Time-series Momentum Strategies Research Articles

시계열 모멘텀과 기술적 분석에 관한 연구

This study comprehensively analyzes time series momentum and moving average strategies for the Korean stock market. The results of the empirical analysis are as follows. The correlation coefficients between the returns of time series momentum and moving average strategies are on average above 0.8, indicating that the two strategies are closely related. When calculating risk-adjusted returns and abnormal returns, Sharpe ratios and Jensen alphas were generally higher for the moving average strategy compared to the time series momentum strategy. Decomposing the difference in returns between time-series momentum and moving average strategies, we found that moving average strategies tend to run earlier and time-series momentum later, with earlier strategies generating positive returns and later strategies generating negative returns. When analyzing stock price crash risk, we found that unlike cross-sectional momentum, neither time series momentum nor moving average strategies were sensitive to stock price crash risk.

A STUDY OF FRACTAL DUAL MOMENTUM INVESTMENT STRATEGY UNDER THE CONSTRAINT OF MULTI-FRACTAL CHARACTERISTICS OF STOCK MARKET

Since the discovery of momentum effect, people have started the journey of using the momentum effect to construct momentum strategies. As a result of coupling cross-sectional and time-series momentum strategy, dual momentum strategy (DM strategy) has been widely used in practice and closely followed by academics. To address the shortcoming of the classical DM strategy that has not considered the multi-fractal characteristics of the stock market, we construct the fractal dual momentum strategy (FDM strategy) from the two aspects of optimizing the ranking index of the cross-sectional momentum strategy by using fractal statistical measures and improving the timing selection of the time-series momentum strategy by using the trend entropy dimension. The empirical results show that the FDM strategy outperforms the DM strategy. Both in terms of the size and stability of the strategy returns, the FDM strategy shows an optimization effect compared with the DM strategy, which is beneficial to provide investors with better decision-making references.

Performance of Time-series Momentum Strategy: US Evidence

This paper examines the effectiveness of the time series momentum strategy in generating positive returns in the US stock market, with a focus on exploring its dynamics and performance using different moving average methods. The author conducted an empirical analysis of the time series momentum strategy using S&P500 data from 2000 to 2022. A regression model was applied to estimate the expected returns and volatility of each as-set, and then an evaluation of momentum trading strategy based on different moving average methods was developed. The author evaluates the performance of the strategy with and without transaction costs. The study contributes to the literature by providing empirical evidence on the effectiveness of the time series momentum strategy in the US stock market and by exploring the performance of different moving average methods on the strategy. The findings of this study can provide insights for investors and portfolio managers interested in implementing momentum strategies in their investment portfolios.

Time-Series Momentum in a Small European Stock Market: Evidence from a New Historical Financial Dataset

In this paper, we examine the Portuguese stock market for indication of time-series momentum effects using a new historical financial dataset that covers about 120 years of data. We find strong time-series momentum effects that cannot be explained by conventional risk factors. The positive return continuation seems to last for a period of 12 months, being heavily concentrated at the first month. At longer investment horizons, returns tend to mean-revert. The market exhibited significant time-series momentum for all look-back and holding periods of 12 months or less. A strategy with a 1-month look-back period and a 12-month holding period is shown to be the most profitable yielding a Sharpe ratio roughly 5.4 times that generated by a passive strategy. Time-series momentum strategies tend to perform best during extreme up-market periods and deliver the worst returns during down markets. This suggests that the strategy may not offer significant diversification benefits. Our findings add to the evidence that time-series momentum effects are not a product of data mining and are difficult to reconcile with the assertion that stock markets follow a random walk.

Cross-asset time-series momentum: Crude oil volatility and global stock markets

We examine the profitability of a cross-asset time-series momentum strategy (XTSMOM) constructed using past changes in crude oil–implied volatility (OVX) and stock market returns as joint predictors. We show that employing the past changes in OVX in addition to past stock returns helps better predict future stock market returns globally. The XTSMOM outperforms the single-asset time-series momentum (TSMOM) and buy & hold strategies with higher mean returns, lower standard deviations, and higher Sharpe ratios. The XTSMOM can also forecast economic cycles. We contribute to the literature on cross-asset momentum spillovers as well as on the impacts of crude oil uncertainty on stock markets.

