Who is it for?
Users with reviewable data or a trade sample Anyone testing assumptions before real capital
Statistical Strategy Analysis
Monte Carlo Lab
Resample R-multiple trade outcomes with bootstrap or permutation to study final-R, drawdown-in-R and recovery distributions.
FreeNo installationEducational & analytical
Free with a TupTrader account
The tool is ready; sign in to run it
The guide and calculation method stay public on this page. Sign-in is only required to open the interactive workspace.
Tool guide
What does Monte Carlo add to strategy analysis?
This Monte Carlo works on an R-multiple outcome sample and builds cumulative-R paths with bootstrap or permutation. The output is therefore not final account equity or an equity forecast; it describes final R, drawdown in R and recovery across resampled paths.
Logic / formula
Bootstrap resamples R outcomes with replacement; permutation uniformly reorders the fixed sample with Fisher–Yates. Percentiles are calculated from the cumulative-R path distribution.
Read the output
Read final-R and max-DD (R) percentiles as sensitivity to ordering and resampling, not as a forecast of account equity or future profit.
TT-FT-018 · versioned engine monte-carlo-strategy@2.0.0 · educational/analytical use; not investment advice.
Full guide, examples, and sources
When should you not rely on it?
A small, incomplete, or cherry-picked sample Using the result as a guarantee of future performance
Logic, formula and methodology
This Monte Carlo works on an R-multiple outcome sample and builds cumulative-R paths with bootstrap or permutation. The output is therefore not final account equity or an equity forecast; it describes final R, drawdown in R and recovery across resampled paths. Bootstrap resamples R outcomes with replacement; permutation uniformly reorders the fixed sample with Fisher–Yates. Percentiles are calculated from the cumulative-R path distribution.
Seeded bootstrap and permutation resampling, configurable path length, cumulative-R path distributions, final-R percentiles, drawdown-in-R percentiles, losing-streak percentiles, threshold probabilities and outcome histograms.
Worked example
Example
200 trades across 5,000 resampled paths
Five thousand resampled paths replace one historical ordering with distributions of final cumulative R and drawdown in R; the goal is to expose sequence risk in the observed R sample.
Five thousand resampled paths replace one historical ordering with distributions of final cumulative R and drawdown in R; the goal is to expose sequence risk in the observed R sample.
How to read the result
No single metric is sufficient for judgement. Sample size and data quality are part of the result.
Read the cumulative-R percentile fan rather than focusing on the average path. Final-R, max-drawdown-in-R, worst-path and recovery distributions describe sequence risk. Threshold probabilities are conditional on the sample and resampling method.
What the TupTrader version adds
The value of this version is not cosmetic; the result, assumptions and workflow are connected to the practical capabilities below.
- Method explainer
- Percentile R-path fan
- Final cumulative-R distribution
- Max-DD distribution
- Worst-path viewer
- Recovery distribution
- Threshold probabilities
- Scenario compare
Turn the result into a workflow
Related tool →Validate the result with a second tool. Run out-of-sample or Monte Carlo analysis. Record executable rules in the plan.
Common mistakes and limitations
- Overfitting or choosing parameters after seeing results
- Omitting costs and unfavourable trades
- Mixing incompatible samples
- Bootstrap assumes the observed sample is a useful proxy for the outcome distribution and may destroy complex temporal structure. Monte Carlo does not repair bad data or an overfit backtest.
Frequently asked questions
When should I use bootstrap versus permutation?
Permutation tests sensitivity to the ordering of the same trades; bootstrap resamples from observed outcomes to create alternative samples.
Why inspect the worst path?
Averages can hide tails; low-percentile and worst paths matter for risk tolerance and account capacity.
Does Monte Carlo prove an edge?
No. It tests path robustness under resampling assumptions; it does not establish causal edge.
Sources and research method
This guide combines the tool’s versioned specification, the earlier TupTrader educational page and the authoritative sources below. External sources are used for definitions and scientific boundaries, not for signals.
- NIST — Bootstrap confidence intervals / resampling documentationEstimating uncertainty from resampled statistics and recognizing unstable estimates.
- Bailey et al. — The Probability of Backtest OverfittingRisk of selecting strategies on in-sample performance and subsequent out-of-sample degradation.
This tool is for education, scenario analysis and process control. It is not investment advice, a buy/sell signal or a guarantee of future performance.