Return Predictability Research Articles

Can a machine learn from behavioral biases? Evidence from stock return predictability of deep learning models

We examine how the return predictability of deep learning models varies with stocks’ vulnerability to investors’ behavioral biases. Using an extensive list of anomaly variables, we find that the long-short strategy of buying (shorting) stocks with high (low) deep learning signals generates greater returns for stocks more vulnerable to behavioral biases, i.e., small, young, unprofitable, volatile, non-dividend-paying, close-to-default, and lottery-like stocks. This performance of deep learning models for speculative stocks becomes pronounced when investor sentiment is high, and when new information is delivered through earnings announcements. Moreover, our nonlinear deep learning signals are negatively associated with analysts’ earnings forecast error especially for speculative stocks, implying that analysts’ forecasts are too low for speculative stocks with high deep learning signals. These results suggest that deep learning models with nonlinear structures are useful for capturing mispricing induced by behavioral biases.

EFFICIENCY MARKET HYPOTHESIS, RANDOM WALK THEORY, AND FINANCIAL RATIOS-STOCK RETURNS: A COMPARATIVE STUDY OF BANKS IN INDONESIA, INDIA, AND CHINA IN THE POST-COVID ERA

This research aims to analyze how financial ratios affect firm value, utilizing the Efficiency Market Hypothesis and Random Walk Theory for analysis. It focuses on banking sector companies listed on the Indonesia Stock Exchange (XIDX), India Stock Exchange (XBOM), and China Stock Exchange (XSHG) from 2022 to 2023. This quantitative study employs purposive sampling, resulting in 184 firm- year samples, with data gathered from annual reports. Regression analysis using SPSS version 26 is the chosen method. The findings suggest that India has the most efficient market, while Indonesia aligns with the random walk theory, indicating weaker market efficiency. Notably, variables like "Return on Average Equity" in Indonesia and India and "Impairment Loans to Equity" in India and China significantly impact stock returns. These varying influences underscore the importance of considering country-specific factors in stock return predictions. However, it's essential to acknowledge tha the models explain only a portion of the variance in the dependent variable.

Solving the Forecast Combination Puzzle Using Double Shrinkages*

AbstractThis study develops a new approach that shrinks the forecast combination weights towards equal weights by using weighted least squares and towards zero weight by using regularization constraints. We reveal the significant predictability of excess returns to the S&P500 index that can be achieved by using this double shrinkage combination (DSC). Furthermore, our DSC approach significantly outperforms the naïve equal‐weighted combination, solving the combination puzzle. The equal‐weight shrinkage has greater effect in economic recessions, whereas the zero‐weight shrinkage dominates in economic expansions. The DSC's superior performance over that of the naïve combination is observed in the application of forecasting macroeconomic indicators.

Efficient predictability of oil price: The role of VIX-based panic index shadow line difference

This paper shows that the panic index shadow line difference (ULD) can be an effective predictor of oil returns. We use a candlestick chart to plot the investor panic index (VIX) and subtract the lower shadow from the upper shadow to obtain ULD. The in-sample analysis shows that the ULD can significantly and negatively predict oil returns. The out-of-sample results show that the inclusion of ULD, an exogenous regression variable, in the model not only substantially improves the predictive accuracy of oil returns, but also yields good economic benefits when using its predicted values for portfolio investment. All bivariate regression models that include ULD as an exogenous regression variable obtain higher prediction accuracy than univariate regression models, both for in-sample and out-of-sample predictions. All the robustness tests done in this paper show that ULD is a powerful predictor that significantly improves the predictability of oil returns.

Return prediction: A tree-based conditional sort approach with firm characteristics

This paper proposes the Augmented Tree-Based Conditional Sort approach to investigate the return prediction problems. We find a momentum effect in the Chinese stock market using monthly data over 5000 stocks from January 2005 to December 2022. The corresponding momentum strategy robustly outperforms the market portfolio and earns an excess return of 10.60 % annually. Besides a full-period momentum effect, we find a sub-period reversal effect before year 2018 and locate year 2018 as a turning point from the reversal effect to the momentum effect. We attribute this phenomenon to the rapid increase of institutional investors since then.