Slow Momentum with Fast Reversion: A Trading Strategy Using Deep Learning and Changepoint Detection

Momentum strategies are an important part of alternative investments and are at the heart of the work of commodity trading advisors. These strategies have, however, been found to have difficulties adjusting to rapid changes in market conditions, such as during the 2020 market crash. In particular, immediately after momentum turning points, when a trend reverses from an uptrend (downtrend) to a downtrend (uptrend), time-series momentum strategies are prone to making bad bets. To improve the responsiveness to regime change, the authors introduce a novel approach, in which they insert an online changepoint detection (CPD) module into a deep momentum network pipeline, which uses a long short-term memory deep-learning architecture to simultaneously learn both trend estimation and position sizing. Furthermore, their model is able to optimize the way in which it balances (1) a slow momentum strategy that exploits persisting trends but does not overreact to localized price moves and (2) a fast mean-reversion strategy regime by quickly flipping its position and then swapping back again to exploit localized price moves. The CPD module outputs a changepoint location and severity score, allowing the model to learn to respond to varying degrees of disequilibrium, or smaller and more localized changepoints, in a data-driven manner. The authors back test their model over the period 1995–2020, and the addition of the CPD module leads to a 33% improvement in the Sharpe ratio. The module is especially beneficial in periods of significant nonstationarity; in particular, over the most recent years tested (2015–2020), the performance boost is approximately 66%. This is especially interesting because traditional momentum strategies underperformed in this period.

Trending Fast and Slow

This article develops a methodology to combine fast and slow time-series momentum signals using machine learning techniques based on market volatility. Starting with the US equity market, the authors find that the performance of a time-series momentum strategy is determined by both its responsiveness and the market volatility regime, among other factors. A decision tree gives a simple and insightful way to determine the threshold in characterizing low- and high-volatility regimes. A slow time-series momentum strategy tends to outperform a fast time-series momentum strategy when market volatility is low. The opposite tends to occur when volatility is high. This pattern of relative performance can be attributed to market-timing alpha and exists in most global equity markets, including both developed and emerging markets.

Portfolio rebalancing based on time series momentum and downside risk

AbstractTo examine the familiar tradeoff between risk and return in financial investments, we use a rolling two-stage stochastic program to compare mean-risk optimization models with time series momentum strategies. In a backtest of allocating investment between a market index and a risk-free asset, we generate scenarios of future return according to a momentum-based stochastic process model. A new hybrid approach, time series momentum strategy controlling downside risk (TSMDR), frequently dominates traditional approaches by generating trading signals according to a modified momentum measure while setting the risky asset position to control the conditional value-at-risk (CVaR) of return. For insight into the outperformance of TSMDR, we decompose each strategy into two aspects, the trading signal and the asset allocation model that determines the risky asset position. We find that 1) weighted moving average can better capture the trend of the stock market than time series momentum computed as past 12-month excess return, 2) mean-risk strategies generally provide better returns whereas risk parity strategies have less investment risk and 3) controlling CVaR limits the investment risk better than controlling variance does.

Asymmetry, tail risk and time series momentum

In this paper, we investigate how to improve the time series momentum strategy by using partial moments. We find that reversals of time series momentum can be partly predicted by tail-distributed upper and lower partial moments derived from daily returns of commodity futures. Based on such information, we propose rule-based approaches to improve the trading signals suggested by the time series momentum strategy. The empirical results based on Chinese commodity futures document statistically significant improvements of the Sharpe ratio in the out-of-sample period. These improvements are robust to different look-back windows.

