This article is educational research about the academic study of momentum as an equity factor. It is not investment advice. SignalStrike provides research and decision-support tools, not personalized recommendations. See the full disclosures at the end of this article.

Momentum is one of the most studied anomalies in modern finance. From Jegadeesh and Titman’s foundational 1993 paper through Fama and French’s four-factor extension, and on to the cross-asset work of Asness, Moskowitz, and Pedersen (2013), three decades of literature have established that past relative strength is a meaningful and persistent predictor of future relative strength.

But the academic finding is not the implementation. Every momentum strategy ever published — and every momentum strategy a researcher might build today — is defined by the same five tunable parameters. Change any one of them and the strategy’s behavior changes, sometimes dramatically. Understanding what these parameters are, and what the literature says about each, is the starting point for any serious work in this area.

The short version. Every momentum investing strategy is defined by five core parameters: the look-back window, the selection methodology, the weighting scheme, the rebalance frequency, and the risk filters layered on top. Adjusting any one of these can change a strategy’s behavior dramatically — which is why understanding them is the foundation of any serious momentum research.

1. The Look-Back Window

The look-back window is the period of historical price action used to rank a universe of stocks. It is the most studied parameter in the momentum literature, and the one that most defines a strategy’s character.

The academic baseline. Jegadeesh and Titman’s 1993 paper, “Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency,” tested look-backs of 3, 6, 9, and 12 months. They found persistent excess returns at every horizon in that range, with the strongest effect concentrated in the 6-12 month window. This range — sometimes referred to as “intermediate-term momentum” — became the academic standard.

The 12-minus-1 convention. A small but important detail: most academic implementations use a “12-1” look-back, meaning the trailing 12-month return excluding the most recent month. The exclusion mitigates the short-term reversal effect documented separately by Lehmann (1990) and Jegadeesh (1990). Skipping the most recent month tends to produce cleaner momentum rankings.

Shorter look-backs. A 1-month or 3-month window captures more recent price action. Strategies built on these horizons turn over more frequently, react faster to regime shifts, and are more sensitive to idiosyncratic news. They are sometimes referred to as “short-term momentum,” but the literature is more mixed on whether shorter windows produce a robust factor in their own right.

Longer look-backs. A 24-month or 36-month window approaches what some researchers call “long-term reversal” territory. DeBondt and Thaler (1985) showed that over multi-year horizons, past losers tend to outperform past winners — the opposite of momentum. Most modern implementations stay within the 6-12 month range to avoid this regime.

The look-back window does not have a single “correct” value. It has a range supported by the academic literature, and the parameter chosen says something about what the strategy is trying to capture.

2. The Selection Methodology

Once the universe is ranked by momentum score, a methodology is needed to select which stocks the strategy will hold. This is more nuanced than it sounds.

Top decile (or quintile) selection. The classic academic approach. Rank the universe, take the top 10 percent (or 20 percent), and treat that group as the “winners” portfolio. This is the methodology used in most foundational momentum papers, and it produces broad, diversified holdings.

Top-N selection. A more concentrated approach. Take the top 10, 20, or 50 names regardless of universe size. This produces more idiosyncratic risk but can amplify returns in trending markets. It also makes the strategy more sensitive to the rebalance frequency.

Filtered momentum. A two-stage approach: first filter the universe (by sector, market cap, liquidity, or fundamental criteria), then apply momentum ranking within the filtered set. Asness’s work on quality-momentum combinations falls into this category.

Cross-sectional vs. time-series. Cross-sectional momentum compares stocks against each other within a universe — the dominant academic approach. Time-series momentum (Moskowitz, Ooi, Pedersen, 2012) compares each stock against its own historical performance. The two methodologies can produce meaningfully different holdings.

The selection methodology is the parameter most likely to be conflated with the look-back window. They are not the same thing. A strategy can use a 12-1 look-back with top-decile selection, top-10 selection, sector-filtered selection, or any combination thereof.

3. The Weighting Scheme

Once stocks are selected, the strategy must decide how much weight to give each one. This is where execution-aware research diverges from theoretical academic returns.

Equal weight. Every selected stock gets the same allocation. Simple, transparent, robust to estimation error. Most academic momentum studies use equal weighting because it isolates the factor effect.

Momentum-weighted. Allocations scale with the momentum score itself. Stocks with stronger relative strength get larger weights. This amplifies the factor’s directional bet and increases sensitivity to ranking noise.

Volatility-adjusted weighting. Allocations are scaled inversely to recent volatility. Higher-volatility names get smaller weights; lower-volatility names get larger weights. This is sometimes called “risk parity” weighting within the momentum sleeve, and it tends to dampen drawdowns at the cost of upside concentration.

Market-cap-tilted. Allocations are partially weighted by market capitalization. This biases the strategy toward larger, more liquid names — useful for execution at scale, but it can dampen the small-cap momentum effect documented in some of the literature.

