Forecast Stock Returns Research Articles

Research and Empirical Evidence of Machine Learning based Financial Statement Analysis Methods

This study presents a novel approach called FinAnalytix which merging machine learning’s prowess in pattern recognition with financial statement analysis. This integrated algorithm combines deep neural networks and recurrent neural networks for predictive accuracy in stock return analysis, alongside logistic regression and random forest models for robust fraud detection in financial statements. The empirical evidence demonstrates FinAnalytix effectiveness in identifying abnormal financial patterns and predicting market reactions to earnings announcements. The study utilizes extensive data from listed companies, ensuring a comprehensive and practical application. FinAnalytix represents a significant advancement in the field, providing a dual approach to financial analysis for enhancing investment strategies through accurate stock return forecasts and bolstering financial integrity by detecting fraudulent activities. The simulation of the study based on the financial data of 100 sample listed companies. This research not only bridges the gap between traditional financial analysis and modern machine learning techniques but also offers a powerful tool for investors and regulatory bodies in navigating the complex financial landscape.

A Comparative Analysis of Deep Learning and Traditional Statistics for Stock Price and Return Forecasting

A Comparative Analysis of Deep Learning and Traditional Statistics for Stock Price and Return Forecasting

Stock return predictability using economic narrative: Evidence from energy sectors

This paper applies the Narrative-based Energy General Index (NEG) to forecast stock returns in the energy industry. The index is constructed using natural language processing (NLP) techniques applied to news topics from The Wall Street Journal. The results indicate that NEG outperforms in predicting future returns of the energy industry in both in-sample and out-of-sample, and the predictive power surpasses that of other macroeconomic variables. The asset allocation exercise demonstrates the substantial economic value of NEG. Furthermore, we document that NEG not only exhibits superior predictive power for energy sector returns but also provides valuable insights for the whole stock market.

Modeling and forecasting stock return volatility using the HARGARCH model with VIX information

AbstractThis study develops a novel approach for improving stock return volatility forecasts using volatility index information with the entropic tilting technique. Unlike traditional linear heteroskedasticity autoregressive methods with option‐implied information, we first derive predictive densities from traditional models, and then tilt using both the first and second moments of the risk‐neutral distribution, which enables us to capture the nonlinear effect in our specification. The empirical findings demonstrate a substantial enhancement in the forecasting accuracy of all models once the first‐ and second‐moment information is considered, where the improvement is both statistically and economically significant. These results have important implications for risk management in well‐established derivatives markets.

Climate Risks and Stock Market Volatility over a Century in an Emerging Market Economy: The Case of South Africa

Because climate change broadcasts a large aggregate risk to the overall macroeconomy and the global financial system, we investigate how a temperature anomaly and/or its volatility affect the accuracy of forecasts of stock return volatility. To this end, we do not apply only the classical GARCH and GARCHX models, but rather we apply newly proposed model-free prediction methods, and use GARCH-NoVaS and GARCHX-NoVaS models to compute volatility predictions. These two models are based on a normalizing and variance-stabilizing transformation (NoVaS transformation) and are guided by a so-called model-free prediction principle. Applying the new models to data for South Africa, we find that climate-related information is helpful in forecasting stock return volatility. Moreover, the novel model-free prediction method can incorporate such exogenous information better than the classical GARCH approach, as revealed by the the squared prediction errors. More importantly, the forecast comparison test reveals that the advantage of applying exogenous information related to climate risks in prediction of the South African stock return volatility is significant over a century of monthly data (February 1910–February 2023). Our findings have important implications for academics, investors, and policymakers.

Causality-driven multivariate stock movement forecasting.

