed upward by a small number of extremely profitable paths where price oscillated perfectly throughout the entire range for the full period. These outlier paths are real possibilities but not representative of typical experience. Relying on the mean overstates the likely outcome.
Key metrics explained
Win rate is the percentage of paths that ended with a positive P&L. A win rate above 50% means the strategy was profitable in the majority of simulated outcomes. A win rate of 75% does not mean the strategy is low-risk — it means 25% of paths lost money, and how much they lost matters as much as the frequency.
Liquidation rate is the percentage of paths that hit the liquidation price before the trade horizon ended. Even a 5% liquidation rate means that in 1 in 20 scenarios the position was completely wiped out. If the liquidation rate is above 10%, the setup should be treated as high-risk regardless of what the median looks like.
Sharpe ratio is the annualised return divided by the annualised standard deviation of daily P&L across paths. A Sharpe above 1.0 is considered reasonable; above 2.0 is strong. A high Sharpe means the strategy generates return efficiently relative to the volatility of its own outcomes. A setup with a moderate median and a high Sharpe is often preferable to one with a higher median but erratic path-by-path results.
Calmar ratio is the annualised return divided by the median maximum drawdown across paths. It measures return relative to the typical worst intra-period loss. A Calmar above 1.0 means the annualised return exceeds the typical drawdown. This is a useful sanity check on whether the strategy can survive the kind of drawdown that is actually likely.
Reading a realistic example
Setup: neutral grid, BTC, $10,000 capital, 5× leverage Range: $90,000 – $110,000 (20%), 20 grids IV: 60% Drift: 0% Funding: 0.01%/8h Horizon: 30 days Results: P50 (median): +$420 P25: -$180 P95: +$1,840 P5: -$1,200 Win rate: 61% Liquidation rate: 4% Sharpe: 1.3 Calmar: 0.8 Median max drawdown: -$520
This is a moderate setup. The median outcome is positive but modest (+4.2% on capital). One in four paths loses money. The liquidation rate of 4% is non-trivial — roughly 1 in 25 paths ends in a wipeout. The Sharpe of 1.3 is reasonable. The Calmar below 1.0 means the median drawdown exceeds the annualised return — the strategy carries meaningful intra-period pain even on profitable paths.
What to adjust when results look wrong
If the fan is implausibly narrow, the IV input is probably too low. If every path is profitable with minimal drawdown, check that leverage is realistic and the range is not so wide that the bot almost never approaches its boundaries. If the liquidation rate is high, either reduce leverage, widen the range, or narrow the capital allocation.
The most common mistake is calibrating inputs to produce a desired result rather than to reflect realistic market conditions. Set IV from observed realised vol, set funding from current market rates, and let the simulation tell you what those inputs imply — rather than adjusting inputs until the chart looks acceptable.
Run a 30-day simulation on any configuration, then change only the IV input from 40% to 80% and observe how the fan width, liquidation rate, and Sharpe change. The relationship between vol and outcomes is not linear.
Launch the simulator →