Quantpedia Premium Update – June 8th

New Strategies:

#1133 – Second-minus-First Spreading Returns Strategy

Period of rebalancing: Weekly
Markets traded: commodities
Instruments used for trading: futures, CFDs
Complexity: Complex strategy
Backtest period: 1980-2022
Indicative performance: 7.62%
Estimated volatility: 5.36%

Source paper:

Rossi, Alberto G. and Zhang, Yingguang and Zhu, Yandi: Short-Term Basis Reversal
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5250499
Abstract: We identify a previously undocumented form of return predictability in commodity futures markets, which we refer to as short-term basis reversal: The return spread between adjacent maturity contracts exhibits systematic negative autocorrelation. Basis reversal is independent of the short-term reversal of the individual contracts. Instead, it stems from the differential price sensitivity to news across the futures curve and is stronger among futures contracts with higher return volatility, more autocorrelated returns, and less correlated returns across maturities. Consistent with a preferred-habitat and limits to arbitrage interpretation, we show that basis reversal is also present in other assets characterized by a term structure, such as stock index futures, corporate bonds, and treasury bonds.

#1134 – Reversal in Low Abnormal Volatility Portfolios

Period of rebalancing: Monthly
Markets traded: equities
Instruments used for trading: stocks
Complexity: Complex strategy
Backtest period: 1996-2024
Indicative performance: 14.25%
Estimated volatility: 21.8%

Source paper:

Bank, Matthias: Relative visibility and attraction in asset pricing
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5199257
Abstract: Numerous studies show that stocks that attract a lot of investor attention tend to have low subsequent returns. Motivated by these findings, I proxy the relative visibility of stocks in the cross section with their ranked abnormal return volatility. I argue that this simple measure is consistent with salience theory as well as the concept of information gaps, curiosity, and motivated attention. For a sample of US large-cap stocks, I show that both high (low) relative visibility alone and combined with high uncertainty, as proxied by idiosyncratic volatility, lead to statistically and economically low (high) subsequent returns. The results are consistent with time-varying expected returns and a large predictable return component.

#1135 – Time Series Momentum in China

Period of rebalancing: Monthly
Markets traded: equities, commodities, bonds
Instruments used for trading: futures, CFDs
Complexity: Complex strategy
Backtest period: 1999-2019
Indicative performance: 16.24%
Estimated volatility: 18.51%

Source paper:

Shi, Chuan and Lian, Xiangbin: Trend Following Strategy: A Practical Guide
https://ssrn.com/abstract=5140633
Abstract: Trend-following strategies, rooted in time-series momentum, have demonstrated enduring efficacy across diverse asset classes and market conditions. This paper provides a comprehensive exploration of the theoretical foundations, risk-return dynamics, and practical implementation of such strategies. Key findings reveal that trend following strategies exhibit convex return profiles with “crisis alpha” properties, thriving during market dislocations while offering diversification benefits. The choice of time scale for trend calculation is paramount, as strategies must align with an asset’s return characteristics. Empirical simulations and comparative analyses of trend measures underscore that methodology matters less than selecting an appropriate time horizon. Empirical evidence from China’s futures markets (1999–2019) confirms the global applicability of trend following, with a combined multi-scale strategy achieving a 16.24% annualized return and a Sharpe ratio of 0.88, significantly outperforming buy-and-hold approaches. However, challenges persist, including periods of underperformance linked to heightened asset correlations, reduced market trends, and leverage sensitivity. Critically, the strategy’s success hinges on disciplined risk management and acceptance of short-term drawdowns. By synthesizing academic insights and practical considerations, this study reaffirms trend following as a resilient, albeit imperfect, tool for portfolio diversification and crisis hedging, urging practitioners to prioritize time-scale alignment and robust risk controls.