Our study aims to investigate the interdependence between international stock markets and sentiments from financial news in stock forecasting. We adopt the Temporal Fusion Transformers (TFT) to incorporate intra and inter-market correlations and the interaction between the information flow, i.e. causality, of financial news sentiment and the dynamics of the stock market. The current study distinguishes itself from existing research by adopting Dynamic Transfer Entropy (DTE) to establish an accurate information flow propagation between stock and sentiments. DTE has the advantage of providing time series that mine information flow propagation paths between certain parts of the time series, highlighting marginal events such as spikes or sudden jumps, which are crucial in financial time series. The proposed methodological approach involves the following elements: a FinBERT-based textual analysis of financial news articles to extract sentiment time series, the use of the Transfer Entropy and corresponding heat maps to analyze the net information flows, the calculation of the DTE time series, which are considered as co-occurring covariates of stock Price, and TFT-based stock forecasting. The Dow Jones Industrial Average index of 13 countries, along with daily financial news data obtained through the New York Times API, are used to demonstrate the validity and superiority of the proposed DTE-based causality method along with TFT for accurate stock Price and Return forecasting compared to state-of-the-art time series forecasting methods.

Charting by machines

We test the efficient market hypothesis by using machine learning to forecast stock returns from historical performance. These forecasts strongly predict the cross-section of future stock returns. The predictive power holds in most subperiods and is strong among the largest 500 stocks. The forecasting function has important nonlinearities and interactions, is remarkably stable through time, and captures effects distinct from momentum, reversal, and extant technical signals. These findings question the efficient market hypothesis and indicate that technical analysis and charting have merit. We also demonstrate that machine learning models that perform well in optimization continue to perform well out-of-sample.

On the Predictability of the DJIA and S&P500 Indexes

We obtained from Standard and Poor�s Corporation the complete 126-year history of the Dow Jones Industrial Average (DJIA) daily closing prices. We are applying rolling window averaging and adaptive learning methodologies, coupled with robust estimation methods, to examine which are the best forecasting models over a broad range of economic and financial conditions during the life of the index, based on daily and monthly stock index prices, and daily, monthly, and semi-annual stock returns. Why is an AR(1) model a reasonable benchmark of stock prices? Why do we have it? What should be our forecasting benchmarks? Do we find forecasting improvements from the Hendry-Castle-Doornik-Clements approach using robust forecasting methodologies and saturation variables in the prices of the index? Given that the DJIA fell over 15% during the first half of 2022, is this one of the worst six-month periods ever? What has happened to the Dow, historically, during such periods in the past with regard to six-month, one-year, and three-year-ahead stock returns? Is capitalism dead or doomed? We report statistically significant forecasting improvement from saturation and robust forecasting techniques during the 1896�June 2022 period. We report forecast stock returns for the next six months and three years that are bullish. In the King�s English, June 30, 2022, was another excellent common stock buying opportunity and Capitalism is not dead.

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.

Expert aggregation for financial forecasting

Machine learning algorithms dedicated to financial time series forecasting have gained a lot of interest. But choosing between several algorithms can be challenging, as their estimation accuracy may be unstable over time. Online aggregation of experts combine the forecasts of a finite set of models in a single approach without making any assumption about the models. In this paper, a Bernstein Online Aggregation (BOA) procedure is applied to the construction of long-short strategies built from individual stock return forecasts coming from different machine learning models. The online mixture of experts leads to attractive portfolio performances even in non-stationary environments. The inclusion of neural networks experts in the aggregation contributes to a better average return, while Ordinary Least Squares with Huber Loss experts contribute to lower risk. The aggregation outperforms individual algorithms, offering a higher portfolio Sharpe ratio, lower shortfall, with a similar turnover. Extensions to expert and aggregation specialisations are also proposed to improve the overall mixture on a family of portfolio evaluation metrics.

Utilitas Piotroski F-Score Untuk Memprediksi Return Saham Pada Periode 1 Dan Perode 2 Penerbitan Index LQ45

This study investigates the predictive capability of the Piotroski F-Score model in forecasting stock returns during two distinct periods following the issuance of the LQ45 index. The rationale behind dividing the testing periods lies in their temporal proximity to the announcement of audited financial statements, with period 1 being close to this announcement and period 2 occurring eight months thereafter. Such differing time intervals can potentially lead to varying investor decisions. The results of regression analysis, conducted on a sample of 30 consistently listed LQ45 stocks spanning from 2020 to 2022, reveal that the Piotroski F-Score exhibits a positive impact on stock returns during period 1, whereas its influence diminishes during period 2."

