Momentum#
Momentum strategies exploit the tendency for assets that have recently outperformed to continue outperforming, and for recent laggards to continue lagging. This phenomenon is one of the most robust and well-documented anomalies in finance, persisting across asset classes, geographies, and time periods spanning over a century. Researchers distinguish between time-series momentum — where an asset’s own past return predicts its future return — and cross-sectional momentum, which ranks assets relative to each other. Both forms have been documented in equities, currencies, futures, commodities, and cryptocurrencies.
The papers and posts collected here span the full range of momentum research: foundational academic work establishing the statistical evidence, practitioner-focused studies on crypto and equity market implementation, and applied frameworks for combining momentum with volatility targeting, regime filters, and risk-adjusted ranking. Key topics include dual momentum, lookback period selection, the relationship between momentum and behavioral biases, and the question of whether momentum is truly a risk premium or an exploitable inefficiency.
Related topics include Trend Following, which shares conceptual roots but focuses on time-series signals and longer-horizon systematic programs, Equity Factors for momentum within factor model contexts, and Mean Reversion, which represents the opposite force and is often combined with momentum for balanced portfolio construction.
Momentum and trend following trading strategies for currencies and bitcoin#
This paper tests EMA-based time-series and cross-sectional momentum strategies on currencies and Bitcoin. The authors find that momentum signals generate positive risk-adjusted returns in both asset classes, with Bitcoin exhibiting particularly strong trending behavior. The paper compares various lookback periods and provides practical guidance on implementation for systematic traders.
Momentum trading in cryptocurrencies: short-term returns and diversification benefits#
This paper studies J/K momentum strategies in cryptocurrency markets, examining whether buying recent winners and selling recent losers generates risk-adjusted returns. Using DCC correlation models to analyze portfolio diversification, the authors document positive momentum payoffs in crypto markets and explore how these strategies interact with traditional asset portfolios.
Pure Momentum in Cryptocurrency Markets#
This paper documents a pure momentum anomaly in cryptocurrency markets, attributing it to the behavioral dynamics of a 24/7 trading environment with no circuit breakers. The authors find momentum effects that are distinct from those in equity markets, driven by the unique microstructure and participant composition of crypto markets.
Optimizing the Persistence of Price Momentum: Which Trends Are Your Friends?#
This paper examines the persistence of price momentum across different market segments, finding that style and industry momentum exhibit different decay patterns than traditional cross-sectional momentum. The authors develop methods for identifying which momentum trends are most likely to persist and provide guidance on lookback period selection.
All Days Are Not Created Equal: Understanding Momentum by Learning to Weight Past Returns#
This paper introduces a flexible framework for weighting past returns when constructing momentum signals, challenging the standard equal-weighting of each month in a formation window. The authors learn optimal past-return weights from data, showing that simple modifications to return weighting can significantly improve momentum signal quality and out-of-sample performance.
Beat the Market: An Effective Intraday Momentum Strategy for S&P500 ETF (SPY)#
This paper documents an intraday momentum effect in S&P 500 ETF trading, showing that the first half-hour return predicts the last half-hour return with sufficient consistency to generate a strategy that delivered approximately 1,985% return over the sample period. The strategy exploits persistent intraday autocorrelation that has proven robust over multiple years of testing.
Risk-adjusted Momentum Strategy Construction and Industry Heterogeneity Analysis Based on STARR Indicator#
This paper proposes using the STARR (Stable Tail Adjusted Return Ratio) indicator to construct risk-adjusted momentum strategies that control for downside risk while capturing momentum premia. The authors document industry heterogeneity in momentum returns and show that risk-adjusted construction improves risk-adjusted performance relative to raw return momentum.
A Tale of Two Anomalies: The Implications of Investor Attention for Price and Earnings Momentum#
This paper studies how investor attention amplifies both price momentum and earnings momentum, finding that assets receiving less investor attention tend to exhibit stronger momentum effects. The results connect momentum to the gradual information diffusion hypothesis, where attention constraints slow the market’s incorporation of fundamental information.