#1136 – Twitter Follower Growth

Period of rebalancing: Daily
Markets traded: equities
Instruments used for trading: stocks
Complexity: Very complex strategy
Backtest period: 2020-2023
Indicative performance: 12.3%
Estimated volatility: 5.91%

Source paper:

Pyun, Chaehyun: Twitter Follower Growth: Social Media Engagement as Investment Indicators
https://ssrn.com/abstract=5171833
Abstract: This paper investigates the relationship between Twitter follower growth and stock returns. Firms with significant follower growth tend to be growth-oriented firms in the consumer goods and high-technology sectors. An investment strategy based on Twitter follower growth demonstrates market outperformance, especially for shorter rebalancing intervals. Large-cap, growth-oriented, and volatile stocks are more likely to have Twitter accounts and exhibit higher long-term returns. Firms with significant growth in Twitter followers are associated with higher future returns, with stronger predictability for large-cap stocks. These findings suggest that social media engagement metrics serve as valuable indicators for stock performance.

#1137 – Generative AI and Fundamentals-Based Exchange Rate Forecasting

Period of rebalancing: Monthly
Markets traded: currencies
Instruments used for trading: forwards, futures, CFDs
Complexity: Very complex strategy
Backtest period: 2000-2024
Indicative performance: 4.35%
Estimated volatility: 7.34%

Source paper:

Izadyar, Amin: Generative AI and Fundamentals-Based Exchange Rate Forecasting
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5228767
Abstract: I use generative artificial intelligence (AI) to forecast currency returns based on economic fundamentals and revisit the exchange rate disconnect puzzle first documented by Meese and Rogoff (1983). Using ChatGPT (and DeepSeek), I analyze a comprehensive dataset of economic data releases for major currency pairs and construct variables that capture the fundamental strength of each currency’s economy. These AI-powered variables exhibit significant cross-sectional predictive power, as a simple trading strategy that goes long on currencies with strong AI-derived fundamentals and short on those with weak AI-derived fundamentals achieves an annualized Sharpe ratio exceeding 0.7. The excess returns of this strategy remain significant after controlling for traditional currency factors. Furthermore, to mitigate look-ahead bias, I implement rigorous tests to ensure that AI’s performance stems from reasoning rather than memorization. Finally, I explore the underlying sources of this predictability and find evidence suggesting that the Taylor rule framework, which central banks use to set interest rates, is a key mechanism connecting exchange rates to economic fundamentals.

#1138 – The Intersection of Expected Returns

Period of rebalancing: Monthly
Markets traded: equities
Instruments used for trading: stocks
Complexity: Very complex strategy
Backtest period: 1956-2022
Indicative performance: 9.74%
Estimated volatility: 14.98%

Source paper:

Sobotka, Austin: The Intersection of Expected Returns
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5244033
Abstract: A relatively small number of stocks plays a disproportionately large role in explaining the performance of 164 cross-sectional asset pricing anomalies. For instance, excluding the top 10% of stocks that are shared across the most anomaly portfolios for a given month reduces the average anomaly’s return and alpha by approximately 40%. These stocks can be identified ex ante and used to form long-short portfolios that generate abnormal returns more than three times larger than that of the average anomaly portfolio. Consistent with prior research, I find evidence that biased investor expectations help explain the returns to these stocks, suggesting that a significant portion of the returns to the 164 anomalies can be attributed to mispricing. My results have implications for traditional asset pricing, behavioral finance, and for investors and practitioners in the factor investing space.

#1139 – Market Timing Nifty 50 Index using S&P500 daily performance

Period of rebalancing: Daily
Markets traded: equities
Instruments used for trading: futures, CFDs, ETFs
Complexity: Simple strategy
Backtest period: 2008-2025
Indicative performance: 94%
Estimated volatility: –

Source paper:

Rahman, Maksudur: Cross-Market Momentum: Leveraging S&P 500 Signals to Predict Nifty 50 Performance
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5177985
Abstract: Stock indices serve as a structured representation of a country’s stock market. In this paper, I examine whether the performance of one country’s index impacts the next day’s performance of another country’s index. I discovered that the overall performance of one index is positively correlated with the performance of another. I developed a strategy of buying (or selling) Nifty 50 index futures based on the previous day’s long (or short) position of the S&P 500. This momentum strategy involved holding ﬁve positions in Nifty 50 for ﬁve consecutive days using the signal from the S&P 500. Among these ﬁve days, the ﬁrst day (the next day) performance based on the S&P 500 signal outperformed all the other four days, signiﬁcantly surpassing the benchmark. Using this strategy, Nifty 50 generated an average annual return of 94%, compared to just 11% for the benchmark (Actual Nifty 50)