The extra value of online investor sentiment measures on forecasting stock return volatility: A large-scale longitudinal evaluation based on Chinese stock market

Numerous studies have purported to show that using online investor sentiment measures can enhance the accuracy of forecasting stock return volatility. However, many of these studies did not validate their findings with out-of-sample data, had insufficient validation samples, or failed to incorporate appropriate controls in their predictive models. None of these studies compared the predictive power of online sentiment measures with commonly used technical sentiment measures, nor did they investigate whether online sentiment measures offer any additional predictive value for volatility forecasting when controlling for technical measures. Using a large stock sample in China A-shares, we contribute to examining the additional predictive power of online sentiment measures for predicting stock volatility for different forecast horizons and different time windows, while controlling for technical measures. Methodologically, through using rolling origins and moving windows evaluation framework, we consider the benefits of using global forecasting models compared to local stock-specific models, and the robustness of our appraisal of the use of sentiment measures. We found that technical sentiment measures generally perform better than online sentiments. When controlling for technical measures, online sentiment measures have extra predictive value only when the forecasting horizon is long (e.g., one week or month), but the magnitudes of the improvement are relatively limited.

Inflation, interest rates and the predictability of stock returns

This paper focuses on the predictive role of inflation and interest rates in forecasting stock returns. We report economically significant in-sample evidence that stock returns are predictable using cyclical components of inflation and interest rates. The out-of-sample analysis results show that these cyclical components have reasonable out-of-sample predictability for the equity risk premium when the comparison is made against a historical average benchmark. We believe this is due to the fact that both inflation and interest rates contain stochastic trends. Accounting for changes in trend inflation (and equilibrium real interest rate) is essential for forecasting stock returns.

Investigation of the Global Stock Trading Based on Visibility Graph and Entropy Weight Method

The increasing complexity and dynamics of the stock trading market are major challenges for the financial industry and are primary dilemmas for all countries nowadays. In addition, the stock trading market has a considerable impact on the global economy, and its importance is self-evident. To cope with the complexity and dynamics of a stock trading market, this paper applies complex network theory and model to explore the topology of the global stock trading network. First, this paper collects stock trading data from 74 countries from 1999 to 2020. It converts the collected stock trading data of these countries into a complex network using a type of algorithm based on the time series visibility graph (VG) algorithm. Then, the data are analyzed by a complex network model, and six analytical metrics are obtained. Finally, the six metrics are analyzed by the entropy weight method to identify the key nodes in the network and to obtain the ranking of each country’s stock trading data. This paper is an effective application of complex network and entropy weight method in stock trend analysis, which mainly includes two contributions. First, the VG algorithm provides a novel research perspective for modeling the global stock trading trend. Second, key nodes in the network are analyzed and identified based on the entropy weight method, and the ranking of key nodes in the stock trading network is obtained, which provides a new method for further research on the stock trading trend, investment portfolio, and stock return forecasting.

How accurate is the CAPM approach compared to the Reward Beta for shares of manufacturing and mining companies in Indonesia's LQ45 Index?

This study evaluates the efficacy of the Capital Asset Pricing Model (CAPM) and Beta Reward Model in forecasting stock returns of companies within the manufacturing and mining sectors listed on Indonesia's LQ45 index. Utilizing monthly closing stock prices from January 2010 to December 2019, the research focuses on ten companies—five from each sector—that consistently appeared in the index throughout the study period. The analysis involves a classic assumption test followed by regression analysis for each company. Key performance indicators, including R-squared, Root Mean Square Error (RMSE), and Mean Absolute Error (MAE), are employed to compare the predictive capabilities of the CAPM and Beta Reward Model. The findings indicate a systematic and superior performance of the Beta Reward Model over the CAPM in predicting stock returns in the Indonesian context. This study contributes to the existing literature on stock return prediction models and provides practical insights for investors and financial analysts in Indonesia.

