Momentum Factor
A-share momentum factors: medium-term relative strength
Momentum ranks stocks by relative performance over a prior window. It can reverse sharply, especially in crowded or fast-changing regimes.
Typical direction
Higher trailing return is commonly ranked higher after a defined skip period.
Data
Adjusted price history, trading status and index membership
Refresh
20, 60 or 120 trading days; often monthly rebalance
Research hypothesis
Write the hypothesis before reading the backtest
Medium-term relative strength may persist for a time, but the relationship must be tested with price-limit, suspension and turnover assumptions.
A factor is a testable research hypothesis, not an investment recommendation or return promise.
Factor health card
Pre-backtest checks for this factor
Research purpose
Test whether medium-term strength remains informative after execution constraints.
Refresh and rebalance
Commonly monthly; test window and skip-period sensitivity.
Data timing
Use one adjusted-price convention and explicit trading-status rules.
Neutralisation
Review sector and size exposure before interpreting results.
Overlapping exposures
Often paired with low volatility or quality; distinguish it from reversal windows.
Check before use
Model fast reversals, limit-up/limit-down fills and turnover costs first.
Definitions
Core measures
Trailing return
Adjusted price(t) ÷ adjusted price(t−N) − 1Define N and the skip period in advance.
Relative strength rank
Cross-sectional rank of trailing returnCan be sector-neutralised.
Momentum spread
Top-group return − bottom-group returnReview both legs and implementability.
Turnover
Portfolio traded value ÷ portfolio valueOften determines net results.
Research protocol
Keep the same research conventions across factors
Data availability
Financial, dividend and share data become available on actual disclosure or implementation dates, not report-period end dates.
Universe and exclusions
Document index membership, listing age, ST, suspensions, delistings and missing-data rules.
Processing and neutralisation
Version winsorisation, standardisation, sector/size neutralisation and missing-value rules.
Tradability
Include price limits, suspensions, participation, fees, slippage and market impact.
Out-of-sample review
Report IC, grouped returns, exposures, turnover and rolling out-of-sample evidence together.
Build and validate
What to test
- 1Compare 20, 60 and 120-day windows.
- 2Test a short skip period separately.
- 3Report gross and net results under tradability constraints.
Common pitfalls
- ×Ignoring sharp reversal regimes.
- ×Assuming limit-hit stocks can trade at the close.
- ×Calling a hot-sector exposure stock selection.
A-share implementation
A-share checks that belong in the backtest
- Use the actual disclosure or implementation date; do not make a field available at the report-period end date.
- State the universe, listing-age, ST, suspension, delisting and missing-data rules before running the backtest.
- Model price limits, suspensions, fees, slippage and participation limits instead of assuming every close can be traded.
- One-word limit-up or limit-down days may make the desired rebalance impossible.
Research prompt
A reviewable starting prompt
“In CSI 300 stocks, build a sector-neutral 120-day momentum signal with a five-day skip period. Model price limits, suspensions, slippage and participation limits; report gross and net performance, turnover and rolling out-of-sample results.”
FAQ
Why use a skip period?
A short skip period can reduce overlap with very short-term reversal. The choice is a testable assumption, not a default truth.
Is momentum tradable in A shares?
It depends on the universe, turnover, price limits and participation assumptions. Assess net results, not only a close-to-close signal.