This is a structural sample, not a real report. It shows what gets tested and how findings are written up — not what any strategy actually returned.
The instrument(s) and timeframe tested, the total history used, and how that history is split between the period the logic was developed against and periods it wasn't exposed to during development. Where relevant, whether the same logic was checked across more than one exchange feed for the same asset.
The strategy's rules are re-tested across a sequence of rolling, non-overlapping windows — each one validated against data that wasn't used to set its parameters. This section reports whether the logic held together across those windows without needing to be manually re-tuned between them, and flags any window where it didn't.
Each core input is varied across a reasonable range either side of its chosen value, one at a time, to check whether the logic's behaviour changes smoothly or whether results hinge on one specific, narrow setting. A strategy that only "works" at one exact parameter value is a strong overfitting signal; this section documents which, if any, parameters show that pattern.
A plain-language assessment of curve-fitting risk: how many free parameters the strategy has relative to the amount of data it was tested against, whether entry/exit conditions look tailored to specific historical events, and whether in-sample and out-of-sample behaviour diverge in a way that suggests the logic learned noise rather than signal.
A plain-language description of what the strategy actually does — the signals it combines, how it reaches a decision, and under what market conditions it's designed to perform well or struggle — so you have a written record of the logic independent of the code itself.
If you've already got something built — on TradingView or elsewhere — I can run this process against it as a standalone service, independent of who built it.