Start with an evidence package
A backtest is more than a score
A return figure by itself leaves out the context needed to understand it. Before comparing versions, preserve the strategy rules and the test conditions that produced the report. This makes a later diagnosis or refinement reviewable rather than a guess.
Strategy baseline
The exact version, explicit rules, market, timeframe or bar type, date window, and long/short settings.
Test assumptions
Sizing, commissions, slippage, session choices, data source, and any platform settings that could change the result.
Backtest output
A supported CSV or XLSX export, or a PNG, JPG/JPEG, or WEBP report image from the platform you tested.
Research question
The specific behavior you want to understand before proposing a change to the rules.
NinjaTrader evidence first
Bring the Strategy Analyzer output you already use
The Strategic Edge AI organizes supported NinjaTrader backtest evidence inside Evolve; TradingView exports remain supported secondarily. It does not run a brokerage account, execute trades, or replace the platform where you independently compile and run the strategy.
| Evidence type | Supported example | Why it belongs with the analysis |
|---|---|---|
| Structured export | CSV or XLSX report | Preserves rows and metrics for a more reviewable discussion of the test. |
| Visual report | PNG, JPG/JPEG, or WEBP | Retains the settings or equity context visible in a platform report. |
| Rule and source context | Strategy Specification and exact NinjaScript version (or supported Pine Script version) | Helps identify whether the tested implementation still matches the intended explicit rules. |
Read the result as a system
Metrics are evidence, not universal passing scores
Profit factor, win rate, maximum drawdown, and Sharpe or Sortino can help describe a fixed historical test. Their meaning depends on the full set of rules, assumptions, sample, and trade-offs; no single threshold makes a strategy ready or reliable.
Year and regime
Separate performance by year and relevant trend, range, and volatility conditions before treating an aggregate as stable.
Long versus short
Compare direction-specific trade count, expectancy, drawdown, and dependence before changing both sides together.
Hour and weekday
Look for time-of-day or day-of-week concentration, while treating small segments as hypotheses rather than filters to add automatically.
Drawdown and loss sequences
Review maximum drawdown, clustered losses, consecutive-loss behavior, and the path taken to reach the final result.
Trade distribution and outliers
Inspect frequency, average win/loss, excursion when available, and whether a small set of trades dominates the outcome.
Cost and parameter sensitivity
Test whether modest slippage, commission, or parameter changes erase the apparent edge and flag brittle regions as overfitting risks.
Evolve from a question
Move from the report to one controlled test
A backtest analyzer should not automatically optimize a strategy. The useful outcome is a documented question that can be independently tested against the same baseline.
Keep the tested version and its assumptions as the baseline.
State the observed behavior in a way that can be checked.
Write one falsifiable rule-change hypothesis.
Run the next independent test and compare the stated trade-offs.
Retain, reject, or investigate the change in a version log.
Keep the comparison clean. If multiple rules, data choices, or test settings change at once, the observed difference may not explain what caused it.
Turn evidence into the next testable question
Start with a strategy version and a real test report, then keep the next change explicit, bounded, and independently reviewable.
Educational strategy-development content only. The Strategic Edge AI does not provide trade signals, automated execution, or financial advice.