Quantpedia Premium Update – 5th May

New strategies:

Period of rebalancing: Monthly
Markets traded: equities
Instruments used for trading: stocks
Complexity: Complex strategy
Backtest period: 2007-2016
Indicative performance: 6.55%
Estimated volatility: 7.17%

Source paper:

Blocher, Jesse and Dong, Xi and Ringgenberg, Matthew C. and Savor, Pavel G.: Short Covering
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2634579
Abstract:
We construct novel measures of gross and net short covering to examine when short sellers exit their positions. We find that idiosyncratic limits to arbitrage, such as adverse stock price movements, volatility, and equity lending fees are associated with significantly higher position closures and lower price efficiency. Moreover, these position closures predict future return movements in the wrong direction, suggesting short sellers may be induced to exit too early. In contrast, we find little evidence that aggregate limits to arbitrage including VIX, funding liquidity, and market volatility affect gross or net short covering. The results show that firm-level limits to arbitrage are important determinants of trading behavior.

#865 – Pairs Trading with Wavelet Transform

Period of rebalancing: Daily
Markets traded: equities
Instruments used for trading: stocks
Complexity: Very complex strategy
Backtest period: 2010-2018
Indicative performance: 36.9%
Estimated volatility: 10%

Source paper:

Eroglu, B. A. and Yener, H. and Yigit, T.: Pairs Trading with Wavelet Transform
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4259385
Abstract:
We show that applying the wavelet transform to S\&P 500 constituents’ prices generates a substantial increase in the returns of the pairs-trading strategy. Pairs trading strategy is based on finding prices that move together, but if there is shared noise in the asset prices, the co-movement, on which one base the trades, might be caused by this common noise. We show that wavelet transform filters away the noise, leading to more profitable trades. The most notable change occurs in the parameter estimation stage, which forms the weights of the assets in the pairs portfolio. Without filtering, the parameters estimated in the training period lose relevance in the trading period. However, when prices are filtered from common noise, the parameters maintain relevance much longer and result in more profitable trades. Particularly, we show that more precise parameter estimation is reflected on a more stationary and conservative spread, meaning more mean reversion in opened pairs trades. We also show that wavelet filtering the prices reduces the downside risk of the trades considerably.

#866 – Narrative-Based Asset Allocation

Period of rebalancing: Daily
Markets traded: equities
Instruments used for trading: stocks
Complexity: Very complex strategy
Backtest period: 2015-2021
Indicative performance: 18.31%
Estimated volatility: 14.38%

Source paper:

Bhargava, R. and Lou, X. and Ozik, G. and Sadka, R. and Whitmore, T.: Quantifying Narratives and their Impact on Financial Markets
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4166640
Abstract:
This paper introduces a media-coverage-based approach to quantify narratives and develops methodologies to explain the extent to which narratives drive financial markets and returns of investment portfolios. We show that media-derived narratives may contain predictive information for market returns beyond traditional macro indicators. Finally, we demonstrate that narrative indicators can be used to enhance asset allocation strategies and to gain or hedge exposure to narratives by constructing portfolios of narrative-sensitive assets.

#867 – Cross Industry Dispersion Factor

Period of rebalancing: Monthly
Markets traded: equities
Instruments used for trading: stocks
Complexity: Moderately complex strategy
Backtest period: 1963-2018
Indicative performance: 6.04%
Estimated volatility: 13.67%

Source paper:

Pinchuk, Mykola, Labor Income Risk and the Cross-Section of Expected Returns
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4334242
Abstract:
This paper explores asset pricing implications of unemployment risk from sectoral shifts. I proxy for this risk using cross-industry dispersion (CID), defined as a mean absolute deviation of returns of 49 industry portfolios. CID peaks during periods of accelerated sectoral reallocation and heightened uncertainty. I find that expected stock returns are related cross-sectionally to the sensitivities of returns to innovations in CID. Annualized returns of the stocks with high sensitivity to CID are 5.9% lower than the returns of the stocks with low sensitivity. Abnormal returns with respect to the best factor model are 3.5%, suggesting that common factors can not explain this return spread. Stocks with high sensitivity to CID are likely to be the stocks, which benefited from sectoral shifts. CID positively predicts unemployment through its long-term component, consistent with the hypothesis that CID is a proxy for unemployment risk from sectoral shifts.

