Financial time series data pattern detection, forecasting and its application

This paper studies the latest techniques for financial time series forecasting by extending the existing work. In addition to historical stock data, sentiment analysis and signal analysis methods are applied to simulate the real-world factors that could potentially affect the stock trends. Three...

Full description

Saved in:
Bibliographic Details
Main Author: Ooi, Yuxuan
Other Authors: Loke Yuan Ren
Format: Final Year Project
Language:English
Published: Nanyang Technological University 2021
Subjects:
Online Access:https://hdl.handle.net/10356/153503
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:This paper studies the latest techniques for financial time series forecasting by extending the existing work. In addition to historical stock data, sentiment analysis and signal analysis methods are applied to simulate the real-world factors that could potentially affect the stock trends. Three LSTM-based models with varied input features and architectures were trained and tested with different popular tech stocks. The experiment result shows that adding a new dimension of public sentiment helps to improve the prediction model to forecast a closing price trend that follows closely to the actual price. Furthermore, this paper proposes a trading platform that applies the prediction model built as a real-world use case. A trading algorithm is proposed to utilize the forecasted results to provide an auto-trading service and serves as the core service of the platform. The platform comes in the form of mobile application and is equipped with useful functionalities with the goal of capturing the market.