2023COSC470Spike

Algorithmic Trading and Short-term Forecast for Financial Time Series with Machine Learning Models Data Warehouse

Introduction

A profitable algorithmic stock trading algorithm will benefit from a forecasting system that can produce accurate short-term forecasts. You will help to build up short-term forecasting models using machine learning (ML) models. The project aims to develop effective algorithmic trading algorithms based on accurate short-term forecasts for financial time series using machine learning models.

To achieve this objective, the project is focused on three sets of activities:

In the guided COSC 470 Spike 5-week project, you will build a data warehouse prototype containing the targeted financial time series and other time series that are considered useful as input to the machine learning models. Automated data acquisition through web-scrapping and APIs, processing, and staging routines required to feed the data warehouse will be evaluated and developed dynamically. A data visualization and business intelligence layer is also built on the data warehouse. A short-term forecasting model based on machine learning will be developed with external researchers and another team at Langara College using the various time series data from the data warehouse. Machine learning algorithms such as neural networks, random forecast, support vector regression, XGBoost, and long short-term memory will be evaluated with several performance criteria to identify the most accurate model from a short-term trading perspective. After having been evaluated in themselves, the stock-value prediction models should be applied to simple trading strategies and thus measured for the improvement they bring to those strategies.

You will use two data sources: Yahoo Finance and the following data provider through the API https://site.financialmodelingprep.com/developer/docs/.

It would be nice to monitor changes in the companies' stocks, but it can be done later in the main project.

Deliverables: (1) Source Code; (2) Working Application(s); (3) Design documentation; (4) A technical report in the form of a research paper submitted to https://arxiv.org (LaTeX).

Clients: Paris East University, Paris, France and Langara College, BC, Canada.