Time-series residual momentum strategies

ABSTRACT We find that the conventional time-series momentum strategy can be revised in two ways to deliver enhanced excess return performance that are reminiscent of the advancements for the cross-sectional momentum strategy. First, we explore the applicability of the residual return for time-series momentum and find that it opens a door for constructing new improved time-series momentum strategies. Second, we examine the pertinence of the auto-covariance property of returns, which is the defining characteristic of time-series momentum, for asset allocation which has almost universally been identified with the equal-weighted one in the literature. We discover that consistent and persistent performance enhancements are earned by both approaches for the industry portfolios.

TIME-SERIES AND CROSS-SECTIONAL MOMENTUM IN INDIAN STOCK MARKET

Present study documents the significant time-series and cross-sectional momentum profits in Indian stock market. These profits remain significant even after adjusting market, size and value factors. Further time-series momentum effect remains significant when we hold securities for longer period signalling that time-series momentum do not reverse in the long run. When we compare the performances of time series and cross-sectional momentum payoffs, we find that time-series momentum strategies generate superior returns than cross-sectional momentum strategies and net long investments in time-series momentum strategies is the main source of difference between the performances of these two approaches.

Return signal momentum

A new type of momentum based on the signs of past returns is introduced. This momentum is driven primarily by sign dependence, which is positively related to average return and negatively related to return volatility. An empirical application using a universe of commodity and financial futures offers supporting evidence for the existence of such momentum. Investment strategies based on return signal momentum result in higher returns and Sharpe ratios and lower drawdown relative to time series momentum and other benchmark strategies. Overall, return signal momentum can benefit investors as an effective strategy for speculation and hedging.

The value-at-risk of time-series momentum and contrarian trading strategies

The value-at-risk of time-series momentum and contrarian trading strategies

A market scoring mechanism for trading of German electricity futures

We present a novel systematic commodity trading model utilizing a time series momentum strategy. The main innovation is a scoring mechanism to generate buy and sell signals, including determining the position size. The model is applied to the German electricity futures market and its price drivers: the momentum and volatility of the German front quarter, the Argus/McCloskey’s Coal Price Index (API 2) coal and emissions. We test the systematic model on data from November 2010 to December 2019 for several strategies. The model outperforms a basic short sale model. The application of the model yields improved risk-adjusted returns. This paper provides, to the best of our knowledge, the first description of a systematic trading approach to German electricity futures.

Cross-asset time-series momentum: Crude oil options and global stock markets

We examine the profitability of a cross-asset time-series momentum strategy (XTSMOM) constructed using past crude oil options and stock market returns as joint predictors. We show that past crude oil options straddle returns negatively predict while past stock returns positively predict future stock market returns globally. The XTSMOM outperforms the single-asset time-series momentum (TSMOM) and buy-and-hold strategies with higher mean returns, lower standard deviations, and higher Sharpe ratios. The XTSMOM is also able to forecast economic cycles. We contribute to the literature on cross-asset momentum spillovers as well as on the impacts of crude oil uncertainty on stock markets.

Short-Term Trend: A Jewel Hidden in Daily Returns

This article examines the performance of time-series momentum strategies using daily returns for 78 futures markets across four major asset classes between January 1985 and December 2017. The authors find that the 252-day, 63-day, and 21-day momentum strategies perform similarly to the previously documented 12-month, 3-month, and 1-month momentum strategies, respectively. The performance is stronger with volatility-based position sizing, robust to implementation considerations such as a one-day gap between signal generation and execution, and persistent across asset classes and subperiods. The authors introduce a shorter duration momentum strategy with a weekly rebalancing frequency, which cannot be replicated using monthly returns. The authors find that the short-term strategy is a strong diversifier to the longer-term strategies, but the benefit may be reduced, or even completely offset, if the quality of trade execution is poor. The authors also find that the positive contribution of short-term momentum is driven by its superior diversifying characteristics rather than by the rebalancing frequency effect. TOPICS:Factor-based models, performance measurement, portfolio construction, style investing Key Findings • The authors examine the performance of time-series momentum using daily rather than monthly returns with standard lookback periods of 1, 3, and 12 months and a rebalancing period of one month. • They introduce a shorter duration momentum strategy with weekly rebalancing frequency. • The authors show that the short-term momentum strategy is a strong diversifier to the longer-term strategies but that the benefit is heavily dependent on the quality of trade execution.