Time Series Momentum#
The seminal Moskowitz, Ooi, and Pedersen paper documenting time-series momentum across 58 liquid futures contracts in equities, currencies, commodities, and fixed income. The paper shows that past 12-month returns positively predict the next month’s returns across all asset classes, with a t-statistic exceeding 5.0. This is the foundational reference for all subsequent time-series momentum research.
By Tobias J. Moskowitz, Yao Hua Ooi, and Lasse Heje Pedersen.
Time Series Momentum and Volatility Scaling#
This paper decomposes the returns of time-series momentum strategies to show that volatility scaling — adjusting position sizes inversely to recent volatility — is the primary source of their superior risk-adjusted performance. The authors demonstrate that the constant-volatility version of TSMOM outperforms the raw version, providing theoretical and empirical support for volatility-targeted implementation.
Is Momentum Really Momentum?#
This paper examines the detailed structure of the momentum effect in equities, finding that the 12-7 month lookback window — which skips the most recent month — is the specification that generates the most robust and statistically significant momentum profits. The paper addresses concerns about data mining and provides guidance on which lookback configurations are most reliable.
Time Series Momentum: Is It There?#
A skeptical reexamination of the time-series momentum evidence, testing whether the phenomenon holds up under alternative specifications, out-of-sample periods, and risk adjustment methods. The paper provides a balanced assessment of the evidence and identifies conditions under which TSMOM profits are most reliably present.
Bitcoin Intraday Time-Series Momentum#
This paper documents an intraday time-series momentum effect in Bitcoin, showing that the first half-hour of Bitcoin trading predicts the last half-hour with statistical significance. The effect is analogous to the SPY intraday momentum documented by Gao et al. and suggests that the mechanisms driving intraday autocorrelation operate similarly in crypto markets.
Value and Momentum Everywhere#
AQR’s landmark cross-asset study documenting value and momentum premia across 8 major asset classes including equities, bonds, currencies, and commodities. The paper shows that momentum and value are pervasively positive across all asset classes and that they are negatively correlated with each other, making them natural complements in a diversified systematic portfolio.
By Clifford S. Asness, Tobias J. Moskowitz, and Lasse Heje Pedersen.
Cryptocurrencies and Momentum#
This paper tests classic cross-sectional momentum strategies across a large universe of 143 cryptocurrencies from 2014 to 2018. The authors find that for the broad universe, winner-minus-loser returns are close to zero and generally insignificant, challenging the straightforward application of equity momentum strategies to crypto markets. Some trimmed-sample specifications produce negative rather than positive momentum returns.
Dynamic Time Series Momentum of Cryptocurrencies#
This paper studies time-series momentum across 20 major cryptocurrencies, finding that the TSMOM effect in crypto is stronger and persists over longer horizons than in traditional equity markets. The authors develop a dynamic version of TSMOM that adapts to the unique distributional characteristics of crypto returns.
A Trend Factor for the Cross Section of Cryptocurrency Returns#
This paper develops the CTREND factor — a trend-based cross-sectional factor for cryptocurrency returns — and shows that it explains a significant portion of the cross-section of crypto returns. The factor captures both time-series and cross-sectional momentum components and provides a unified framework for understanding trend-related return predictability in crypto markets.
Nonlinear Time Series Momentum#
This paper applies machine learning methods to capture nonlinear features of time-series momentum signals, documenting that momentum signals exhibit nonlinear dynamics that linear models miss. The authors show that ML-enhanced momentum strategies outperform their linear counterparts, particularly in periods of regime change.
Simple Crypto Momentum Strategy: Buying Corrections#
A straightforward crypto trading approach shared by Pavel demonstrating that simple strategies can work effectively in cryptocurrency markets. The strategy involves: buying corrections on the highest momentum coins, holding positions for one day, and applying a basic regime filter to avoid unfavorable market conditions.
The equity curve shown suggests consistent performance with this systematic approach. The strategy exemplifies how basic momentum and mean-reversion concepts can be combined—buying dips in strong trending assets while using regime filtering to stay out during adverse market conditions.