The cross-sectional stock return predictions via quantum neural network and tensor network

In this paper, we investigate the application of quantum and quantum-inspired machine learning algorithms to stock return predictions. Specifically, we evaluate the performance of quantum neural network, an algorithm suited for noisy intermediate-scale quantum computers, and tensor network, a quantum-inspired machine learning algorithm, against classical models such as linear regression and neural networks. To evaluate their abilities, we construct portfolios based on their predictions and measure investment performances. The empirical study on the Japanese stock market shows the tensor network model achieves superior performance compared to classical benchmark models, including linear and neural network models. Though the quantum neural network model attains the lowered risk-adjusted excess return than the classical neural network models over the whole period, both the quantum neural network and tensor network models have superior performances in the latest market environment, which suggests capability of model’s capturing non-linearity between input features.

Salience theory and cryptocurrency returns

The salience theory of choice under risk shows that investor behavior drives cross-sectional cryptocurrency returns. Investors place too much weight on salient payouts, causing overvaluation of cryptocurrencies with upward salience returns and undervaluation of those with downward salience returns, leading to negative expected returns for the former and positive expected returns for the latter. The salience effect in the cryptocurrency market is more pronounced than in equity markets, making it a significant risk factor for explaining other cross-sectional returns in the cryptocurrency market. Unlike other documented return predictors, the salience theory uniquely contributes to understanding the cryptocurrency market.

Limits of arbitrage and their impact on market efficiency: Evidence from China

This paper examines the impact of limits of arbitrage (LOA) on market efficiency by considering a large sample of Chinese stocks. Intraday market efficiency is measured using two widely used measures: the intraday return predictability measure and the variance ratio measure. We find that LOA plays a major role in driving market efficiency. We also use an exogenous shock, the introduction of Shenzhen Stock Connect, which increases trade volume and lowers LOA, to examine the impact of LOA on market efficiency. Our results indicate that this event improves market efficiency regarding both our measures. Therefore, we conclude that policy makers in China are acting in the right direction to elevate China's status in world markets by adapting policies that inherently make the country more attractive to global investors by lowering LOA and enhancing market efficiency.

Inferring Intermediary Risk Exposure from Trade

We propose a novel measure of intermediary risk exposure based on the fraction of all trade that is conducted between dealers, called the interdealer trade (IDT) measure. Intuitively, when dealers’ aggregate risk exposure rises, they trade more with each other to redistribute inventory shocks. Consistent with risk exposures relating to expected returns, market-specific IDT measures add incremental return predictability across five different markets. For example, one-standard-deviation increases in the Treasury and foreign exchange (FX) IDT measures, respectively, forecast a 1.8% higher annual excess return on a five-year bond and a 3.7% higher annual excess return on currency-specific FX trades. This paper was accepted by David Sraer, finance. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2021.01831 .

Risk-neutral skewness and stock market returns: A time-series analysis

This paper investigates whether the change of average risk-neutral skewness (RNS), which is the average of monthly risk-neutral skewness across firms, can predict subsequent aggregate stock returns. We find that average RNS positively and significantly predicts future aggregate stock returns, consistent with the firm-level evidence. Our findings are robust after controlling for other well-documented financial and economic stock return predictors. Moreover, we document that the robustness of predictability still holds in out-of-sample settings. Finally, we show that the forecasting ability of average RNS stems from its better performances during the economic recession rather than economic expansion and its pronounced predictability among stocks that are more speculative and difficult to arbitrage.

Imaging of muscle injuries in soccer.

Accurate diagnosis of muscle injuries is a challenge in everyday clinical practice and may have profound impact on the recovery and return-to-play decisions of professional athletes particularly in soccer. Imaging techniques such as ultrasound and magnetic resonance imaging (MRI), in addition to the medical history and clinical examination, make a significant contribution to the timely structural assessment of muscle injuries. The severity of a muscle injury determined by imaging findings has a decisive influence on therapy planning and affects prognosis. Imaging is of high importance when the diagnosis or grade of injury is unclear, when recovery is taking longer than expected, and when interventional or surgical management may be needed. This narrative review will discuss ultrasound and MRI for the assessment of sports-related muscle injuries in the context of soccer, including advanced imaging techniques, with the focus on the clinical relevance of imaging findings for the prediction of return to play.

