The Enduring Effect Of Time Series Momentum On Stock Returns Over Nearly 100-Years
New York University – Leonard N. Stern School of Business
David Einhorn’s Greenlight had a strong fourth quarter; Gains on Neubase Therapeutics [Full Q4 Letter]
David Einhorn's Greenlight Capital was up 5.2% in 2020, underperforming the S&P 500's 18.4% return. For the fourth quarter, the fund was up 25%, which was its best quarterly result ever. Longs contributed 42% during the fourth quarter, while shorts detracted 15% and macro detracted 1%. Q4 2020 hedge fund letters, conferences and more Growth Read More
Manchester Business School; New York University – Stern School of Business
Lancaster University – Management School; New York University (NYU) – Leonard N. Stern School of Business
January 22, 2016
This study documents the significant profitability of “time series momentum” strategies in individual stocks in the US markets from 1927 to 2014 and in international markets since 1975. Unlike cross-sectional momentum, time-series stock momentum performs well following both up- and down-market states, and it does not suffer from January losses and market crashes. An easily formed dual-momentum strategy, combining time-series and cross-sectional momentum, generates striking returns of 1.88% per month. We test both risk based and behavioral models for the existence and durability of time series momentum and suggest the latter offers unique insights into its continuing factor dominance.
The Enduring Effect Of Time Series Momentum On Stock Returns Over Nearly 100-Years – Introduction
Time series momentum is perhaps the most observable form of any asset return factor as it can be visually detected by any investor – smart money or dumb money, value or growth orientated, quantitative algorithm or human stock picker. An investor can look at a price graph or table and detect directional price movement – if time series momentum exists, then if that prior and current asset price goes up it should tend to continue to go up in future (trend) and vice versa. This trend requires “memory” beyond the traditional academic one-time period of Markov models and the normal distribution of independent identical periods.
Existing studies primarily focus on time series momentum across asset classes (Moskowitz, Ooi, and Pedersen, 2012; Baltas and Kosowski, 2013), its relation with volatility states (Petterson, 2014), and how it could be optimally implemented by managed-futures hedge funds, commodity trading advisors and certain macro traders (Hurst, Ooi, and Pedersen, 2013, 2014; Levine and Pedersen, 2015; Baltas and Kosowski, 2015). Moskowitz, Ooi, and Pedersen (2012) were among the first to document the time series momentum effect in the future market. They show that prior-year returns of a futures contract is a positive predictor of its future return for the next year, and that the strategy of financing the acquisition of up-trend futures by selling those down-trend futures generates substantially abnormal returns. Yet, the academic literature has devoted surprisingly little attention to the most conventional asset class. We document a strongpresence of time series momentum in individual stocks over a significant time period without any significant reduction in its factor persistence. We also uncover its unique characteristics and features, which supplement the properties of well-known cross-sectional momentum.
This paper assesses the durable strength of a specific type of time series momentum by creating a rudimentary trading strategy around stocks. We then compare the strategy returns relative to market returns to determine its effectiveness. Our paper is in contrast to other momentum literature which has more complex return factor combinations (e.g. how do collective investors define value or cheapness without including biases and combine it momentum?) and utilize many moments of an asset’s return distribution. We test both risk based and behavioral explanations for the existence and durability of time series momentum and suggest the latter offers unique insights into its continuing factor dominance even after the literature has grown.
Specifically, we show that time-series stock momentum strategies produced significant profits in the US markets throughout the 88-year period from 1927 to 2014 exceeding the returns from other return factors such as value and size. A strategy of going long on stocks with positive returns in the prior year and going short on stocks with negative returns during the same period yields the average monthly return of 0.55% (t-statistic = 5.28) for value weighting and 0.58% (t-statistic = 5.05) for equal weighting. Further, time series momentum also prevails in the international stock markets. We find a strong presence of time series momentum in 10 out of the 13 major international stock markets examined, including those in Austria, Canada, Denmark, France, Germany, Italy, Netherlands, Norway, Switzerland, and the United Kingdom. For instance, the same strategy described generated the value-weighted monthly return of 1.15% per month (t-statistic = 5.06) in Denmark over the 1975?2014 sample period.
We find that time-series stock momentum is profitable regardless of formation and holding periods for 16 different combinations. Even when we use market-adjusted excess returns instead of raw returns and a different weighting system (e.g., inverse volatility weighting) to form long-short portfolios, time-series stock momentum profits remain statistically and economically significant. The robustness of time-series stock momentum in the global equity market is a contradiction to the conventional wisdom of the random walk theory, which predicts that a stock’s past price movement or direction cannot be used to predict its future movement.
Time-series stock momentum exhibits three unique characteristics. Firstly, our regression analyses shows that time-series stock momentum can fully subsume cross-sectional stock momentum, while cross-sectional stock momentum cannot capture time-series stock momentum. Secondly, cross-sectional momentum is existent in up markets only (Cooper Gutierrez and Hameed, 2004), while time series momentum is present following both up and down markets. Specifically, time series momentum produces an average monthly raw return of 0.57% (t-statistic = 2.09) following down market states and 0.54% (t-statistic = 5.30) following up market states. Thirdly, cross-sectional momentum has a seasonal component, being profitable in all months except in January but suffering considerable losses in January (Jegadeesh and Titman, 1993; Grundy and Martin, 2001; Yao, 2012), whereas time-series momentum does not experience significant losses in January.
See full PDF below.