A Time Series Momentum Hybrid That Controls Downside Risk

To examine the familiar tradeoff between risk and return in financial investments, we use a rolling two-stage stochastic program to compare mean-risk optimization models with time series momentum strategies. In a backtest of allocating investment between a market index and a risk-free asset, we generate scenarios of future return according to a momentum-based stochastic process model. A new hybrid approach, time series momentum strategy controlling downside risk (TSMDR), frequently dominates traditional approaches by generating trading signals according to a modified momentum measure while setting the risky asset position to control the CVaR of return. For insight into the outperformance of TSMDR, we decompose each strategy into two aspects, the trading signal and the asset allocation model that determines the risky asset position. We find that 1) weighted moving average can better capture the trend of the stock market than time series momentum computed as past 12-month excess return, 2) mean-risk strategies generally provide better returns whereas risk parity strategies have less investment risk, and 3) controlling CVaR limits the investment risk better than controlling variance does.

Interest Rate Momentum Everywhere Across Global Yield Curves

This paper explores time series momentum in interest rates across developed and emerging market countries and various interest rate maturities. With a one-year lookback window, almost all countries in our sample of G10 developed countries and 18 emerging market countries strategies have statistically significant positive time series momentum strategy returns. Shorter tenor (2-year) interest rate swaps have greater momentum returns compared to longer tenor (5-year and 10-year) interest rate swaps which arguably is a result of investor underreaction to monetary policy cycles. A significantly greater share of the positive momentum returns across all tenors comes from falling rate environments versus rising rate environments as a result of the secular decline in interest rates in recent decades. This suggests that if low interest rates continue to persist, future interest rate time series momentum strategy returns could be lower.

Improving market timing of time series momentum in the Chinese stock market

ABSTRACT This paper is the first study to present firm-level evidence that the time-series momentum (TSMOM) strategies with look-back-period k of 10 to 200 days outperform the buy-and-hold strategy (BH) on individual stocks in the Chinese stock market. We document that the optimal k* generating the best performance is different across assets and varies over time. We hence propose a model to predict the asset-specific and time-dependent k*, and examine the performance of the TSMOM strategies with the predicted k*. Our analysis shows that using the time-varying predicted k* substantially improves the predictability of the TSMOM strategies. Our new model and findings shed the light on trading strategy for both academia and applied investment practitioners.

Deep Reinforcement Learning for Trading

In this article, the authors adopt deep reinforcement learning algorithms to design trading strategies for continuous futures contracts. Both discrete and continuous action spaces are considered, and volatility scaling is incorporated to create reward functions that scale trade positions based on market volatility. They test their algorithms on 50 very liquid futures contracts from 2011 to 2019 and investigate how performance varies across different asset classes, including commodities, equity indexes, fixed income, and foreign exchange markets. They compare their algorithms against classical time-series momentum strategies and show that their method outperforms such baseline models, delivering positive profits despite heavy transaction costs. The experiments show that the proposed algorithms can follow large market trends without changing positions and can also scale down, or hold, through consolidation periods. TOPICS:Futures and forward contracts, exchanges/markets/clearinghouses, statistical methods, simulations Key Findings • In this article, the authors introduce reinforcement learning algorithms to design trading strategies for futures contracts. They investigate both discrete and continuous action spaces and improve reward functions by using volatility scaling to scale trade positions based on market volatility. • The authors discuss the connection between modern portfolio theory and the reinforcement learning reward hypothesis and show that they are equivalent if a linear utility function is used. • The authors back test their methods on 50 very liquid futures contracts from 2011 to 2019, and their algorithms deliver positive profits despite heavy transaction costs.