Rates and Predictors of Returns to Homelessness Among Veterans, 2018–2022

IntroductionThe progress made by the U.S. Department of Veterans Affairs (VA) towards ending veteran homelessness requires attention be paid to preventing returns to homelessness. MethodsUsing national VA data on 293,820 exits from a VA homeless program to a permanent housing destination between January 2018 thru December 2022, rates and predictors of returns to homelessness among veterans were examined. Analyses were conducted June-August 2023. A return to homelessness was operationally defined as a return encounter with a VA homeless program. Results5.8% of successful exits to permanent housing resulted in a return to homelessness within 6 months, 10.2% return within 12 months, and 16.7% return within 24 months. In the total sample, veterans who were male (hazard ratio [HR]= 1.47), widowed (HR= 1.29), had diagnoses of drug use disorder (HR= 1.40) or psychotic disorder (HR= 1.20), and had used more inpatient or urgent care services in the year prior (HR= 1.05-1.15) were at significantly greater risk of returning to homelessness. Many of these predictors remained significant in subgroup analyses of female veterans, veterans aged 65 or older, and veterans in the Housing and Urban Development-Veterans Affairs Supportive Housing (HUD-VASH) program. ConclusionsMost homeless veterans served by VA who exit to permanent housing do not return to homelessness within 2 years. The most critical period seems to be the first year, when 1 in 10 veterans return to homelessness. Knowledge of these risk factors may be important in planning secondary and tertiary prevention efforts for homelessness.

Predictable EPS growth and the performance of value investing

AbstractPrevious research finds that EPS growth rates are difficult to predict and reasons that much of the observed cross-sectional variation in valuation ratios is due to variation in implied future stock returns. Yet the observed cross-sectional relation between valuation ratios and realized future stock returns is weak. We revisit these findings using a refined measure of expected EPS growth rates and document robust evidence of predictability in EPS growth rates. Moreover, we find that this predictable growth extends beyond two years into the future and is strongly reflected in observed valuation ratios. We show that combining valuation ratios with our refined measure of expected EPS growth rates improves forecasts of stock returns, though return predictability remains weak. Thus, we conclude that most of the variation in valuation ratios is driven by predictable EPS growth.

Improving SMEs Performance Based on Strategy Management

This study aims to analyze the expected return of stocks in the pharmaceutical sub-sector based on the capital asset pricing model (CAPM), arbitrage pricing theory (APT), and the Three Factor Model (TFM) during the COVID-19 pandemic from March 2020 to December 2023, and test the accuracy of the expected return prediction using the Mean Average Deviation (MAD). Determination of the pharmaceutical sub-sector as a research sample with the consideration of producing positive return growth during the COVID-19 pandemic, while at the same time, other companies were experiencing a downturn. The results showed that the average expected return was 3.69% based on the CAPM, 3.88% based on the APT formula, and 3.66% based on TFM calculations. However, there is no significant difference in the resulting expected return. The regression analysis concludes TFM is better than other asset pricing models. Beta significantly explains the average weekly pharmaceutical stock return. The measure of the MAD test concluded the CAPM is a model that can predict the expected return more accurately by the smallest MAD value. TFM is a model with the highest MAD value. The highest level of MAD, the less accurate prediction of the expected return.

Option gamma and stock returns

Stocks with high net gamma exposure systematically underperform stocks with low net gamma exposure. This effect is distinct from other well-known return predictors, and survives many robustness checks. We show that stocks with low net gamma exposure negatively predict future realized volatility, and argue that investors command a risk premium to hold low net gamma exposure stocks, which are riskier. Lastly, we show that the volatility predictability stems from a non-informational channel, and not from private information.