By Pavel (Robuxio).
A Delta-Neutral Cross-Sectional Momentum Strategy in Crypto#
Research Article #73 from Trading Research Hub. Explores a delta-neutral cross-sectional momentum strategy applied to cryptocurrency markets. Cross-sectional momentum ranks assets against each other based on their relative performance, going long the strongest performers and short the weakest to create a market-neutral portfolio.
The strategy aims to capture the momentum premium in crypto while eliminating directional market exposure. By maintaining delta neutrality, the approach seeks to profit from the spread between winners and losers regardless of overall market direction, making it particularly relevant for volatile crypto markets where directional risk can overwhelm individual alpha signals.
By Pedma.
Systematic Momentum in Crypto: 249% Returns with 82% Less Volatility#
Research Article #61 from Trading Research Hub. Presents a systematic momentum strategy that delivered 249% returns with 82% less volatility than the broader crypto market. The article addresses the core challenge of crypto investing: extreme drawdowns exceeding 90% and volatility above 80% that make buy-and-hold approaches impractical for most investors.
The strategy uses a systematic momentum approach with volatility targeting to deliver steady growth while avoiding the worst of crypto’s drawdown episodes. The article demonstrates how disciplined risk management can capture a significant portion of crypto’s upside while making the return stream suitable for investors who cannot tolerate the asset class’s native risk profile.
By Pedma.
A Counter-Intuitive RSI Strategy Delivering 602% Returns at Half the Drawdown#
Research Article #60 from Trading Research Hub. Presents a systematic RSI-based trading strategy that achieved 602.91% total return while limiting maximum drawdown to -33.22% compared to the market’s -92.58%. The strategy delivered a Sharpe Ratio of 0.9925, significantly outperforming benchmark risk-adjusted metrics.
The approach transforms traditional RSI interpretations by using a methodical, counter-intuitive framework that turns market volatility from a threat into an opportunity. Rather than relying on complex algorithms or high-frequency trading, the strategy demonstrates how systematic application of a well-tested RSI variant can capture meaningful crypto upside while maintaining remarkably lower risk metrics.
By Pedma.
Applying Faber’s Relative Strength Framework to Cryptocurrency Markets#
Research Article #57 from Trading Research Hub. Applies the relative strength investing framework from Mebane Faber’s 2010 paper “Relative Strength Strategies for Investing” to cryptocurrency markets. The study tests whether the momentum-based ranking and allocation rules that have worked in traditional markets translate to the unique dynamics of crypto assets.
By Pedma.
Harvesting Risk Premia Through Dual Momentum: Reviewing Antonacci’s Framework#
Research Article #23 from Trading Research Hub. An in-depth review of Gary Antonacci’s 2012 paper on dual momentum, which combines relative momentum (comparing assets against each other) with absolute momentum (an asset’s own past performance) to harvest risk premia across asset classes.
The article explains how momentum has been documented across equities, currencies, bonds, and other asset classes, while acknowledging that the reasons for its persistence remain debated between rational (risk premia) and behavioral (herding, anchoring) explanations. The dual momentum framework is analyzed for its applicability to cryptocurrency portfolio construction and systematic trading strategies.
By Pedma.
A Simple Monthly Momentum Strategy for US Stocks#
A beginner-friendly systematic trading strategy using monthly momentum on Nasdaq 100 stocks. The approach focuses on robustness over complexity, framing the strategy as premium harvesting of beta rather than alpha mining — fewer moving parts mean fewer things that can break compared to overoptimized, datamined strategies.
The exact rules: trade only when NDX is in an uptrend (Close > MA200), filter the Nasdaq 100 universe for stocks also in uptrend with positive 6-month returns, rank by 6-month return once a month, buy the top 5 equally weighted at 20% each, hold for one month, then repeat. No stop loss, no profit target. The author positions this as a solid core sleeve for a systematic portfolio that can be improved by combining it with uncorrelated strategies like long mean reversion or by adding better filters and ranking criteria.
By Petr Podhajsky.
