Establish the evidence
Record the EA version, settings, data, platform, broker profile, period and test assumptions so the result can be understood and reproduced.
The RangeState method
Evidence is useful only when the assumptions stay visible.
Every factual conclusion is tied to the customer-supplied version, settings, record and written test scope.
Six stages
From supplied record to technical evidence report.
Challenge the backtest
Review trade count, modelling quality, parameter dependence, stagnation and whether the result relies on a narrow market period.
Stress execution
Increase spread, commission and slippage to locate the point where the observed edge becomes unreliable.
Test uncertainty
Use out-of-sample periods and Monte Carlo analysis to estimate plausible drawdown, losing streaks and path variation.
Calculate portfolio relationships
Measure correlation, shared exposure, combined historical drawdown and concentration for the customer-defined portfolio.
Compare supplied records
Where matching records exist, compare observed trades, costs, latency, settings or market periods and document material differences.
Boundaries
What the method cannot do.
Historical and simulated records cannot reproduce every future market, cost or execution event. Monte Carlo analysis describes variation within a model; it does not remove uncertainty. A Supported label applies only to the written technical proposition and is not permission or advice to trade.
Read the operating boundary ↗