Quant Investment Research Expert

Problem: Why Do Pretty Backtests Fail in Production?

Quant strategies often fail because assumptions are vague, signals cannot be reproduced, data leakage goes unnoticed, and parameters are overfit. This skill treats each strategy idea as a testable hypothesis rather than a backtest screenshot.

How It Works

• Define the research question: hypothesis, universe, holding period, rebalancing logic, and expected transmission mechanism.
• Check economic or asset-pricing logic, separating factor exposure, behavioral mispricing, and statistical pattern matching.
• Specify signal inputs, ranking, thresholds, lags, and implementation timing so the signal is reproducible.
• Enforce data hygiene and test design: point-in-time correctness, survivorship bias, look-ahead bias, missing data, and in-sample/out-of-sample splits.
• After backtesting, run robustness, cost, capacity, turnover, and attribution checks before classifying the idea as continue, conditionally viable, overfit, or reject.

Scope and Caveats

Targeted at A-share research workflows, with attention to T+1, price limits, suspensions, ST, new-stock effects, and roughly 0.2% round-trip transaction costs. Outputs mark evidence strength with [actual], [inferred], [assumed]; when sample size, regime coverage, or parameter sensitivity is weak, the conclusion should be downgraded rather than oversold.