Gold Momentum Strategy#
A simple dual-momentum approach for timing gold investments. The strategy combines 12-month total return signals from both gold (GLD) and 10-year U.S. Treasuries (IEF) to filter out weak periods and reduce drawdowns.
The trading rules are straightforward: when both gold and Treasury returns are positive at month-end, go long gold; if either is negative, stay in cash. This deliberate simplicity—with minimal parameters—reduces the risk of overfitting. Based on research from Allocate Smartly’s “Gold Cross-Asset Momentum,” historical backtesting over 50 years shows the strategy has generally reduced exposure during losing phases and outperformed a basic gold-only momentum approach.
By QuantifiedStrategies.com.
Backtesting a Community-Sourced Momentum Strategy#
Strategy Backtest #14 from Trading Research Hub. Tests a momentum-based trading strategy through systematic backtesting. Part of an ongoing series designed to make strategy research accessible by collecting backtests of different approaches in one place, saving readers hours of independent research time.
The article tests specific strategy rules from the community, providing honest performance analysis including both the promising and concerning aspects of the results. It emphasizes the importance of working with practitioners who have proven live track records, and the distinction between historical backtest performance and actual trading outcomes.
By Pedma.
Systematic Regime Detection for Momentum Strategy Timing#
An article on the systematic approach to regime targeting for momentum trading strategies. The article argues that trading against the general market direction is one of the most common causes of losses, and presents a framework for identifying favorable market regimes.
The article covers how to define and detect market regimes, how to adjust strategy exposure based on the current regime, and how to avoid trading during unfavorable conditions. The practical framework helps momentum traders align their positions with the broader market trend while knowing when to reduce exposure or “go fishing.”
By Pedma.
NDX Momentum#
A momentum-based rotational strategy focused on Nasdaq-listed stocks. The system ranks stocks by their momentum scores and rotates into the strongest performers while applying a market regime filter to reduce exposure during unfavorable conditions. The post details the ranking methodology, rebalancing frequency, and historical performance characteristics of this systematic approach to capturing momentum in technology-heavy equities.
By Peter, CrackingMarkets.
US Stock Momentum Trading System for Retail Traders#
A comprehensive deep research report on building a US stock momentum trading system suitable for retail traders. The post was generated using ChatGPT Pro’s Deep Research functionality and covers extensive academic and practical sources on momentum strategies. It addresses universe selection, momentum measurement, portfolio construction, rebalancing schedules, and risk management, providing a thorough starting point for retail traders looking to implement systematic momentum strategies.
By Peter, CrackingMarkets.
RealTest Code for SPX Momentum Rotational System#
Complete RealTest code for an SPX momentum rotational system, backtested using Norgate Data with commissions and delisted stocks included. The system checks a market regime filter based on the balance between stocks with positive and negative longer-term momentum, then opens up to 10 positions ranked by momentum. The post provides the full implementation code for site supporters.
By Peter, CrackingMarkets.
Dynamic Asset Allocation for Practitioners, Part 3: Risk-Adjusted Momentum#
A comprehensive study testing 13 different risk-adjusted momentum indicators across 16,116 simulation variations to determine whether accounting for volatility and drawdowns improves momentum-based asset allocation. The metrics tested include Sharpe ratio, Omega ratio, Sortino ratio, Calmar ratio, DVR (Sharpe x R-squared of equity curve), VaR, CVaR, max loss ratio, average drawdown ratio, high-low differential, Ulcer Performance Index, gain-to-pain ratio, and fractal efficiency.
The key finding is that all 13 methods delivered virtually identical median results with an average pairwise correlation of 0.945, suggesting investors inherently exhibit risk awareness when expressing momentum preferences. However, risk-adjusted approaches showed much smaller dispersion than pure price momentum, and aggregating all approaches into a combined index delivered superior risk-adjusted performance with a Sharpe ratio of 1.29 — outperforming any single metric alone. The research demonstrates that diversifying across multiple highly correlated risk-adjusted momentum indicators provides measurable risk-reduction benefits without sacrificing returns.
