Momentum vs mean reversion: two research lenses

They ask different questions about price history

Momentum looks for relative strength in past returns. Mean-reversion research asks whether weakness or a deviation from a reference level deserves investigation. Neither approach guarantees that a trend will continue or that a price will recover.

The lookback period and the exact construction matter. For example, Kenneth French’s published daily momentum factor uses prior returns from months 2 through 12 and a particular portfolio construction. Polydesk’s watchlist ranking uses different inputs and is not a reproduction of that factor or its historical returns.

How Polydesk ranks momentum

Polydesk compares one-month, three-month, six-month and one-year returns. Its default weights are 10%, 20%, 30% and 40%, respectively. Higher returns receive higher relative scores. At least two enabled return metrics must be available, unless you enable only one.

This helps you ask whether a company’s strength appears across several windows or is concentrated in one recent move. The score is calculated relative to eligible companies in your list. It is not a forecast of next month’s return, and a high score does not say whether the business is attractively valued.

How Polydesk ranks mean reversion

The current mean-reversion ranking uses one-month performance, distance from the 200-day moving average, drawdown from the 52-week high, and 14-day RSI. Default weights are 20%, 30%, 30% and 20%. Lower values receive higher relative scores. Normally at least three enabled metrics must be available; if fewer are enabled, all of them are required.

This is a weakness-oriented research screen, not a fitted model proving that a company’s price must return to an estimated fair value. A large decline can reflect a lasting change in the business. Read the disclosures behind the move before interpreting weakness as an opportunity.

Keep the two results separate

Choose the research question first, then inspect the factors behind each result. A stock can look strong over a year but weak over a month. That is not necessarily a contradiction: the methods emphasize different observations.

Avoid treating agreement between rankings as independent confirmation; several metrics come from the same price history. Scores also depend on the peer set and weights. Record the settings alongside your notes so that a later result can be compared meaningfully.

Finish with an unanswered business question rather than a trade instruction. Neither ranking models your personal financial circumstances, trading costs or ability to bear a loss. Historical price patterns can change, and an investment can lose value.

Worked example

These two return columns are insufficient to reproduce either full ranking. No future return is implied.