New research papers related to existing strategies:

#183 – Optimized Currency Portfolios

Fan, Minyou and Kearney, Fearghal and Li, Youwei and Liu, Jiadong: Rethinking Currency Factors: The Case for Mean-Variance Optimisation
https://ssrn.com/abstract=5240833
Abstract: We show that a simple mean-variance (MV) optimisation can substantially enhance the performance of established currency factor strategies such as Carry, Value, and Momentum. We find that the improved performance is due to the stronger cross-sectional predictability of the optimised strategies. International diversification plays a key role in optimisation due to 1) the low correlation between developed and emerging currencies, and 2) the low level of comovement across emerging currencies. We also find that the outperformance of our proposed MV optimised factor portfolios is positively related to the standard deviation of currency abnormal returns over time. Our asset pricing tests suggest that the MV optimised factors subsume the corresponding plain currency factors.

#453 – Machine Learning Adaptive Portfolio Asset Allocation

Abouseir, Amine and Le Manach, Arthur and El Mennaoui, Mohamed and Zheng, Ban: Integration of Macroeconomic Data into Multi-Asset Allocation with Machine Learning Techniques
https://ssrn.com/abstract=3586040
Abstract: In this paper, we propose a new way to predict market returns for multi-assets (equity, fixed-income and commodity) by extracting features from macroeconomic data and performing machine learning algorithms for both regression and classification. Our approach aims to select robust models to build alternative risk premia portfolio. We apply machine learning algorithms to our investment universe and then apply different portfolio allocation methods. We discover the importance of integrating macroeconomic data to build portfolio, especially with classification techniques which enhance the Sharpe ratios of strategies.

#38 – Accrual Anomaly

Halim, Asyraf Abdul and Abd Sukor, Mohd Edil: Accrual-Based Strategies and Alpha Persistence: Insights from U.S. Shariah-Compliant Stocks
https://ssrn.com/abstract=5245430
Abstract: This study examines asset pricing anomalies in U.S. Shariah-compliant (SC) stocks, focusing on the role of accrual-based strategies in generating statistically significant alphas. While prior literature highlights the persistence of accrual anomalies in conventional stock markets, the asset pricing behavior of SC stocks remains underexplored. Using a dataset spanning 1993–2023, we construct 175 size-accrual portfolios and test their performance across standard asset pricing models, including the CAPM, Fama-French three-factor (FF3), five-factor (FF5), and six-factor (FF6) models. Using our newly proposed Trifecta Conditions, we find that the CAPM suffers heavily in model misspecification, and subsequently, should be not used. At the same time, we also found that no single model amongst the FF3, FF5 and FF6 have the best model specification to explain all of our size-accrual portfolios. In other words, there exists no one-size-fits-all model to explain the average excess returns of portfolios in our US SC stocks sample, instead, each model performs best in different sorts of accruals. Then, our results indicate that most accrual definitions generate statistically insignificant alphas, indicating that US SC stocks do not conduct earnings management practices, consistent with its fundamental Islamic values. For certain accrual definitions however, statistically significant alphas may yet be found, this is usually the case for narrow accrual definition related to current and non-current operating assets. Additionally, our findings suggest that the size, value and momentum premiums remain the most critical risk factors in explaining average excess returns US SC stocks, while the market, profitability and investment premiums exhibit negligible explanatory power. These results have important implications for investors seeking to implement accrual-based strategies within the US SC stock universe. The findings suggest that while some accrual anomalies persist, the broader US SC market may exhibit pricing dynamics that differ from US conventional stocks due to unique corporate financial behaviors, wider investor base and narrower investment base. By shedding light on these anomalies, this paper contributes to the ongoing debate on the efficiency of US SC stock markets and the applicability of accrual-based investment strategies in faith-compliant financial environments.