By ReSolve Asset Management.
Dynamic Asset Allocation for Practitioners Part 2: Risk-Adjusted Momentum#
The second installment in a series on dynamic asset allocation, investigating how 13 risk-adjusted momentum metrics compare to raw momentum for portfolio construction. The research tests across a 10-asset universe spanning commodities, gold, US/European/Japanese/emerging market stocks, international REITs, and treasuries from 1995 onward, using multiple portfolio concentrations (2–5 holdings) and systematic asset removal to minimise selection bias.
The DVR and return-to-max-loss ratio delivered the highest Sharpe ratios, while the Ulcer Performance Index achieved the best return-to-maximum-drawdown performance. A counterintuitive finding: risk-adjusted momentum portfolios exhibited lower Sharpe ratios than raw momentum systems from Part 1, suggesting that disaggregating risk management from momentum signals may produce better results. The gain-to-pain ratio demonstrated the greatest consistency across different asset universes and concentrations.
By Adam Butler, GestaltU.
Risk-Adjusted Momentum Oscillator (RAMO)#
An open-source TradingView indicator that combines traditional momentum calculation with real-time risk assessment, addressing a gap in conventional momentum indicators by incorporating drawdown metrics into momentum calculations. The core innovation is a self-regulating system that automatically adjusts signal sensitivity based on current risk conditions using the formula: Risk_Factor = 1 - (Current_Drawdown / Maximum_Drawdown × Scaling_Factor), with a floor of 0.05 to prevent complete signal suppression.
The indicator supports three momentum calculation modes (rate of change, price momentum, log returns) and employs adaptive EMA smoothing where the alpha parameter adjusts based on volatility percentile — faster response during volatile periods, stability during calm markets. Z-score normalisation clamps values to [-3.5, 3.5] for outlier handling. Additional features include momentum acceleration (second derivative) for early trend change detection, linear regression prediction for leading signals, and volume-based exhaustion detection for identifying potential reversals on declining volume. The risk environment classification (low/medium/high based on drawdown depth) filters long signals during high-risk conditions.
By EdgeTools.
Systematic Allocation in International Equity Regimes#
A research article examining systematic tactical allocation between US equities and international markets (EAFE and Emerging Markets) using momentum-based signals. The study uses a 56-year dataset (1969–2025) to test whether a signals-based framework can systematically time exposure to EAFE equities amid changing macroeconomic conditions.
Two strategies are tested: a pure spread momentum approach using rate-of-change signals across 6, 12, 24, and 36-month lookback periods, and a trend-conditioned approach applying simple moving average filters for allocation signals. The research finds intermediate-term windows (12–36 months) generate optimal risk-adjusted returns by balancing signal responsiveness against noise, and that systematic overlays can generate alpha orthogonal to equity-market betas — particularly during major secular market transitions.
By Cyril Dujava, Quantpedia. Mentioned by Radovan Vojtko in this discussion.
A 46-Year Backtest on a Carhart Momentum Strategy: Why Tiny Edges Compound#
A LinkedIn post reporting a 46-year (1980–2026) backtest of a cross-sectional Carhart momentum strategy run on a 100-stock S&P 500 universe with monthly rebalancing. The strategy produces a 24.7% CAGR versus the S&P 500’s 9.6%, an information ratio of 0.81, and positive returns in 55% of 389 monthly rebalances. The central argument is that a statistically tiny edge — a mean information coefficient (IC) of just 0.013, effectively noise in isolation — compounds into meaningful alpha when applied across many independent bets. Invoking Grinold’s Fundamental Law of Active Management (IR ≈ IC × √Breadth), the author shows that 0.013 × √389 ≈ 0.26 per period scales into a strong information ratio, framing systematic investing as a contest won by breadth and discipline rather than by heroic single forecasts.