Comparison of Different Machine Learning Prediction about Stock based on Multi-Factor Input-Taking the Nasdaq Index as an Example

Stock investment has always been widely concerned, and the prediction of future stock trends is what many investors look forward to. There are numerous techniques that can be used to predict stocks as machine learning advances. The commonly used method is support vector machine, random forest, linear regression, etc. Recurrent neural networks, multi-layer perceptron, single-layer LSTM networks, naive Bayes networks, convolutional neural networks, back propagation networks, etc. are examples of deep learning techniques. In historical studies, researchers have tended to predict directly from stock prices or used time series as an independent variable to build a forecasting model for stock prices. In this study, we propose input training parameters based on stock indicators and build a sliding window of it to predict the future price. Based on LSTM, ANN, support vector machine regression, Linear Regression to make predictions on price, and analyze the differences between them. Our research is based on the Nasdaq and the evaluation values show that the neural network approach is effective for stock return forecasting.

Investment and Risk Management with Online News and Heterogeneous Networks

Stock price movements in financial markets are influenced by large volumes of news from diverse sources on the web, e.g., online news outlets, blogs, social media. Extracting useful information from online news for financial tasks, e.g., forecasting stock returns or risks, is, however, challenging due to the low signal-to-noise ratios of such online information. Assessing the relevance of each news article to the price movements of individual stocks is also difficult, even for human experts. In this article, we propose the Guided Global-Local Attention-based Multimodal Heterogeneous Network (GLAM) model, which comprises novel attention-based mechanisms for multimodal sequential and graph encoding, a guided learning strategy, and a multitask training objective. GLAM uses multimodal information, heterogeneous relationships between companies and leverages significant local responses of individual stock prices to online news to extract useful information from diverse global online news relevant to individual stocks for multiple forecasting tasks. Our extensive experiments with multiple datasets show that GLAM outperforms other state-of-the-art models on multiple forecasting tasks and investment and risk management application case-studies.

Research on Stock Return Forecasting Methods based on Time Series Models

Accurately predicting the trend of stock return rate is a hot research issue. With the development of artificial intelligence, machine learning, big data and other technologies, it brings new potential to the prediction of the stock market. In order to accurately predict the trend of stock return, this paper mainly constructs the time series ARMA model and random forest model, uses the stacking method to fuse the models, and predicts the daily return of Yangtze River Electric Power stock. The final fusion model has an MSE of 1.757 on the training set and 1.274 on the test set. The overall prediction error of the model is within an acceptable range. At the same time, the fused model can weaken the problem of underfitting of a single model, which provides a valuable reference for model optimization research.

Two-Stage Portfolio Optimization Integrating Optimal Sharp Ratio Measure and Ensemble Learning

The traditional portfolio theory has relied heavily on historical asset returns while ignoring future information. Based on ensemble learning and maximum Sharpe ratio portfolio theory, this paper proposes a two-stage portfolio optimization method by considering asset forecast information, aiming to improve the performance and robustness of a portfolio. In the first stage, concerning the underlying asset selection, we integrate six individual prediction models using the ensemble learning method to forecast the future return of assets where the assets with higher potential returns are selected for portfolio optimization. In the second stage, we propose a novel investment strategy by combining the forecasted returns of selected assets with the maximum Sharpe ratio portfolio model. In the empirical analysis, we employ the constituent stocks of the China Securities Index 300 (CSI 300) to test the out-of-sample performance of the proposed strategy with several traditional portfolio strategies, including minimum variance portfolio strategy, traditional maximum Sharpe ratio portfolio strategy, 1/N portfolio strategy, and CSI 300 index. Our analytical results show that (1) compared to individual forecasting models, the ensemble learning method is more accurate in forecasting stock returns, and (2) the proposed portfolio strategy largely outperforms most of its competitors in terms of the Sharpe ratio, Sortino ratio, Omega ratio, and Calmar ratio. This indicates that the proposed two-stage portfolio optimization method is of potential to construct a promising investment strategy due to its trade-off between historical and future information of assets.

Forecasting Stock Return Volatility Using the Realized Garch Model and an Artificial Neural Network

Forecasting Stock Return Volatility Using the Realized Garch Model and an Artificial Neural Network