Exportar Publicação

Costa, A., Ramos, F.R., Mendes, V. & Mendes, V. (2019). Forecasting financial time series using deep learning techniques. IO2019 - XX Congresso da APDIO 2019.

A. R. Costa et al.,  "Forecasting financial time series using deep learning techniques", in IO2019 - XX Congr.o da APDIO 2019, Tomar, 2019

@misc{costa2019_1714820354291,
	author = "Costa, A. and Ramos, F.R. and Mendes, V. and Mendes, V.",
	title = "Forecasting financial time series using deep learning techniques",
	year = "2019",
	howpublished = "Ambos (impresso e digital)",
	url = "https://ciencia.iscte-iul.pt/publications/forecasting-financial-time-series-using-deep-learning-techniques/63229?lang=en"
}

TY  - CPAPER
TI  - Forecasting financial time series using deep learning techniques
T2  - IO2019 - XX Congresso da APDIO 2019
AU  - Costa, A.
AU  - Ramos, F.R.
AU  - Mendes, V.
AU  - Mendes, V.
PY  - 2019
CY  - Tomar
UR  - https://ciencia.iscte-iul.pt/publications/forecasting-financial-time-series-using-deep-learning-techniques/63229?lang=en
AB  - In this article, we propose the use of recurrent neural networks based on long short-term memory (LSTM) architecture in order to forecast financial time series. Artificial neural networks have proven to be efficient in forecasting financial time series. In particular, recurrent neural networks have been able to store past inputs to produce the currently desired output, which justifies their application in financial time series prediction. First, we study the main features of the Standard and Poor’s 500 index, S&P500, such as the linearity, stationarity, descriptive statistics, Hurst exponents, among others. Secondly, we train several types of recurrent neural networks for the S&P500 index and use the models to make short-term forecasts. Finally, we compare the out-of-sample forecast error (MAE) for the employed models, in order to conclude about the forecasting performance
ER  -