বিষয়বস্তু · 4 / 7
Backtests and historical observations
হালনাগাদ 4 আগস্ট 2026 · 2 মিনিটের পাঠ · 10 বিষয়বস্তু
সহজ ভাষায় অর্থ
A backtest walks an explicit technical rule through stored historical bars and summarizes what happened under stated entry, exit and cost assumptions. It is a historical observation, not predictive accuracy or a promise that the pattern will repeat.
Where it appears #
Backtest is available from Screener and Strategy Lab for supported price/technical conditions. Results show the tested window, trades/observations, exits, returns, drawdowns and assumptions where available.
Step-by-step workflow #
1. Freeze the rule and universe. 2. Choose dates. 3. Review entry timing, exit mode, target, stop, maximum holding period and costs. 4. Run. 5. Inspect observation count and unavailable rows. 6. Compare with a simple baseline. 7. Test a different period without rewriting the original result.
Worked example #
A rule produces 20 completed observations: 11 positive and 9 negative before costs. The descriptive hit rate is 55%, but two large losses can still make total return negative. Five unresolved or incompatible cases remain unable and are excluded—not counted as zero-return wins or losses.
How to interpret the result #
Positive average return and higher hit rate are historical summaries, not recommendations. Negative results show the tested rules/assumptions underperformed in that sample. A small observation count is fragile. Unavailable means valid point-in-time evidence was insufficient.
Limitations and common mistakes #
Look-ahead bias, survivorship bias, parameter fitting, omitted costs, partial candles, corporate actions and changed liquidity can invalidate conclusions. Reusing the same data for discovery and evaluation overstates confidence. Unlike currencies must not be added without a disclosed conversion.
Market-specific differences #
Trading sessions, costs, price limits, liquidity and data history differ by venue. Backtests use completed exchange sessions and compatible persisted price provenance. Missing benchmark sessions or incompatible price bases should yield unable, never an invented outcome.
Educational use only #
This guide is descriptive education, not investment advice, a prediction, a recommendation, or an instruction to buy, sell, rebalance or place an order.