“…Because profits are not proportional to the performance of the classification or regression, we use a simple trading strategy named long/short strategy, as suggested by Zhou et al [47] , to examine the profitability of our proposed framework. Fig.…”
Section: Our Proposed Frameworkmentioning
confidence: 99%
“…Based on the predicted value from the LSTM-Attention model, this strategy introduces a buy-threshold and a sell-threshold to decide whether to change the position from to 1, or from 1 to , or just to keep the position. Following Zhou et al [47] , in our experiments without considering transaction costs, the values of the buy-threshold and sell-threshold are set as 0.50. Fig.…”
Section: Our Proposed Frameworkmentioning
confidence: 99%
“…Although these models have advantages in theoretically describing the underlying data generating process based on statistical logic, they contain some strong assumptions about the noise terms (such as that variables are independent and identically distributed, or follow a t -distribution) or explanatory variables (such as stationarity or exogeneity) that are not fully satisfied in the real world. On the other hand, many machine learning models have been successfully developed over recent decades, without imposing restrictive assumptions, to learn from and forecast financial time series, such as the support vector machine (SVM) [34] , gradient boosted decision trees (GBDTs) [47] , neural networks (NNs) [1] , [45] , and the ensemble model by cascading logistic regression (LR) onto GBDT [47] . However, one key limitation of most existing machine learning models is their lack of an explicitly declared mechanism to handle nonlinearity and non-stationarity in time series [48] , which may lead to inaccurate predictions.…”
The recent availability of enormous amounts of both data and computing power has created new opportunities for predictive modeling. This paper compiles an analytical framework based on multiple sources of data including daily trading data, online news, derivative technical indicators, and time-frequency features decomposed from closing prices. We also provide a real-life demonstration of how to combine and capitalize on all available information to predict the stock price of BGI Genomics. Moreover, we apply a long short-term memory (LSTM) network equipped with an attention mechanism to identify long-term temporal dependencies and adaptively highlight key features. We further examine the learning capabilities of the network for specific tasks, including forecasting the next day’s price direction and closing price and developing trading strategies, comparing its statistical accuracy and trading performance with those of methods based on logistic regression, support vector machine,gradient boosting decision trees, and the original LSTM model. The experimental results for BGI Genomics demonstrate that the attention enhanced LSTM model remarkably improves prediction performance through multi-source heterogeneous information fusion, highlighting the significance of online news and time-frequency features, as well as exemplifying and validating our proposed framework.
“…Because profits are not proportional to the performance of the classification or regression, we use a simple trading strategy named long/short strategy, as suggested by Zhou et al [47] , to examine the profitability of our proposed framework. Fig.…”
Section: Our Proposed Frameworkmentioning
confidence: 99%
“…Based on the predicted value from the LSTM-Attention model, this strategy introduces a buy-threshold and a sell-threshold to decide whether to change the position from to 1, or from 1 to , or just to keep the position. Following Zhou et al [47] , in our experiments without considering transaction costs, the values of the buy-threshold and sell-threshold are set as 0.50. Fig.…”
Section: Our Proposed Frameworkmentioning
confidence: 99%
“…Although these models have advantages in theoretically describing the underlying data generating process based on statistical logic, they contain some strong assumptions about the noise terms (such as that variables are independent and identically distributed, or follow a t -distribution) or explanatory variables (such as stationarity or exogeneity) that are not fully satisfied in the real world. On the other hand, many machine learning models have been successfully developed over recent decades, without imposing restrictive assumptions, to learn from and forecast financial time series, such as the support vector machine (SVM) [34] , gradient boosted decision trees (GBDTs) [47] , neural networks (NNs) [1] , [45] , and the ensemble model by cascading logistic regression (LR) onto GBDT [47] . However, one key limitation of most existing machine learning models is their lack of an explicitly declared mechanism to handle nonlinearity and non-stationarity in time series [48] , which may lead to inaccurate predictions.…”
The recent availability of enormous amounts of both data and computing power has created new opportunities for predictive modeling. This paper compiles an analytical framework based on multiple sources of data including daily trading data, online news, derivative technical indicators, and time-frequency features decomposed from closing prices. We also provide a real-life demonstration of how to combine and capitalize on all available information to predict the stock price of BGI Genomics. Moreover, we apply a long short-term memory (LSTM) network equipped with an attention mechanism to identify long-term temporal dependencies and adaptively highlight key features. We further examine the learning capabilities of the network for specific tasks, including forecasting the next day’s price direction and closing price and developing trading strategies, comparing its statistical accuracy and trading performance with those of methods based on logistic regression, support vector machine,gradient boosting decision trees, and the original LSTM model. The experimental results for BGI Genomics demonstrate that the attention enhanced LSTM model remarkably improves prediction performance through multi-source heterogeneous information fusion, highlighting the significance of online news and time-frequency features, as well as exemplifying and validating our proposed framework.
“…Experiments on Indian stock market demonstrated the effectiveness of the fusion prediction models. Zhou et al [30] developed a learning architecture by cascading the logistic regression model onto the GBDT for predicting the stock indices. Cao and Wang [13] established a hybrid prediction model, which consists of CNN and SVM, to make stock market predictions.…”
The prediction of stock price movement direction is significant in financial studies. In recent years, a number of deep learning models have gradually been applied for stock predictions. This paper presents a deep learning framework to predict price movement direction based on historical information in financial time series. The framework combines a convolutional neural network (CNN) for feature extraction and a long short-term memory (LSTM) network for prediction. We specifically use a three-dimensional CNN for data input in the framework, including the information on time series, technical indicators, and the correlation between stock indices. And in the three-dimensional input tensor, the technical indicators are converted into deterministic trend signals and the stock indices are ranked by Pearson product-moment correlation coefficient (PPMCC). When training, a fully connected network is used to drive the CNN to learn a feature vector, which acts as the input of concatenated LSTM. After both the CNN and the LSTM are trained well, they are finally used for prediction in the testing set. The experimental results demonstrate that the framework outperforms state-of-the-art models in predicting stock price movement direction.
“…LRMs are some of the most classic and commonly used methods, which have the advantages of minimal computation, high detection speed, and good adaptability [11]. LRMs have extensive applications in forecasting [12,13] and susceptibility assessment [14,15]. These studies provide a solid foundation for the susceptibility assessment of landslides, and advance the knowledge of susceptibility assessment with machine learning algorithms.…”
Digital elevation models (DEMs) are fundamental data models used for susceptibility assessment of landslides. Due to landscape change and reshaping processes, a DEM can show obvious temporal variation and has a significant influence on assessment results. To explore the impact of DEM temporal variation on hazard susceptibility, the southern area of Sichuan province in China is selected as a study area. Multitemporal DEM data spanning over 17 years are collected and the topographic variation of the landscape in this area is investigated. Multitemporal susceptibility maps of landslides are subsequently generated using the widely accepted logistic regression model (LRM). A positive correlation between the topographic variation and landslide susceptibility that was supported by previous studies is quantitatively verified. The ratio of the number of landslides to the susceptibility level areas (RNA) in which the hazards occur is introduced. The RNA demonstrates a general decrease in the susceptibility level from 2000 to 2009, while the ratio of the decreased level is more than fifteen times greater than that of the ratio of the increased level. The impact of the multitemporal DEM on susceptibility mapping is demonstrated to be significant. As such, susceptibility assessments should use DEM data at the time of study.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
