Taxonomy of Velocity: Defining the Types of Momentum Trading

Navigating the professional landscape of directional persistence, relative strength, and quantitative alpha models.

Cross-Sectional (Relative) Momentum

Relative Momentum, also known as cross-sectional momentum, is the methodology of ranking a universe of assets against one another. The objective is to identify the "leaders" within a group. It asks: Which horse is the fastest in the race? This strategy relies on the dispersion of returns across different securities. By buying the top decile of performers and frequently selling the bottom decile, the trader captures the behavioral herding effect that occurs as capital rotates toward winners.

In an institutional equity factor model, relative momentum typically uses a 12-month lookback (often excluding the most recent month to avoid short-term mean reversion). Because this strategy is beta-dependent, the portfolio remains long the "best" stocks even during broad market collapses. While the strategy generates significant alpha in bull markets, it is susceptible to "Momentum Crashes" when the market regime shifts suddenly and the previous leaders become the primary targets of liquidation.

Institutional Context: Relative momentum is the primary "Selection Factor" for active equity funds. It assumes that information travels sequentially through a sector, and those stocks reacting first provide a signal for the subsequent movement of the entire group.

Time-Series (Absolute) Momentum

Absolute Momentum, or Time-Series Momentum (TSMOM), analyzes each asset in isolation relative to its own history. It ignores the "horse race" and asks a binary question: Is the price higher than it was [N] periods ago? This strategy focuses on Directional Persistence. If an asset has a positive 12-month return, the system is long; if negative, the system moves to cash or initiates a short position.

The primary utility of absolute momentum is capital preservation. Because it can move entirely into cash or short positions, it provides a structural defense against secular bear markets. This is the foundation of the Managed Futures industry and is uniquely capable of generating Crisis Alpha—profits during periods of extreme market dislocation where correlations across all asset classes tend to converge.

Dual Momentum: The Composite Model

Popularized by Gary Antonacci, Dual Momentum synthesizes relative and absolute models into a single decision engine. This framework uses relative momentum to select what to buy (the leader of the pack) and absolute momentum to decide when to be in the market (the directional filter).

# The Dual Momentum Rule Logic 1. Relative Signal: Compare 12-month returns of [US Stocks] vs. [International Stocks]. Select the Winner (W). 2. Absolute Signal: If Winner (W) return > T-Bill return: ACTION: Go Long (W). 3. Defense Trigger: If Winner (W) return < T-Bill return: ACTION: Move to Cash/Aggregate Bonds.

Tactical and Technical Momentum

While institutional models focus on multi-month windows, Tactical Momentum operates on shorter timeframes—intraday to multi-day. This type of momentum is identified through geometric price structures and volatility profiles.

Gap and Go Momentum

Intraday momentum triggered by news catalysts. It identifies assets opening above structural resistance and uses the first 5-15 minutes of session liquidity to capture immediate displacement.

Volatility Contraction (VCP)

Swing-based momentum where price "coils" into a narrow range. The burst from this contraction signals institutional supply absorption and the start of a new high-velocity trend.

Macro and Sector Rotation

Sector Rotation applies momentum to the eleven GICS sectors (Technology, Healthcare, Energy, etc.). It assumes that macroeconomic cycles—driven by interest rates and inflation gradients—impact industries sequentially. By monitoring the Relative Strength Gradient of these sectors, macro allocators move capital into the "line of least resistance" as the business cycle evolves from expansion to contraction.

Quantitative and Linear Regression

Linear Regression Momentum moves beyond point-to-point percentage changes to analyze Trend Quality. It fits a best-fit line through log-prices to measure the slope (velocity) and the R-squared (consistency).

A high-quality momentum trade is defined by a "smooth" ascent. Professional quants penalize "jumpy" momentum—price spikes driven by thin liquidity—and prioritize assets where price hugs the regression line. This mathematical precision reduces whipsaws and identifies trends supported by steady institutional accumulation rather than speculative mania.

Machine Learning and Neural Momentum

The frontier of the field involves Non-Linear Momentum models using deep learning architectures like LSTMs (Long Short-Term Memory) or Random Forests. These models do not look for a single factor; they identify high-dimensional patterns across thousands of features including volume flow, sentiment, and cross-asset leads.

Random Forest models use multiple decision trees to identify the specific market state (volatility, volume, price speed) that supports persistent trends. By averaging these "votes," the model reduces signal noise and identifies structural momentum that traditional indicators miss.

Neural architectures like LSTMs maintain "memory" of previous price sequences. They distinguish between a breakout that is a legitimate trend reversal and a "bull trap" by analyzing the temporal context of the preceding consolidation period.

Systematic Comparison Matrix

Type	Core Question	Lookback	Primary Benefit
Relative	Who is the fastest?	6 - 12 Months	Alpha in bull markets
Absolute	Is it moving forward?	12 Months	Bear market protection
Technical	Is the coil bursting?	1 - 20 Days	High reward-to-risk
Quant	How smooth is the path?	90 - 250 Days	Reduced noise/whipsaws
Machine Learning	What is the hidden shape?	Multi-dimensional	Regime adaptability

Strategic Synthesis

Momentum trading is not a singular strategy but a Spectrum of Philosophy. The choice between absolute, relative, or neural models depends entirely on the investor's objective—whether it is capital preservation, maximum alpha capture, or high-frequency income.

The most robust portfolios utilize an ensemble approach, combining the defensive "Absolute" filters of Time-Series models with the high-alpha "Relative" selection of Cross-Sectional models. By understanding the physical laws of price velocity and institutional liquidity displacement, the trader transforms the market from a chaotic random walk into a structured environment of probabilistic momentum.

Institutional Risk Disclosure: All momentum strategies involve significant exposure to directional risk. Relative models are subject to momentum crashes; absolute models are lagging by design. Past performance of price velocity factors is not indicative of future results. All systematic frameworks require rigorous out-of-sample testing before capital deployment.