This study reconsiders the timing of the equity risk factors by using the flexible neural networks specified for image recognition to determine the timing weights. The performance of each factor is visualized to be standardized price and volatility charts and `learned' by flexible image recognition methods with timing weights as outputs. The performance of all groups of factors can be significantly improved by using these ``deep learning--based'' timing weights. In addition, visualizing the volatility of factors and using deep learning methods to predict volatility can significantly improve the performance of the volatility-managed portfolio for most categories of factors. Our further investigation reveals that the timing success of our method hinges on its ability in identifying ex ante regime switches such as jumps and crashes of the factors and its predictability on future macroeconomic risk.

Developing an early warning model for company financial crises holds critical significance in robust risk management and ensuring the enduring stability of the capital market. Although the existing research has achieved rich results, the disadvantages of insufficient text information mining and poor model performance still exist. To alleviate the problem of insufficient text information mining, we collect related financial and annual report data from 820 listed companies in mainland China from 2018 to 2023 by using sophisticated web crawlers and advanced text sentiment analysis technologies and using missing value interpolation, standardization, and data balancing to build multi-source datasets of companies. Ranking the feature importance of multi-source data promotes understanding the formation of financial crises for companies. In the meantime, a novel Attention-based Fuzzy Neural Network (AFNN) was proposed to parse multi-source data to forecast financial crises among listed companies. Experimental results indicate that AFNN exhibits significantly improved performance compared to other advanced methods.

We propose a novel measure, DOT, to capture belief divergence on extreme tail events in stock returns. Defined as the standard deviation of expected probability forecasts generated by distinct information processing functions and neural network models, DOT exhibits significant predictive power for future stock returns. A value-weighted (equal-weighted) long-short portfolio based on DOT yields an average return of -1.07% (-0.98%) per month. Furthermore, we document novel evidence supporting a risk-sharing channel underlying the negative relation between DOT and the equity premium following extreme negative shocks. Finally, our findings are also in line with a mispricing channel in normal periods.

The healthy operation of the stock market relies on sound liquidity. We utilize the semantic information from disclosure texts of listed companies on the China Science and Technology Innovation Board (STAR Market) to construct a daily corporate network. Through empirical tests and performance analyses of machine learning models, we elucidate the relationship between the similarity of company disclosure text contents and the temporal and spatial correlations of stock liquidity. Our liquidity indicators encompass trading costs, market depth, trading speed, and price impact, recognized across four dimensions. Furthermore, we reveal that the information loss caused by employing Minimum Spanning Tree (MST) topology significantly affects the explanatory power of network topology indicators for stock liquidity, with a more pronounced impact observed at the document level. Subsequently, by establishing a neural network model to predict next-day liquidity indicators, we demonstrate the temporal relationship of stock liquidity. We model a liquidity predicting task and train a daily liquidity prediction model incorporating Graph Convolutional Network (GCN) modules to solve it. Compared to models with the same parameter structure containing only fully connected layers, the GCN prediction model, which leverages company network structure information, exhibits stronger performance and faster convergence. We provide new insights for research on company disclosure and capital market liquidity.

This paper proposes a new forecasting framework for the stock market that combines machine learning algorithms with several technical analyses. The paper considers three different algorithms: the Random Forests (RF), the Gradient-boosted Trees (GBT), and the Deep Neural Networks (DNN), and performs forecasting tasks and statistical arbitrage strategies. The portfolio weight optimization strategy is also proposed to capture the model's return and risk information from output probabilities. The paper then uses the stock data in the Chinese A-share market from January 1, 2011, to December 31, 2020, and observes that all three machine learning models achieve significant returns in the Chinese stock market. The DNN achieves an average daily return of 0.78% before transaction costs, outperforming the 0.58% of the RF and 0.48% of the GBT, far exceeding the general market level. The performance of the weighted portfolio based on the ESG score is also improved in all three machine learning strategies compared to the equally weighted portfolio. These results help bridge the gap between academic research and professional investments and offer practical implications for financial asset pricing modelling and corporate investment decisions.

A deep learning method is applied to predict stock portfolio allocation in the Chinese stock market. We use 6 original price and volume series as benchmark model settings and further explore the model's predictive performance with social media sentiment. Our results show that our model can achieve a high out-of-sample Sharp ratio and annual return. Moreover, social media sentiment could increase the performance for both Sharp ratio and annual return while reducing annual volatility. We provide an end-to-end stock portfolio allocation model based on deep neural networks.

Stock market prediction provides the decision-making ability to the different stockholders for their investments. Recently, stock technical indicators (STI) emerged as a vital analysis tool for predicting high-dimensional stock data in various studies. However, the prediction performance and error rate still face limitations due to the lack of correlational analysis between STI and stock movement. This paper proposes a correlational strategy to overcome these challenges by analyzing the correlation of STI with stock movement using neural networks with the feature vector. This strategy adopts the Pearson coefficient to analyze STI and close index of stock data from 8 Chinese companies in the Hong Kong stock market. The results reveal the price prediction of BiLSTM outperformed the GRU and LSTM in various datasets and prior studies.

In this paper we use artificial neural network (ANN) to empirically analyze the liquidity values of the non-floating shares and the influencing factors to China’s stock market in the background of China’s listed companies split share stricture reform. We try to use a proportion which the company’s non-floating shareholders offer compensation to the floating shareholders to test the liquidity values of the non-floating shares and use MATLAP establish a feed-forward BP neural network model to analyze and forecast according to the data of the companies which have announced and actualized their split stricture reform plans. In expansion analysis, we use the perturbation method to measure the influence of these parameters on the liquidity values of the non-floating. As result, the character of the shares, the share structure and the ratio of the shares by the principal shareholder held are the main influencing factors.

It is often documented, based on autocorrelation, variance ratio and power spectrum, that exchange rates approximately follow a martingale process. Because autocorrelation, variance ratio and spectrum check serial uncorrelatedness rather than martingale difference, they may deliver misleading conclusions in favor of the martingale hypothesis when the test statistics are insigniÞcant. In this paper, we explore whether there exists a gap between serial uncorrelatedness and martingale difference for exchange rate changes, and if so, whether nonlinear time series models admissible in the gap can outperform the martingale model in out-of-sample forecasts. Applying the generalized spectral tests of Hong (1999) to Þve major currencies, we Þnd that the changes of exchange rates are often serially uncorrelated, but there exists strong nonlinearity in conditional mean, in addition to the well-known volatility clustering. To forecast the conditional mean, we consider the linear autoregressive, autoregressive polynomial, artiÞcial neural network and functional-coefficient models, as well as their combination. The functional coefficient model allows the autoregressive coefficients to depend on investment positions via an moving average technical trading rule. We evaluate out-of-sample forecasts of these models relative to the martingale model, using four criteria– the mean squared forecast error, the mean absolute forecast error, the mean forecast trading return, and the mean correct forecast direction. White’s (2000) reality check method is used to avoid data-snooping bias. It is found that suitable nonlinear models, particularly their combination, do have superior predictive ability over the martingale model for some currencies in terms of certain forecast evaluation criteria.