The momentum factor that this strategy harvests is the same one Mark Carhart added to the Fama-French three-factor model in 1997, building on the cross-sectional winners-minus-losers anomaly documented by Jegadeesh and Titman (1993). Its strengths are well established: momentum is one of the most robust and pervasive anomalies in finance, persisting across asset classes, geographies, and over a century of data, and — as the post illustrates — it offers high breadth, making it well suited to systematic, diversified implementation. Its weaknesses are equally well documented. The factor lacks a clean theoretical foundation (Carhart treated it as an empirical, atheoretical add-on); it suffers rare but catastrophic crashes (e.g. roughly -73% over three months in 2009) that occur during sharp market rebounds and make returns highly non-normal; it carries high turnover, so transaction costs can erode or even eliminate paper profits; and its premium has decayed sharply, averaging only ~2% annually over 2014–2024 versus ~10% in the 1990s, with crowding via factor ETFs frequently blamed. Practitioner remedies include residual/idiosyncratic momentum, volatility scaling, and dynamic crash-risk management.
The post’s own discussion surfaces the key caveats. The Grinold framework assumes bets are independent, frictionless, and have constant IC — assumptions that rarely hold. One commenter (Alexander Hübbert) notes that consecutive 12-month momentum signals overlap by eleven months, so the 389 monthly rebalances are far from 389 independent bets, which inflates the effective breadth and overstates the implied information ratio; the 100 S&P 500 names are also highly correlated rather than orthogonal. Other commenters question multi-factor strategies’ recent real-world performance and raise survivorship-bias concerns about a backtest restricted to current/historical large-cap constituents. Together these points are a useful reminder that a backtested IR of 0.81 on overlapping, correlated, frictionless bets is an optimistic upper bound rather than a deliverable live result.
By Gilberto Pellerano. Shared in this LinkedIn discussion.
Demystifying Time-Series Momentum Strategies: Volatility Estimators, Trading Rules and Pairwise Correlations#
Baltas and Kosowski (Imperial College / Oxford-Man Institute, 2019) study three methodological levers in the standard time-series momentum (TSMOM) strategy used by managed futures funds — the volatility estimator that scales positions, the trading rule that detects price trends, and the treatment of pairwise correlations across portfolio constituents — and show that each can materially improve net-of-cost performance over the naive Moskowitz–Ooi–Pedersen (2012) and Hurst–Ooi–Pedersen (2013) baseline. The paper is a methodological follow-up that complements “Demystifying Managed Futures”: instead of asking “what factor drives CTA returns?”, it asks “how should you actually build a TSMOM strategy if you have to pay transaction costs and your assets are correlated?”
The first two contributions are turnover reductions. Replacing the standard close-to-close volatility estimator with the Yang–Zhang (2000) range-based estimator (which uses open-high-low-close prices) reduces strategy turnover by 17% (t-stat 5.33) without statistically significant performance degradation. Replacing the binary sign trading rule with a continuous trading signal derived from the t-statistic of past daily log-returns over 12 months further reduces turnover by 24% (t-stat 8.49). Combined, these two changes cut turnover by 36% (t-stat 13.19) while leaving Sharpe ratios statistically indistinguishable from the standard TSMOM portfolio. The third contribution is a “correlation-adjusted” TSMOM variant that dynamically scales overall portfolio leverage based on the contemporaneous average pairwise signed correlation of constituents, reducing leverage in high-correlation regimes (and raising it in low-correlation regimes). This adjustment safeguards against downside risk and is most valuable in the post-2008 period, when futures-market correlations rose sharply.
Our summary: this is the natural sequel to Hurst-Ooi-Pedersen and the methodological reference for anyone implementing TSMOM. The paper is most useful for three reasons. First, it shows the Sharpe-1.8 gross headline number of the “Demystifying Managed Futures” paper is fragile to standard implementation choices and that careful volatility estimation and trading-rule design recover most of the gap that transaction costs open up. Second, it provides one of the few formal frameworks for handling the rising-correlation problem that has plagued CTAs since 2008 — relevant to the diagnosis offered in Babu et al. (2020) “You Can’t Always Trend When You Want”. Third, it introduces a transaction-cost model for futures-based strategies that separately accounts for roll-over and rebalancing costs, which has since become a standard reference for CTA cost modelling.
