Integrating principle component analysis and weighted support vector machine for stock trading signals prediction

Portfolio formation with preselection using deep learning from long-term financial data

“…In recent years, with the development of big data and artificial intelligence (AI) technology, more and more scholars start to use AI as support for their research solutions and prove that AI methods deal with problem of nonlinear, nonstationary characteristics better than traditional statistical models. For example, a number of researches based on SVM (Paiva et al, 2019), PCA (Chen and Hao, 2018;Zbikowski, 2015), GA or random forest Mousavi, 2014), ANN (Patel et al, 2015;Chong et al, 2017) to classify, predict and optimise complex financial assets. Among these technologies, the deep learning is thought to be an appropriate method for the financial time-series forecasting solution, since it is good at processing complex, high-dimensional data as well as extracting abstract characteristics from mass data without depending on any assumptions.…”

“…In this research, we collect daily stock data from the UK Stock Exchange 100 Index (FTSE 100) from March 1994 until March 2019, covering 25 years. Since the majority of related studies have been conducted over a period of 10 years or less (Kara et al, 2011;Patel et al, 2015;Chen and Hao, 2018), 15 years (Paiva et al, 2019; Almahdi and Yang; 2017), or 25 years (Fischer and Krauss, 2018), our samples spanning 25 years can be considered to provide a sufficiently large volume data to generate statistically significant results. Our sample data involves the historical series of adjusted open prices, close prices, the highest prices, the lowest prices, and the trading volume of assets.…”

“…In the light of previous literatures, technical indicators are effective features to describe and reflect the real market situation. For instance, Chen and Hao (2018) Also, financial time-series forecasting is always explained by the lagged observations. For example,…”

“…Therefore, incorporating forecasting theory into the portfolio formation will be promising in financial investment (Kolm et al, 2014). Forecasting financial time-series is always regarded as one of the most challenging tasks because of the dynamic, nonlinear, unstable and complex nature with longterm fluctuations of the financial market (Chen and Hao, 2018;Paiva et al, 2019). But a reliable investment decision should rely on long-term observations and patterns of behaviour of asset data rather than short-term (Chourmouziadis and Chatzoglou 2016;Chong et al, 2017).…”

Optimal Prediction of Foreign Exchange Market Exploring the Effect of Water Cycle Algorithm Strategy on Data Segregation

“…Hence, the data is not of fixed pattern. Beforehand factual methodologies, for example, moving normal, weighted moving normal, autoregressive moving normal (ARMA), autoregressive coordinated moving normal (ARIMA) have been utilized to forecast trend [3]. But those techniques were not capable enough to grab the non linear and dynamic nature of forex rates for the prediction task as they were working on the assumption that data are co related and linear in nature.…”