Residual Momentum: A Cleaner Momentum Signal

The standard academic momentum signal — sort stocks by trailing 6-to-12-month return, hold the winners, avoid the losers — is one of the most-replicated effects in finance. It also has a noisy structure: a stock’s raw past return mixes its idiosyncratic trend with its passive exposure to other factors. Strip out the factor exposures, rank by what’s left, and the signal becomes meaningfully cleaner. The technique has a name: residual momentum.

The short version. Residual momentum ranks stocks by the part of their past return that is not explained by standard factor exposures — typically market, size, and value. The remaining return is the stock’s idiosyncratic trend, isolated from the noise of other factor drifts. Blitz, Huij, and Martens (2011) showed that residual momentum produces comparable returns to standard momentum but with substantially lower crash-period drawdowns. It is now a common implementation choice in academic and quantitative momentum work.

The signal-noise problem in raw momentum

The raw 12-month return of a stock is a useful momentum signal — and a slightly noisy one. The reason is that the return measured over the past year isn’t only the stock’s own trend; it also reflects its passive exposure to other factors that happened to be doing well or poorly during that window.

If small-cap stocks broadly outperformed in a given year, every small-cap stock will look “high momentum” by raw return — not because its own trend was strong, but because its size exposure was rewarded. Similarly, if low-beta stocks led in a defensive regime, low-beta names accumulate “momentum” they didn’t really earn from their own idiosyncratic drift.

This isn’t a fatal flaw; raw momentum still works on long-run average. But it means the raw momentum ranking is contaminated by drift from other factor exposures. When those other factors mean-revert — which happens periodically, often suddenly — the apparent momentum can evaporate quickly. The big momentum crashes documented by Daniel and Moskowitz (2016) are partly explained by this dynamic: stocks ranked as winners because of their exposure to factors that then reversed get hammered when the reversal hits.

Residual momentum is the academic literature’s most direct fix.

How residual momentum is computed

The basic recipe is well-established and has minor variations across implementations:

1. Run a factor-model regression for each stock over a recent window (commonly 36 months of monthly returns). The standard choice is the Fama-French three-factor model: market excess return, size (SMB), and value (HML).
2. Extract the residuals — the part of each stock’s return in each period that is not explained by its loadings on those factors.
3. Compute the momentum signal on the residuals — typically the sum or average of the residual returns over the past 11 months, skipping the most recent month (the standard 12-1 momentum window).
4. Standardize across the cross-section — divide by the standard deviation of each stock’s residuals over the window, so the signal is comparable across stocks with different residual volatility.
5. Rank and select — use this residual-momentum score as the basis for portfolio construction in the same way you’d use raw momentum.

Different papers tweak the factor model (some add quality or low-volatility factors), the window length, and the standardization. The core idea is invariant: rank by the idiosyncratic part of past return, not the total.

The Blitz, Huij, and Martens (2011) result

The defining paper on residual momentum in U.S. equities is David Blitz, Joop Huij, and Martin Martens’s “Residual Momentum” (2011, Journal of Empirical Finance).

They constructed both a standard 12-1 momentum portfolio and a residual-momentum portfolio using the Fama-French three-factor model, over a long U.S. sample. Two findings stood out:

1. Comparable long-run returns. The residual-momentum strategy generated long-run returns broadly in line with standard momentum. Stripping out factor exposures didn’t kill the signal — it preserved most of the underlying alpha.

2. Substantially lower drawdowns. The residual-momentum strategy’s worst periods were materially less severe than standard momentum’s. In particular, the crash regimes that produced the most damage in standard momentum were significantly muted in the residual version.

The paper’s interpretation: residual momentum captures the part of momentum that’s driven by idiosyncratic trend persistence — the actual behavioral underreaction story — while pulling out the part driven by drift in other factor exposures, which is what makes standard momentum vulnerable to the kind of factor reversals that fuel crashes.

Subsequent research has largely confirmed this. Residual momentum is now treated as one of the cleaner ways to harvest the momentum premium, and is commonly used in academic and quantitative implementations seeking a higher Sharpe ratio than the raw signal can deliver on its own.

How residual momentum fits alongside other risk-aware techniques

Residual momentum is one of three distinct ways the academic literature manages momentum’s known risks. They are not mutually exclusive — they address different problems:

• Residual momentum modifies the signal itself, ranking by idiosyncratic rather than total past return. (Blitz–Huij–Martens, this article.)
• Time-series volatility scaling modifies the exposure, sizing the strategy based on its own recent volatility. (Barroso–Santa-Clara; see Risk-Managed Momentum: Scaling Exposure to Volatility.)
• Cross-sectional risk-adjusted ranking modifies the selection, ranking stocks by a risk-adjusted measure rather than raw return.

Each operates at a different point in the strategy. A program that took all three seriously would (a) build the signal from residual rather than raw returns, (b) rank stocks by a risk-adjusted version of that signal at the selection layer, and (c) scale total exposure by the strategy’s own recent volatility at the portfolio layer. That’s the maximalist version. Most real implementations choose a subset based on what they can build and operate well.