Data: 56 futures contracts across four asset classes — 25 commodities, 14 developed-country equity indices, 8 currency pairs, and 9 government bond futures — over December 1974 to February 2013, with the cross-section growing over time. No public source code is released; the volatility estimator and trading rule are fully specified in the paper and Yang–Zhang (2000) is well documented in the literature. Implementation is reproducible given an equivalent futures dataset, though specific results depend on continuous-futures roll methodology, which is not exhaustively documented.
Key metrics: gross annualised performance of the baseline TSMOM strategy is 16.12% over the sample, with transaction costs representing about 10% of that figure. The improved specification reduces transaction costs to roughly 105 basis points (from a higher baseline) without a statistically significant fall in Sharpe ratio. The combined Yang–Zhang + continuous-signal variant cuts turnover by 36.23% (t-stat 13.19); the correlation-adjusted variant outperforms the naive implementation, with the outperformance most pronounced in the 2009–2013 post-crisis sub-period when the default specification exhibits a negative after-cost Sharpe ratio. 49 of the 56 futures contracts have a positive unconditional mean TSMOM excess return.
By Nick Baltas and Robert Kosowski.
Momentum Factor Investing: Evidence and Evolution#
Baltussen, Dom, Van Vliet, and Vidojevic (Erasmus School of Economics / Northern Trust Asset Management, October 2025) present a comprehensive review and empirical analysis of the momentum factor across domestic and global stock markets spanning up to 159 years of data (1866-2024). The paper establishes momentum as an “eternal feature of financial markets” by demonstrating its robustness across multiple sample periods, 46 countries, and over 4,000 different portfolio construction specifications. Beyond traditional price momentum, the authors document that momentum is fundamentally multi-dimensional — present in firm fundamentals (earnings surprises, analyst revisions, news sentiment), firm-specific residual returns, anchor-based measures (52-week high), industry and network peer returns, and factor returns themselves.
The paper’s central contribution is showing that combining these diverse momentum signals into a multi-dimensional composite produces superior risk-adjusted returns that cannot be explained by traditional price momentum alone. The authors also address crash risk — momentum’s primary weakness — demonstrating that risk-managed approaches using volatility scaling substantially reduce maximum drawdowns while preserving or enhancing returns. On the question of post-publication decay, the paper finds no statistically significant evidence that momentum profits have diminished since the original Jegadeesh and Titman (1993) publication, further supported by 61 years of pre-CRSP out-of-sample data where arbitrage was impossible. The behavioral explanation — investor underreaction and gradual information diffusion — is favored over risk-based explanations.
Data uses CRSP (1927-2024), a pre-CRSP hand-collected dataset (1866-1926), and JKP international data covering 46 countries (1990-2024). No source code is provided, but the methodology is fully specified and reproducible with standard academic datasets.
Key metrics for the winner-minus-loser long-short portfolio: annualised return of 9.24% VW / 9.75% EW over the full CRSP sample (1927-2024) with t-statistics of 4.31 and 4.66. Over the deepest 159-year sample (1866-2024), returns average 8.83% VW / 8.95% EW with t-statistics around 5. The median Sharpe ratio across 4,096 portfolio design choices equals 0.61 (range 0.38-0.94). Global momentum premium: 7.77% (world), 7.57% (ex-US), 6.20% (emerging markets), positive in all 31 countries with continuous data. Maximum drawdown for standard price momentum is -88.41% VW. Risk-managed momentum (RM_MOM) increases annualised return to 17.87% while reducing max drawdown to -49.37%. The multi-dimensional EW_ALL composite achieves 9.65% EW annualised with a t-statistic of 8.22 and lower volatility (11.62%) versus standard price momentum (20.71%).
Mentioned by Darren (@ReformedTrader) in this discussion.
Real-Time Profitability of Published Anomalies: An Out-of-Sample Test#
Jing-Zhi Huang and Zhijian Huang test whether published stock market anomalies remain profitable out-of-sample in real time. The paper constructs a long-only strategy that each year recursively selects the best past-performing anomaly from a pool of published anomalies over a given training period — effectively applying momentum to anomalies themselves.
