Search

Predicting stock market trends is the basic daily routine task that investors should perform in the stock trading market. Traditional market trends prediction models are generally based on hand-crafted factors or features, which heavily rely on expensive expertise knowledge. Moreover, it is difficult to discover hidden features contained in the stock time series data, which are otherwise helpful for predicting stock market trends. In this paper, we propose a novel stock market trends prediction framework SMART with a self-supervised stock technical data sequence embedding model S3E. Specifically, the model encodes stock technical data sequences into embeddings, which are further trained with multiple self-supervised auxiliary tasks. With the learned sequence embeddings, we make stock market trends predictions based on an LSTM and a feed-forward neural network. We conduct extensive experiments on China A-Shares market and NASDAQ market to show that our model is highly effective for stock market trends prediction. We further deploy SMART in a leading financial service provider in China and the result demonstrates the effectiveness of the proposed method in real-world applications. [ABSTRACT FROM AUTHOR]

The stock price prediction is an important issue in stock markets because it will result in significant benefits and impacts for investor. In contrast to traditional time series, fuzzy time series can solve the forecast problem with historical data of linguistic values. In order to improve forecast performance of fuzzy time-series models, this study replaced fuzzy logical relationships with rule-based algorithm to extract forecast rules from time-series observations. Therefore, this paper developed a novel fuzzy time-series model based on rough set rule induction for forecasting stock index, and this study has four contributions to improve forecast accuracy and provide investment point (in right time) to investors: (1) Proposed a novel fuzzy time-series model to improve forecast accuracy, (2) rough sets are employed to generate forecasting rules to replace fuzzy logical relationship rules based on the lag period, (3) utilized adaptive expectation model to strengthen forecasting performance, and based on the meaning of adaptive parameter to observe stock fluctuation and oscillation, and (4) proposed buy and sell rules to calculate the profit and based on three different scenarios to provide investment suggestion to investor as references. For evaluating the proposed model, we practically collected TAIEX, Nikkei, and HSI stock price from 1998 to 2012 years as experimental dataset, and compared the listing models under three error indexes and profits criteria. The results show that the proposed method outperforms listing models in error indexes and profits. [ABSTRACT FROM AUTHOR]

The lack of regulations and disclosures regarding intangible capital has made it rather difficult for investors and creditors to evaluate a firm׳s intangible value before making the associated investment and loan decisions. This study represents an initial attempt to compare/contrast different types of machine learning techniques and identify the optimal prediction model for intangible assets. In addition, this paper shows that machine learning can be used effectively for the problem of intangible assets evaluation. To be specific, five classification algorithms are considered: decision trees (DT), artificial neural networks (ANN), naïve Bayes, support vector machines (SVM) and k -Nearest Neighbors ( k -NN). Consequently, thirty prediction models are constructed for comparison, including five single classifiers, boosting and bagging based classifier ensembles, and the combination of k -means clustering, single classifiers and classifier ensembles. The experimental results show that prediction models combining k -means with boosting/bagging based classifier ensembles perform much better than the other methods in terms of prediction accuracy, ROC Curve, as well as Type I and II errors. In particular, while the best single classifier, k -NN provides 78.24% prediction accuracy, k -means+bagging based DT ensembles provide the best performance to predict intangible assets with a prediction accuracy of 91.60%, 96.40% of ROC Curve and 18.65% of Type I and 6.34% of II errors, respectively. [ABSTRACT FROM AUTHOR]

The interaction between stock price process and market news has been widely analyzed by investors on different markets. Previous works, however, focus either on market news purely as exogenous factors that tend to lead price process or on the analysis of how past stock price process can affect future stock returns. To take a step forward, we quantitatively integrate information from both market news and stock prices in order to improve the accuracy of prediction on stock future price return in an intra-day trading context. In this paper, we present the design and architecture of our approach for market information fusion. By means of multiple kernel learning, the hidden information behind the two sources is effectively extracted, and more importantly, seamlessly integrated rather than simply combined by a single kernel approach. Experiments on comprehensive comparisons between our approach and three baseline methods (which use only one type of information, or naively combine the two sources) have been conducted on the intra-day tick-by-tick data of the Hong Kong Stock Exchange and market news archives of the same period. It has been shown that for both cross-validation and independent testing, our approach is able to achieve the best results. [ABSTRACT FROM AUTHOR]

Abstract: In this paper, we present a cellular learning automata model for the investment behavior in the stock market. In this model, investors decide to hold, buy, or sell the stocks based on the evolution rules and learn how much they can trust on other investors based on the learning rules. We analyze the effects of imitation, reliability and macrofactors on the stock market and compare the obtained results with the previous approach that is based on cellular automata. [Copyright &y& Elsevier]