News-driven commodity trading

The Hong Kong University of Science and Technology 
Department of Computer Science and Engineering 

Final Year Thesis Oral Presentation

Title: "News-driven commodity trading"

by 

Mr. Cui WANG


Abstract:

Traditional trading systems adopt strategies like mean-reversion, 
momentum and arbitraging which are all based on market inefficiency. 
News-driven trading, in contrast, become more effective with more market 
participants because of the effect to reflexity raised by George 
Soros.

Unlike many existing news-driven trading strategies which trade 
equities, this thesis examines news-driven trading on crude oil futures 
contracts, because commodities trend better, further and more often, with 
recurring news coming out every week.

The whole system starts from reading news feeds and labelling them as 
positive, negative and neutral. Three different labelling methods are 
examined on a random sample.

Combining labelled news and price data, the strategy will make decisions 
of going long and short. Three different strategies are examined, and one 
of them is further tested using a large data set.

The resulting strategy makes 0.78% per 15-minute trade. Assume one trade 
per week, the annual return on investment is 40.56%. 3 out of 4 trades 
make money, which indicates the strong predictive power of news feeds.

  
Date                 : 26 April 2016 (Tuesday)

Time                 : 4:30pm to 5:30pm

Venue                : Room 5506 (lift 25/26)

Advisor              : Prof. Huamin QU
                              
2nd Reader           : Dr. Xiaojuan MA