The key finding is that this recursive best-performer strategy outperforms the equity market even after transaction costs, suggesting that published anomalies exhibit performance persistence. The result is relevant to meta-strategy research: if individual anomalies show momentum, then dynamically rotating capital toward recently profitable strategies has an empirical basis. The paper also addresses data-snooping bias, a critical concern when selecting among many anomalies. Published in the Quarterly Journal of Finance, 2014.
Time-Series and Cross-Sectional Momentum in Anomaly Returns#
Wang, Yan, and Zheng document strong evidence of both time-series and cross-sectional momentum in the long-short returns of a comprehensive sample of stock market anomalies. Strategies that exploit this persistence — going long recently profitable anomalies and short recently unprofitable ones — deliver significant abnormal returns that are robust to the standard stock-level momentum effect and cannot be explained by traditional asset-pricing models.
The momentum in anomaly returns is more pronounced when arbitrage capital is scarcer or market liquidity is lower, consistent with limits-to-arbitrage explanations. This paper provides the direct empirical foundation for the meta-strategy approach of rotating across trading strategies based on recent performance. Published in European Financial Management, Volume 27, 2021.
Factor Momentum and the Momentum Factor#
Ehsani and Linnainmaa show that momentum in individual stock returns emanates from momentum in factor returns. Most factors are positively autocorrelated: the average factor earns a monthly return of 6 basis points following a year of losses and 51 basis points following a positive year. A factor momentum strategy that goes long recent winner factors and short recent loser factors is highly profitable and subsumes the standard stock momentum factor.
The central implication is that momentum may not be a distinct risk factor — it times other factors. This reframes the meta-strategy question: allocating to recently winning strategies may work because the underlying factors those strategies capture are themselves persistent. Published in the Journal of Finance, 2022.
Two Centuries of Multi-Asset Momentum#
Geczy and Samonov extend momentum tests to the longest available histories of global financial asset returns — 215 years (1800–2014) — covering equities, bonds, currencies, commodities, sectors, and individual stocks. The paper confirms the significance of the momentum premium both within and across asset classes over this extended sample, making it one of the most comprehensive out-of-sample validations of momentum as a pervasive market phenomenon.
The breadth of the evidence — spanning multiple countries, asset classes, and two centuries of data — strengthens the case that momentum is a robust empirical regularity rather than a sample-specific artifact. The multi-asset perspective is directly relevant to meta-strategy construction, where the question is whether momentum operates at the strategy level just as it does across individual securities and asset classes.
Is There Momentum in Factor Premia? Evidence from International Equity Markets#
Zaremba and Shemer examine whether momentum exists in the returns of factor premia — market, value, size, momentum, low-risk, and quality — across 24 international equity markets from 1990 to 2016. They find strong evidence that top-performing factors continue to outperform the worst-performing factors, both within individual equity markets and in a cross-country framework.
However, the paper also shows that momentum in factor premia is largely explained by the classic stock-level momentum effect, raising questions about whether factor momentum is a distinct phenomenon or a manifestation of the same underlying force. This is a cautionary result for meta-strategy builders who assume factor-level momentum provides independent diversification. Published in Research in International Business and Finance, Volume 46, 2018.
Factor Performance 2010-2019: A Lost Decade?#
Blitz examines whether the 2010–2019 decade represents a structural break in factor investing, documenting that value, size, and low-volatility factors all underperformed during this period while momentum remained profitable but with elevated crash risk. The paper investigates potential explanations including crowding, changes in market structure, and the prolonged low-interest-rate environment.
The practical implications are significant for meta-strategy builders: factor performance exhibits extended regime dependence, and a momentum overlay that rotates into recently winning factors would have concentrated into the few factors that worked (momentum, quality) while avoiding the many that didn’t. The paper serves as a cautionary case study on the risks of performance chasing at the factor level and the importance of understanding whether decay is temporary or permanent. Published in the Journal of Portfolio Management, 2020.