#868 – Industry-Relative Stock Beta in Chinese Equities

Period of rebalancing: Monthly
Markets traded: equities
Instruments used for trading: stocks
Complexity: Moderately complex strategy
Backtest period: 2001-2020
Indicative performance: 12.55%
Estimated volatility: 19.04%

Source paper:

Tang, Guohao and Wu, Yiyong and Lou, Guanyu, Extrapolation beyond Peers: An Asset Pricing Perspective
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4234406
Abstract:
We construct a new measure related to biased beliefs based on industry relative stock market beta (IRSB) in the Chinese stock market. We find that stocks in the highest IRSB decile generate 12.84% more annualized return compared to stocks in the lowest IRSB decile. The IRSB premium is not driven by well-known risk factors and other return predictors and persists in alternative beta estimation and large firms. We argue that the outperformance of high IRSB stocks primarily stems from price inflation due to investors’ extrapolation bias.

New research papers related to existing strategies:

#137 – Trend-following in Futures Markets
#645 – Statistical Arbitrage With CNN and Transformer Networks

Lezmi, Edmond and Xu, Jiali: Time Series Forecasting with Transformer Models and Application to Asset Management
https://ssrn.com/abstract=4375798
Abstract:
Since its introduction in 2017 (Vaswani et al., 2017), the Transformer model has excelled in a wide range of tasks involving natural language processing and computer vision. We investigate the Transformer model to address an important sequence learning problem in finance: time series forecasting. The underlying idea is to use the attention mechanism and the seq2seq architecture in the Transformer model to capture long-range dependencies and interactions across assets and perform multi-step time series forecasting in finance. The first part of this article systematically reviews the Transformer model while highlighting its strengths and limitations. In particular, we focus on the attention mechanism and the seq2seq architecture, which are at the core of the Transformer model. Inspired by the concept of weak learners in ensemble learning, we identify the diversification benefit of generating a collection of low-complexity models with simple structures and fewer features. The second part is dedicated to two financial applications. First, we consider the construction of trend-following strategies. Specifically, we use the encoder part of the Transformer model to construct a binary classification model to predict the sign of an asset’s future returns. The second application is the multi-period portfolio optimization problem, particularly volatility forecasting. In addition, our paper discusses the issues and considerations when using machine learning models in finance.

#726 – Technical Indicators Predict Cross-Sectional Expected Stock Returns

Un, Kuok Sin and Li, He: The Aggregate Profitability of Technical Analysis
https://ssrn.com/abstract=4359825
Abstract:
We propose an aggregate technical trading index by extracting the most relevant forecasting information contained in 7,846 technical trading rules to predict equity risk premium in the U.S. The proposed method is based on the false discovery rate and partial least squares approaches that alleviate the impact of data-snooping bias and idiosyncratic noise components in technical indicators. We find evidence that the aggregate technical trading index can deliver sizable economic gains for mean-variance investors in asset allocation analysis after taking both transaction costs and data snooping bias into account.

#496 – Using Machine Learning to Pick the Right Combination of Risky and Risk-Free Asset

Goodarzi, Milad and Meinerding, Christoph: Asset Allocation with Recursive Parameter Updating and Macroeconomic Regime Identifiers
https://ssrn.com/abstract=4391262
Abstract:
This article studies long-horizon dynamic asset allocation strategies with recursive parameter updating. The parameter estimates for the regime-switching dynamics vary as more and more datapoints are observed and the sample size increases. In such a setting, the globally optimal portfolio strategy cannot be determined due to computational complexity. Among a set of suboptimal strategies, the portfolio performance can be improved substantially if the dynamics of the regimes are estimated from fundamental macroeconomic instead of financial return data. Especially after highly uncertain times, the estimation based on financial market data identifies extreme regimes, leading to extreme hedging demands against regime changes.

#710 – Quantile Curves and the VRP
#539 – Historical and Implied Volatility in FX Options
#742 – Risk-Reversal Options Strategy

Heston, Steven L. and Todorov, Karamfil: Exploring the Variance Risk Premium Across Assets
https://ssrn.com/abstract=4373509
Abstract:
This paper explores the variance risk premium in option returns across twenty different futures, including equities, bonds, currencies, and commodities (energy, metals, and grains). We implement a novel model-free methodology that constructs tradable option portfolios, which replicate realized variance. In the period 2006–2020, most assets had significant variance risk premiums, but the realized S&P 500 variance risk premium was not significantly different from zero. Within a particular asset, option prices across different strikes are related to the level of volatility and the correlation of volatility with futures returns. Returns to variance are not associated with systematic risk, but are related to fat tails, consistent with option dealers demanding a premium for holding idiosyncratic volatility risk. Contrary to Bollerslev et al. (2009), we find that option-implied variance does not positively predict underlying futures returns for the majority of assets. However, implied variance does predict returns to variance-sensitive option portfolios.

