Why a BigQuant Stock Strategy Can Backtest Without Live Signals
Summary
This forum post presents a Chinese stock strategy that reportedly produced backtest trades but no live signals. Its example queries daily stock factors for non-ST, non-suspended companies outside specified listing sectors, then applies filters for minimum float market capitalization, listing age, profit and revenue growth, liabilities relative to assets, price-to-earnings ratio, and price-to-book ratio. On each trading day, the handler sorts qualifying stocks by total market capitalization, takes up to three, and submits equal-weight target orders on a counter-based schedule.
The post supplies code and the reported symptom, but no accepted diagnosis or demonstrated fix. The sample includes a likely logic issue: it sets target weights before checking whether the daily selection is empty, which can divide by zero. It also queries a date range tied to the backtest context and may depend on data availability or live-run date alignment. Since the example leaves the per-day top-stock grouping commented out and does not explain live signal requirements, readers would need to inspect data coverage, dates, scheduling, and order handling in their own deployment. No performance evidence is offered.
Key ideas
- The example filters stocks using size, listing age, growth, leverage, valuation, trading status, and sector fields.
- It ranks each day’s qualifying stocks by total market capitalization and targets up to three positions.
- Orders are submitted when a counter reaches the specified rebalance interval.
- The code sets weights before checking for an empty selection, creating a potential division-by-zero error.
- The post reports a backtest-to-live signal mismatch but provides no verified diagnosis or fix.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.