The weighting scheme is the parameter most often glossed over in popular treatments of momentum, but it is one of the most important determinants of realized return distribution.

4. The Rebalance Frequency

Momentum is not a static strategy. The whole point is that the ranking changes over time, which means the strategy has to be reconstituted on some cadence.

Monthly. The academic consensus middle ground. Frequent enough to capture changes in relative strength; infrequent enough that turnover and transaction costs remain manageable. Most published momentum strategies in the literature use monthly rebalancing.

Quarterly. Lower turnover, lower realized transaction costs, but slower response to regime shifts. Some institutional momentum implementations use quarterly rebalancing for cost efficiency.

Weekly or daily. Higher turnover, higher transaction-cost burden, faster response. The marginal benefit of higher rebalance frequency is heavily debated in the literature; most studies find that monthly captures most of the available factor return without the cost drag.

Calendar vs. signal-driven. A calendar rebalance happens on a fixed schedule. A signal-driven rebalance happens only when the ranking changes meaningfully — for example, when a stock drops out of the top decile by more than a threshold amount. Signal-driven rebalancing is harder to backtest cleanly but can reduce unnecessary turnover.

The choice of rebalance frequency is a trade-off between responsiveness and cost. The academic literature is most supportive of monthly cadence, but the right answer for any given research project depends on the universe, the look-back window, and the implementation environment.

5. The Risk Filters

The first four parameters define the strategy. The fifth — risk filters — defines what happens when things go wrong.

Pure momentum is known to be vulnerable to “momentum crashes.” Daniel and Moskowitz (2016) documented that the momentum factor experiences sharp drawdowns at market inflection points, particularly when the market rebounds violently after a downturn. Risk filters are how researchers attempt to manage this.

Volatility-based filters. Shift the strategy’s exposure based on realized or implied market volatility. When volatility spikes, the strategy reduces gross exposure or moves toward cash.

Liquidity and market-cap floors. Restrict the universe to names above a minimum market cap or average daily volume threshold. This is less about return and more about ensuring the strategy is implementable.

Sector and industry constraints. Cap the strategy’s exposure to any single sector or industry. This dampens concentration risk when momentum loads heavily into one part of the market.

Drawdown overlays. Activate a defensive posture (cash, treasuries, low-volatility holdings) when the strategy’s own drawdown exceeds a threshold. This is more common in tactical asset allocation than in pure cross-sectional equity momentum.

Regime-based shifts. Use a separate signal — bond-equity correlation, term structure, credit spreads — to determine whether the momentum strategy should be active at all. The academic work on regime-conditional momentum is comparatively recent and still evolving.

Risk filters are the parameter most often invisible in popular discussions of momentum, but they are usually what determines whether a strategy is researchable on paper versus deployable in practice.

Why These Parameters Matter for Research

The five parameters above are not abstract. They are the actual variables a researcher tunes when building or evaluating a momentum strategy. Two strategies labeled “momentum” can have radically different risk and return profiles depending on how each parameter is set.

This is also why methodology transparency matters in momentum research. A backtest is only as credible as the parameter set it documents. A momentum study that says “we used a 12-month look-back” but does not specify the selection methodology, weighting, rebalance frequency, or risk filters is incomplete — and possibly not reproducible.

A serious momentum research environment lets the user adjust each of these parameters independently, document the choice, and produce backtests that show the consequences. That is the analytical workflow the academic literature has been quietly demanding for thirty years.

How SignalStrike Approaches Parameter Research

SignalStrike is built as a research and decision-support environment for momentum strategies. The platform exposes each of the five parameters above as a user-controllable input:

Every backtest the platform produces is tied to a specific, documented parameter set. Results are reproducible, configurations are auditable, and the universe coverage spans roughly 1,600 to 2,000 US equities across the S&P 500, NASDAQ 100, and Russell 1000. The user defines the strategy. The platform produces the analysis. Any decision to act on any analysis remains entirely with the user.

If you are a researcher, advisor, or self-directed investor who wants to study how each of these parameters affects strategy behavior, that is the work SignalStrike was built for.

Further Reading

Related Research

Disclosures

This article is educational research about the academic study of momentum as an equity factor. It is not investment advice and does not constitute a recommendation to buy, sell, or hold any security. The strategies and parameters described are research configurations, not personalized portfolio guidance.

SignalStrike is a software platform providing research, screening, and backtesting tools. It is not a registered investment advisor (RIA) and does not provide personalized investment advice. Backtested results discussed in connection with the platform are hypothetical, do not represent actual trading, and may not reflect the impact of material economic and market factors. Past performance is not indicative of future results. All investing involves risk, including the loss of principal. SignalStrike does not custody funds or execute trades on behalf of users; users execute through their own brokerage accounts at their sole discretion.