#334 – Volatility-Adjusted Momentum in Corporate Bonds
#426 – 1 Month Momentum in Bonds

Keshavarz, Javad and Sirmans, Stace: Bond Factor Momentum and Its Predictability for Stock Returns
https://ssrn.com/abstract=5242278
Abstract: This paper investigates momentum effects in corporate bond factors and spillover into equity markets. We find robust evidence of momentum within bond factors, with past winners outperforming losers by economically significant margins (up to 11.2% per year). This factor-level momentum subsumes security-level bond momentum, suggesting that bond return predictability is primarily a systematic rather than idiosyncratic phenomenon. We also discover significant cross-market momentum spillovers from bond to stock factors, with stronger effects for factors exposed to credit risk. These bond-to-stock spillovers intensify during market stress and improve with bond market liquidity. Our findings suggest bond markets play a unique role in processing systematic information, particularly related to credit conditions, challenging the notion of equity market informational dominance.

#558 – Quality Strategy in the Indian Market

Jagarlapudi, Chaitanya and Gupta, Vatsal and Gupta, Rudraksh: How many stocks for maximizing risk-adjusted return? Perspective from Indian Stock Market
https://ssrn.com/abstract=3972603
Abstract: Portfolio managers and investors alike continuously grapple with trying to outperform the benchmarks, while keeping in mind the right number of stocks in the portfolio for optimal diversification. Through our analysis of the Indian stock market, we show that by random selection of stocks, the odds of outperforming the index are very low. Instead, if investors incorporate the ‘quality’ factor in stock selection, the probability of outperforming the index improves substantially and is statistically significant – at around 5% outperformance per annum at 90% confidence. By incorporating the quality factor, they can own a fairly small basket of 15 to 25 stocks, which captures 90% of the benefit of owning the benchmark at 90% confidence and generates strong outperformance compared to the index. This is much lower than the 45 stocks required for reducing only the diversifiable risk by 90% with a 90% confidence, as the previous studies suggest. These could serve as practical guidelines for portfolio construction and improving returns.

#1 – Asset Class Trend-Following

Zarattini, Carlo: Global Tactical Asset Allocation Updated Results and Real-Market Implementation Using Python and IBKR
https://ssrn.com/abstract=5230603
Abstract: This study revisits and expands on one of the most influential investment research papers of the past two decades: A Quantitative Approach to Tactical Asset Allocation, authored by Meb Faber in 2006 and published in The Journal of Wealth Management in 2007. We begin with a concise overview of the original strategy, then present updated historical performance results using data through March 2025. Next, we explore how different rebalancing frequencies can affect outcomes, and propose a tranche-based approach to help practitioners reduce the impact of rebalance timing luck. Finally, we show how investors can put this strategy into practice with a simple Python script that automates portfolio rebalancing through Interactive Brokers.

And several interesting free blog posts that have been published during the last 2 weeks:

Quantpedia Awards 2025 – Winners Announcement

This is the moment we all have been waiting for, and today, we would like to acknowledge the accomplishments of the researchers behind innovative studies in quantitative trading. So, what do the top five look like, and what will the authors of the papers receive?

Let’s find out …

Can We Finally Use ChatGPT as a Quantitative Analyst?

In two of our previous articles, we explored the idea of using artificial intelligence to backtest trading strategies. Since then, AI has continued to develop, with tools like ChatGPT evolving from simple Q&A assistants into more complex tools that may aid in developing and testing investment strategies—at least, according to some of the more optimistic voices in the field. Over a year has passed since our first experiments, and with all the current hype around the usefulness of large language models (LLMs), we believe it’s the right time to critically revisit this topic. Therefore, our goal is to evaluate how well today’s AI models can perform as quasi-junior quantitative analysts—highlighting not only the promising use cases but also the limitations that still remain.

Pre-Announcement Drift for BoE, BoJ, SNB: Do Markets Move Before the Word Is Out?

We’ve previously examined how central bank policy decisions—particularly those by the Federal Reserve and the European Central Bank (ECB)—impact stock market behavior. The price drift in U.S. equities around the Federal Open Market Committee (FOMC) meetings is a well-documented phenomenon. Likewise, our research study of the ECB revealed a pre-announcement drift, underscoring the anticipatory nature of equity markets ahead of key policy events and the potential opportunities for trading strategies. But are such price drifts unique to the Fed and ECB? In this article, we broaden the scope to investigate whether similar market behavior occurs around monetary policy announcements by other major central banks: mainly the Swiss National Bank (SNB), the Bank of England (BoE), and the Bank of Japan (BoJ).