#710 – Quantile Curves and the VRP

Zhao, Yanhui and Borochin, Paul: The Economic Value of Equity Implied Volatility Forecasting with Machine Learning
https://ssrn.com/abstract=4341968
Abstract:
We evaluate the importance of nonlinear and interactive effects in implied volatility innovation forecasting by comparing the performance of machine learning models that can search for interactive effects relative to classical ones that cannot, measuring the economic significance of these predictions in crosssectional and time series pricing tests of delta-hedged option returns. Machine learning models offer superior out of sample performance. Since the predictive variables are the same across all models, these performance differences likely capture the value of nonlinear and interactive effects in implied volatility forecasts. Our results are robust to look-ahead bias and model overfitting.

#628 – Social Media Sentiment Factor

Kellner, Tobias and Maxa, Alexander: Sentiment, Social Reach and Deep Learning. Can Financial Microblogging Services Predict Future Stock Returns?
https://ssrn.com/abstract=4342030
Abstract:
Microblogging platforms are playing an increasingly important role in the exchange of information and opinions among investors. Therefore, this study investigates the causal relationship between investor’s sentiment on the StockTwits platform and US stock returns. In the analysis, we also incorporate the social reach of users and tweets which has been largely neglected in previous literature. We demonstrate that our deep-learning model outperforms established dictionary and machine learning approaches in terms of sentiment classification accuracy. While past price movements strongly influence current sentiment, especially in intraday periods, the reverse direction is scarcely statistically detectable. The inclusion of social reach does not increase the explanatory power of sentiment. In more extreme market phases, tweet volume is a stronger predictor, but with mixed implications regarding the sign. An analysis of the prediction success rate of stock prices shows that only slightly less than half of the users are correct with their prediction.

#626 – Image Recognition in Stock Price Charts Predicts Stock Returns

Yang, Keer and Zhang, Guanqun and Bi, Chuan and Guan, Qiang and Xu, Hailu and Xu, Shuai: Improving CNN-Based Stock Trading by Considering Data Heterogeneity and Burst
https://ssrn.com/abstract=4387842
Abstract:
In recent years, there have been quite a few attempts to apply intelligent techniques to financial trading, i.e., constructing automatic and intelligent trading framework based on historical stock price. Due to the unpredictable, uncertainty and volatile nature of financial market, researchers have also resorted to deep learning to construct the intelligent trading framework. In this paper, we propose to use CNN as the core functionality of such framework, because it is able to learn the spatial dependency (i.e., between rows and columns) of the input data. However, different with existing deep learning-based trading frameworks, we develop novel normalization process to prepare the stock data. In particular, we first empirically observe that the stock data is intrinsically heterogeneous and bursty, and then validate the heterogeneity and burst nature of stock data from a statistical perspective. Next, we design the data normalization method in a way such that the data heterogeneity is preserved and bursty events are suppressed. We verify out developed CNN-based trading framework plus our new normalization method on 29 stocks. Experiment results show that our approach can outperform other comparing approaches.

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

Evaluating Factor Models in China

Today, we will evaluate some specifics that are akin to the now second-largest market in the world – China. The abundance of “shell companies” creates a problem when researchers try to uncover sources of alpha in the Chinese market. We present recent research by Zhiyong Li and Xiao Rao (2022) that proposes a new alternative filter, which excludes the stocks with a high estimated shell probability when constructing equity factor models.

Price Momentum or Factor Momentum: What Leads What?

Continuing our research of different factor allocations and models, we will look at the evergreen momentum effect closer. Cakici, Fieberg, Metko, and Zaremba’s (January 2023) paper contributes to the never-ending debate of the chicken-or-egg problem of what comes first: Does the stock price momentum originate from the factor momentum? The study reexamined the relationship between the factor and price momentum on an extensive sample of 95 years of data from 51 countries. And what are the main takeaways? Let’s find out…