Time-Series Momentum
Time-series momentum (TSMOM) is the empirical finding that an asset's own recent return predicts its future return in the same direction—assets that have risen tend to continue rising, and assets that have declined tend to continue declining, over horizons of 1 to 12 months. It is the foundational signal of trend-following strategies and is distinct from cross-sectional momentum, which ranks assets relative to each other.
Key takeaways
- Time-series momentum is defined by the sign and magnitude of an asset's return over a lookback period (typically 1–12 months); a positive return predicts a long position, and a negative return predicts a short position.
- TSMOM has been documented across equities, bonds, commodities, and currencies, with the strongest signals typically observed at the 6–12 month lookback horizon and with weaker evidence at very short (1-month) and very long (60-month) horizons.
- Moskowitz, Ooi, and Pedersen (2012) documented TSMOM in 58 liquid futures markets from 1985–2009, finding positive returns in all major asset classes and significant diversification benefits—particularly crisis-alpha during equity bear markets.
- The risk-adjusted returns from TSMOM are eroded by trading costs, execution slippage, and the transaction costs of high turnover, making practical implementation (choice of lookback, rebalancing frequency, position sizing) critical to achieving the gross documented performance in net terms.
- The behavioral explanation for TSMOM involves initial underreaction to news (causing trends to develop) followed by overreaction and eventual reversal, consistent with investor herding and anchoring biases.
Explanation
Time-series momentum is one of the most robust return anomalies in finance, having been documented in out-of-sample data across multiple asset classes, geographies, and time periods spanning over a century. Its core empirical finding, as established by Moskowitz, Ooi, and Pedersen's landmark 2012 paper, is that instruments with positive trailing 12-month returns continue to outperform over the subsequent month, and instruments with negative trailing returns continue to underperform, with statistical significance that survives multiple testing adjustments and transaction cost estimates. This persistence of trends, over the 1–12 month horizon, contradicts the weak form of the Efficient Market Hypothesis (which implies that past returns should not predict future returns) and has attracted substantial theoretical attention.
The implementation of a TSMOM strategy is conceptually straightforward. Define a lookback period L (typically 12 months, excluding the most recent month to avoid the 1-month reversal effect). For each instrument in the universe (equity index futures, bond futures, commodity futures, currency forwards), compute the sign of the L-month return. Go long instruments with positive L-month returns and short instruments with negative L-month returns, sized according to a position sizing rule (typically inverse volatility weighting to equalize risk contributions across assets). Rebalance monthly as returns are updated. The resulting strategy has historically generated average annual returns of 10–15% with Sharpe ratios of 0.7–1.2 before fees, though these metrics are sensitive to lookback horizon, rebalancing frequency, and execution quality.
The performance characteristics of TSMOM are particularly attractive from a portfolio diversification perspective. TSMOM exhibits positive skewness (large gains are more common than large losses, contrary to most equity strategies), positive convexity with respect to market extremes (the strategy tends to perform best during extended bull and bear markets when trends are strongest), and low or negative correlation with equity markets during major equity bear markets. This 'crisis alpha' property makes TSMOM a valued diversifier for institutional portfolios dominated by equity risk. The 2022 environment—characterized by sustained trends in rising interest rates, falling equities, and surging commodities—was particularly favorable for TSMOM strategies, with leading managed futures funds returning 25–40% while the 60/40 portfolio declined 15–20%.
Theoretical explanations for TSMOM have evolved across behavioral and institutional frameworks. The behavioral view posits that investors initially underreact to positive information (anchoring to prior prices, gradual diffusion of information across market participants), causing trends to develop as the market slowly incorporates new information; the trend eventually overshoots (overreaction due to herding and extrapolation bias) before reversing at long horizons (18–60 months). The institutional view emphasizes momentum-generating mechanisms: trend-following funds themselves amplify price trends as they add to winning positions and exit losing positions; hedging activities by commercial participants create systematic order flows that sustain trends; and risk management models using backward-looking volatility estimates create momentum in realized returns as high-volatility (often trending) environments cause position cuts that exacerbate moves.
The two main implementation risks for TSMOM strategies are whipsaws and crowding. Whipsaws occur when markets reverse sharply before a trend has been confirmed, generating losses on both the entry and exit of positions that momentarily appeared profitable. Longer lookback periods reduce whipsaw frequency but increase the delay in adapting to trend reversals. Crowding occurs when too many trend-following managers hold the same positions simultaneously; in a crowded trend, the reversal risk is amplified as multiple managers simultaneously exit, creating overshooting in the opposite direction. The August 2007 'quant quake,' while primarily affecting statistical arbitrage strategies, illustrated how correlated position unwinds can create extreme short-term volatility in strategies with similar signal structures.
Formula
TSMOM Signal_t = sign(R_{t-L,t-1}); Position_i = signal_i / σ_i × (target vol / N)
Example
A systematic CTA implements a 12-month time-series momentum strategy across 50 futures markets (25 financial, 25 commodity). In January 2022, the 12-month return signal is computed for each market: 10-year US Treasury futures have a trailing 12-month return of −5% (negative signal: go short), S&P 500 futures are flat (no clear signal), WTI crude oil futures have a trailing 12-month return of +55% (positive signal: go long), and the Euro/USD currency forward is down −5% (negative signal: go short EUR, long USD). Positions are sized at 1% portfolio volatility each (using 60-day realized volatility). The resulting portfolio is short rates, long energy, and short EUR. As these trends accelerated through 2022—rates rose sharply, oil surged to $120 before pulling back, and EUR continued its decline—the strategy captured sustained trend returns across all three positions, generating approximately +30% gross return for the year.
Related terms
Alpha Arbitrage Bond Cointegration Convexity Correlation Cross Sectional Momentum Diversification Efficient Market Hypothesis Equity Equity Index Hedging