Return predictability of prospect theory: Evidence from the Thailand stock market

Using Thailand stock market data, we find that prospect theory has strong predictive power for returns. This predictive power is strengthened during crises and bear and bull markets. The loss aversion component is the main contributor to the increased predictive power during crises and bear markets. In contrast, the probability weighting and concavity/convexity components contribute more to the predictive power during bull markets. Prospect theory has stronger predictive power in the Market for Alternative Investment than in the Securities Exchange of Thailand, providing evidence that individual investors prefer the mental presentation effect and evaluate risk in a way described by prospect theory.

Forecasting cryptocurrencies returns: Do macroeconomic and financial variables improve tail expectation predictions?

This study aims to jointly predict conditional quantiles and tail expectations for the returns of the most popular cryptocurrencies (Bitcoin, Ethereum, Ripple, Dogecoin and Litecoin) using financial and macroeconomic indicators as explanatory variables. We adopt a Monotone Composite Quantile Regression Neural Network (MCQRNN) model to make one- and five-steps-ahead predictions of Value-at-Risk (VaR) and Expected Shortfall (ES) based on a rolling window and compare the performance of our model against the Historical simulation and the standard ARMA(1,1)-GARCH(1,1) model used as benchmarks. The superior set of models is then chosen by backtesting VaR and ES using a Model Confidence Set procedure. Our results show that the MCQRNN performs better than both benchmark models for jointly predicting VaR and ES when considering daily data. Models with the implied volatility index, treasury yield spread and inflation expectations sharpen the extreme return predictions. The results are consistent for the two risk measures at the 1% and 5% level both, in the case of a long and short position and for all cryptocurrencies.

ECONOMETRIC ASSESSMENT OF THE EFFICIENCY AND VOLATILITY OF THE STOCK MARKET IN UKRAINE

The validity of published research findings is related to their demand from both investors and issuers and from the side of research scientists. The methods used make it possible to determine the general trend of market movement in the direction of increasing or decreasing its efficiency. For investors, such an assessment of the efficiency of the stock and bond markets creates a basis for optimizing the allocation of assets when building portfolio strategies, and for issuers, it is the basis for building a strategy for emission activity. It is necessary to analyze the factors affecting the efficiency of the Ukrainian market and show that the illiquidity of the Ukrainian bond and stock markets has a negative impact on the overall efficiency of the market, and therefore it is necessary to increase liquidity, which is an important factor for improving the efficiency of the market. The analysis of the efficiency of stock and bond markets is presented as an impact on the markets themselves. For research scientists, the predictability of returns, or the lack of such predictability, is critical to creating models that can accurately describe risks and returns in financial markets.The paper proposes a different approach to the study of stock market efficiency, using a time-varying parameter model that can estimate indicators, efficiency and inefficiency as parameters change.

Unifying Estimation and Inference for Linear Regression with Stationary and Integrated or Near-Integrated Variables

Abstract There is a discrepancy in the limiting distributions of least-squares estimators for stationary and integrated variables. For statistical inference, it must be decided which distribution should be used in advance. This motivates us to develop a unifying inference procedure based on weighted estimation. The asymptotic distributions of the proposed estimators are developed and a random weighting bootstrap method is proposed for constructing confidence regions. The proposed method outperforms existing methods (with time constant or time-varying error variance) in simulations. We further study the predictability of asset returns in a setting where some of our state variables are endogenous.

Technological advancements and noise trading: A case of investors’ sentiments at the Pakistan Stock Exchange

This study uses data from the KSE-100 stock index from 2009 to 2021 to investigate the influence of technological advancements, investors' sentiments, and behavioral factors on stock market inefficiency. The study employs a threshold regression model to distinguish between different thresholds of these factors and a Wald test is used to determine the significance of the variances among thresholds. The results indicate that technological advancements, investors' sentiments, and behavioral factors significantly affect the predictability of returns in the KSE-100 index. This study is unique in the context of Pakistan and provides a comprehensive framework for understanding stock market operations. The findings have practical applications for individuals, researchers, managers, institutions, and policymakers seeking to better understand the impact of technology, society, and behavioral biases on stock markets. Furthermore, this study provides unique insights into addressing the issue of market inefficiency by identifying the factors responsible for such behavior.