The use of machine learning techniques to predict the next-day stock direction is established. To make prediction models more robust, a common approach is to combine historical time series and news sentiment analysis. Most of the trading simulations performed in this field rely on trend following strategies, which are aimed at identifying and following an ongoing price trend that is likely to persist in the next days. Conversely, a more limited effort has been devoted to applying machine learning techniques to predict trend reversal, i.e., changes in price directions. This paper investigates the relevance of news information and time series descriptors derived from technical analysis to predict trend reversal in the next days. It compares the performance of various classification models trained on (i) technical indicators, which indicate short-term overbought or oversold conditions, (ii) news sentiment descriptors, which express the opinion of the financial community, (iii) the historical time series, to highlight recurrences in price trends, and (iv) a combination of the above. The results achieved on an 11- year dataset related to the stocks of the U.S. S&P 500 index show that the strategies combining the historical values of news sentiment and stock price indicators averagely perform better than all the other tested combinations. Hence, news information is worth considering by trend reversal strategies.
Combining news sentiment and technical analysis to predict stock trend reversal / Cagliero, Luca; Attanasio, Giuseppe; Garza, Paolo; Baralis, ELENA MARIA. - ELETTRONICO. - (2019), pp. 514-521. ((Intervento presentato al convegno 9th ICDM Workshop on Sentiment Elicitation from Natural Text for Information Retrieval and Extraction tenutosi a Beijing, Cina nel November 8, 2019.
Titolo: | Combining news sentiment and technical analysis to predict stock trend reversal |
Autori: | |
Autori: | Cagliero, Luca; Attanasio, Giuseppe; Garza, Paolo; Baralis, ELENA MARIA |
Numero degli autori: | 4 |
Data di pubblicazione: | 2019 |
Presenza coautori internazionali: | no |
Nome del convegno: | 9th ICDM Workshop on Sentiment Elicitation from Natural Text for Information Retrieval and Extraction |
Luogo del convegno: | Beijing, Cina |
Anno del convegno: | November 8, 2019 |
Pagina iniziale: | 514 |
Pagina finale: | 521 |
Numero di pagine: | 8 |
Supporto: | ELETTRONICO |
Lingua: | Inglese |
Parole Chiave: | news sentiment analysis, quantitative trading, classification, trend reversal prediction |
Rilevanza: | Internazionale |
Abstract: | The use of machine learning techniques to predict the next-day stock direction is established. To... make prediction models more robust, a common approach is to combine historical time series and news sentiment analysis. Most of the trading simulations performed in this field rely on trend following strategies, which are aimed at identifying and following an ongoing price trend that is likely to persist in the next days. Conversely, a more limited effort has been devoted to applying machine learning techniques to predict trend reversal, i.e., changes in price directions. This paper investigates the relevance of news information and time series descriptors derived from technical analysis to predict trend reversal in the next days. It compares the performance of various classification models trained on (i) technical indicators, which indicate short-term overbought or oversold conditions, (ii) news sentiment descriptors, which express the opinion of the financial community, (iii) the historical time series, to highlight recurrences in price trends, and (iv) a combination of the above. The results achieved on an 11- year dataset related to the stocks of the U.S. S&P 500 index show that the strategies combining the historical values of news sentiment and stock price indicators averagely perform better than all the other tested combinations. Hence, news information is worth considering by trend reversal strategies. |
URL: | https://ieeexplore.ieee.org/abstract/document/8955650 |
Revisione (peer review): | Esperti anonimi |
Digital Object Identifier (DOI): | http://dx.doi.org/10.1109/ICDMW.2019.00079 |
Codice identificativo Scopus: | 2-s2.0-85078806684 |
Tesi di Dottorato in Cotutela: | no |
Editore: | IEEE Computer Society Press |
Titolo del libro: | IEEE ICDM 2019 Workshops Proceedings |
Sezione: | contributo |
Data di presentazione: | 2020-01-22T10:42:46Z |
Appare nelle tipologie: | 4.1 Contributo in Atti di convegno |
File in questo prodotto:
File | Descrizione | Tipologia | Licenza | |
---|---|---|---|---|
postprint_editoriale.pdf | Articolo principale (postprint editoriale) | 2a Post-print versione editoriale / Version of Record | Non Pubblico - Accesso privato/ristretto | Administrator Richiedi una copia |
cameraready_SENTIREWORKSHOP_ICDM2019.pdf | Articolo principale (postprint) | 2. Post-print / Author's Accepted Manuscript | PUBBLICO - Tutti i diritti riservati | Visibile a tuttiVisualizza/Apri |
http://hdl.handle.net/11583/2753472