News-driven stock prediction investigates the correlation between news events and stock price movements.
To improve the prediction of share price for aerospace industry sector and well understand the impact of various indicators on stock prices, we provided a hybrid prediction model by the combination of Principal Component Analysis (PCA) and Recurrent Neural Networks.
We adopt BERT with multitask learning which additionally predicts the worthiness of the news and propose a metric called Polarity-Over-Time to extract the word polarity among different event periods.
We consider the problem with history-dependent experts, in which each expert uses the previous $d$ days of history of the market in making their predictions.
Compared to other recent applications of partial differential equations to prediction, ours has a new element: there are two timescales, since the recent history changes at every step whereas regret accumulates more slowly.
This paper proposes to use sentiment analysis to extract useful information from multiple textual data sources and a blending ensemble deep learning model to predict future stock movement.
Bayesian Networks (BN) are probabilistic graphical models that are widely used for uncertainty modeling, stochastic prediction and probabilistic inference.
However, it is well known that an individual stock price is correlated with prices of other stocks in complex ways.
Thanks to the use of attention over news events, our model is also more explainable.
We present a method to automatically identify financially relevant news using stock price movements and news headlines as input.