Practical considerations for implementing residual momentum

Residual momentum is a meaningful step up in implementation complexity from raw momentum. A few real-world considerations:

• Factor model choice. The standard three-factor (market, size, value) is the most-cited reference. Some implementations use four- or five-factor variants. Adding factors removes more drift contamination but also strips out some of the underlying signal. There is no single “right” choice — it’s an implementation tradeoff.
• Estimation window. Longer windows produce more stable factor loadings but slower adaptation when a stock’s true exposures change. Shorter windows are more responsive but noisier.
• Survivorship and look-ahead bias. Backtesting residual momentum requires a clean point-in-time factor estimate, not a forward-looking one. The most rigorous implementations re-estimate the factor model at each period using only data that would have been available at the time.
• Capacity and turnover. Residual momentum sometimes produces more concentrated bets in idiosyncratic-trend names, which can mean higher turnover and lower nominal capacity than raw momentum at the same notional size.

These are not arguments against residual momentum. They are reasons it’s a more involved implementation choice that benefits from a serious quantitative process behind it.

How this relates to SignalStrike

To be precise: SignalStrike does not currently implement residual momentum as a ranking mode. The platform’s available ranking choices include raw past return (pure momentum), composite scores that blend momentum with other factors, and a volatility-adjusted ranking that risk-scales each stock’s score by its own realized volatility. Residual momentum — ranking by the regression residual against a factor model — is a different technique and is not a feature of the platform today.

That precision matters. An advisor evaluating a momentum platform for client use deserves to know what the platform does and does not implement, not a generous reinterpretation of the academic literature into product claims.

What the platform does implement is the cross-sectional risk-aware approach — a different technique that addresses related concerns. Volatility-adjusted ranking divides each stock’s momentum score by its own realized volatility, tilting selection away from the most fragile high-beta names. It’s the same risk-aware spirit as residual momentum — manage the rough edges of the raw signal rather than swallowing them whole — applied with a different mechanism.

For advisors who want a residual-momentum implementation specifically, this article is educational context — a description of what the technique is and what the research shows — rather than a product claim. If you want to see how SignalStrike’s actual ranking modes behave on real historical data, you can build a configuration with any of them and backtest it. For a methodology walk-through that’s precise about what is and isn’t implemented, the full methodology is available to review, and the public portfolio tracker shows how the founders run their own strategies in live markets.

Frequently Asked Questions

What is residual momentum?

Residual momentum is a momentum strategy that ranks stocks by the part of their past return not explained by standard factor exposures — typically market, size, and value. The idea is to isolate each stock’s idiosyncratic trend from the drift caused by passive exposure to other factors, producing a cleaner signal. Blitz, Huij, and Martens documented the technique formally in 2011, showing it produced comparable long-run returns to standard momentum with substantially lower crash-period drawdowns.

How is residual momentum different from raw momentum?

Raw momentum ranks stocks by their total past return — which mixes the stock’s own trend with its exposure to other factors that happened to do well or poorly. Residual momentum ranks by the residual return after a factor-model regression, isolating the idiosyncratic component. The two approaches typically agree on most names but differ in their extremes, with residual momentum often less concentrated in stocks whose apparent momentum is really just factor drift.

Does residual momentum work better than standard momentum?

The empirical record suggests residual momentum produces comparable or slightly better long-run returns than standard momentum, but with materially lower drawdowns during the crash periods that hurt standard momentum most. The result is a higher Sharpe ratio in most samples. The tradeoff is more implementation complexity — residual momentum requires running a point-in-time factor-model regression for every stock, every period.

What factor model is used for residual momentum?

The standard academic implementation uses the Fama-French three-factor model: market excess return, size (SMB), and value (HML). Variants add quality, low-volatility, or other factors. More factors in the model strip out more drift contamination but also remove more of the underlying signal — there is no single optimal choice, and different implementations make different tradeoffs depending on their objectives.

Is residual momentum the same as beta-adjusted momentum?

They are related but not identical. Beta-adjusted momentum specifically removes the market-beta component of each stock’s past return, leaving everything else. Residual momentum is the more general case — it removes whatever factor exposures the chosen factor model includes, which for the standard implementation is market, size, and value. Beta-adjusted momentum is essentially a one-factor residual momentum; full residual momentum strips out more.

Further Reading

• Blitz, D., Huij, J., & Martens, M. (2011). “Residual Momentum.” Journal of Empirical Finance.
• Jegadeesh, N., & Titman, S. (1993). “Returns to Buying Winners and Selling Losers.” Journal of Finance.
• Daniel, K., & Moskowitz, T. J. (2016). “Momentum Crashes.” Journal of Financial Economics.
• Fama, E. F., & French, K. R. (1993). “Common Risk Factors in the Returns on Stocks and Bonds.” Journal of Financial Economics.
• For the broader context on momentum as a factor, see our pillar overview: Momentum Investing: How It Works and Why It Persists.

Related Research

• What Is Momentum Investing? A Complete Guide — Learn how momentum investing works - the academic research, why baskets beat individual picks, and why monthly rebalancing outperforms trailing stops.
• Risk-Managed Momentum: Scaling Exposure to Volatility — Risk-managed momentum scales exposure based on the strategy's own volatility. Why it improves outcomes, and how cross-sectional and time-series approaches differ.
• Momentum Investing Research: 30 Years of Academic Findings — The most-cited academic papers on momentum investing — what each study showed, why it matters, and how the field has evolved across three decades of research.

Disclosures