Project Reference: CS74

Student’s Name: Minglong Shen

Project title: Automated Trading in Stock Market

Course Title: Computer Science

Supervisor’s Name: Shadi Basurra

The project aims to create an automated algorithm bot to predict the stock market trend using TrendSpider platform parameters, option flow data, and using machine learning to synergise the two data to predict the trend’s overall performance accurately. sentiment analysis technique to analyse the latest news headline

The stock market is one of the fastest ways to make money; billions, trillions of dollars have flowed in the stock market every day. High rewards also come with high risks. Many factors can be determining the stock price movement, economic status, political factors, and companies own decisions, so highly accurate data is essential. Using the data to precisely predict the financial market is beneficial for decision-making purposes; therefore, a well-made prediction method is needed to predict market fluctuations. I explore the predictive power of sentiment analysis and time series analysis on the market based on forecast financial news sentiment and seasonality trends to predict the market price performance. I learned that news sentiment could predict the short term trend reasonably accurate because everyone can see it and drive up or down the price based on how good or bad the news is. However, using time series forecasting to predict the market is impossible. There are three main factors when using time series to forecast something, 1 “”how well we understand factors that contribute to it;”” 2 “”How much data is available”” 3 “”whether the forecast can affect the thing we are trying to forecast””. For example, if we were going to forecast flight ticket sales, it would be very accurate because all conditions are met. We know that flight tickets are a great sale in holiday seasons and provide a decent amount of historical data on flight tickets demand. We can develop a good model linking ticket sale demand and key driver variables. On the other hand, when forecasting the stock market, only one condition is met; there is a good amount of available data. However, we cannot use the historical data to predict the market because it does not have a seasonality (repeating trends); they are more of a cyclical trend, there are no trends with a set of repetition. Meaning there is no difference to toss a coin and try to guess it is a head or a tail.

A business opportunity of selling the service to retailer trader in the stock trading community.