How the Data Quality Guard Works
The [Data Quality Guard (DQG)] assesses the reliability of the statistical sample before any profitability rating. It contributes to the [Investability Grade]. If any module has status REJECT, the entire strategy's data suitability is forced to zero, regardless of returns.
1. What the block shows
Score (0–100%) and Verdict (PASS / FAIL /REJECT). DQG Factor and Contribution show how this block feeds into the overall grade. A table lists each module (Gap Density, Outlier Influence, Look-Ahead Bias, Spread/Liquidity, Sampling & Over-fitting, Price Integrity) with its score and verdict.
2. Verdicts: PASS, FAIL, REJECT, N/A
- PASS: Check passed; data or logic meets the required thresholds.
- FAIL: Check did not pass but is not a critical blocker; interpret with caution.
- REJECT: Critical failure. This module contributes 0 and can force the entire DQG to fail (e.g. Look-ahead bias, too few trades).
- N/A: Insufficient raw data; score and verdict not computed.
3. Modules
Gap Density: Bar series continuity; no large holes. Outlier Influence: Profit distribution; no single trade dominates. Look-Ahead Bias: Trades executed after signal time only. Spread/Liquidity: Position size vs available volume. Sampling & Over-fitting: Enough trades and trades-per-parameter ratio. Price Integrity: No anomalous bars (spikes/flatlines). Gap Density and Price Integrity require candle data; without it they show N/A.
4. Robust Net Edge (Safe Edge)
When available, this line summarizes the Outlier Influence outcome (e.g. profit excluding top trades). It helps judge whether results depend on a few outliers.
We do not rate returns until data quality is confirmed. Fix REJECTs first, then interpret performance.
[Kiploks analysis methodology] – formulas, glossary, and FAQ.