| Recently, financial markets have experienced a shift from the traditional way of trading using human brokers to algorithmic trading. Trading algorithms proved to be more efficient than conventional approaches as they provide for lower latency, larger volume, and higher market coverage degree. During the last years, it has been acknowledged the need to use news information to help trading algorithms make better decisions. The biggest challenge is to allow machines to identify and use the news information that is relevant for technical trading. The research question addressed in this proposal is how to identify financial events in news and use this information for improving the returns generated by algorithmic trading? Due to the multidisciplinary nature of the project our research is positioned at the confluence of three fields: text mining, the Semantic Web, and finance. First, we specify in an ontology the financial events and their contextual information. Then, by applying text mining techniques, we discover instances of the ontology events in news items. Last, we investigate the use of the recognized financial events as an additional input aiming to improve the returns generated by algorithmic trading. To our knowledge the recognition and usage of financial events captured in news items for algorithmic trading has not been thoroughly explored. The innovative aspects that play a key role here are: developing a financial ontology for algorithmic trading, devising powerful lexico-semantic rules for recognizing financial events in news, and using the financial events to improve the returns generated by trading algorithms. |
