Homepage of Garnett Wilson

Ph.D., Computer Science

Adjunct Assistant Professor,
Faculty of Computer Science, Dalhousie University

Ph.D. Supervisory Committee:
Research Aptitude Defense: Alexander Loginov, "On Increasing The Scope of Genetic Programming Trading Agents"
May 2015

Faculty of Computer Science, Dalhousie University, www.cs.dal.ca

Abstract.  The goal of this study is to investigate the potential for widening the scope of decisions made by an automated trading agent beyond just the construction of decision trees for expressing buy, hold, sell decisions. To do so, both technical indicators and decision trees are co-evolved under the machine learning paradigm of genetic programming. We show how decision trees can make use of Stop-Loss and Take-Profit orders with retracement levels. Moreover, given that potential users are generally interested in the collective performance of a portfolio of stocks, the optimization of the weightings associated with different stocks is also addressed. The resulting approach constructs multiple trading agents that act as an ensemble to make trading decisions under minute-to-minute candlestick data. On the one hand, this makes the task significantly more difficult than discovering efficient trading policies from periods of longer duration (say, 1 hour intervals) because there is a lot more market noise in the price information. Conversely, the shorter duration between consecutive prices means that local (or short term) trends can be detected earlier, potentially resulting in an agent being able to make use of trends before other agents. A benchmarking study considers the effectiveness of the approach under a portfolio of stocks from the NASDAQ using performance criteria expressing profitability, drawdown, hit rate and reward to risk ratio. The computational cost of the framework is also considered, with the evolutionary cycle to identify trading agents for each stock shown to take less than 7 seconds on a regular desktop computer.

Master's Supervisory Committee:
Master's Thesis: Alexander Loginov, "On The Utility of Evolving Forex Market Trading Agents with Criteria Based Retraining"
March 2013

Faculty of Computer Science, Dalhousie University, www.cs.dal.ca

Abstract.  This research investigates the ability of genetic programming to build profitable trading strategies for the Foreign Exchange Market (FX) of one major currency pair (EURUSD) using one hour prices from July 1, 2009 to November 30, 2012. We recognize that such environments are likely to be non-stationary and we do not expect that a single training partition, used to train a trading agent, represents all likely future behaviours. The proposed adaptive retraining algorithm - hereafter FXGP - detects poor trading behaviours and trains a new trading agent. This represents a significant departure from current practice which assumes some form of continuous evolution. Extensive benchmarking is performed against the widely used EURUSD currency pair. The non-stationary nature of the task is shown to result in a preference for exploration over exploitation. Moreover, adopting a behavioural approach to detecting retraining events is more effective than assuming incremental adaptation on a continuous basis. From the application perspective, we demonstrate that use of a validation partition and Stop-Loss (S/L) orders significantly improves the performance of a trading agent. In addition the task of co-evolving of technical indicators (TI) and the decision trees (DT) for deploying trading agent is explicitly addressed. The results of 27 experiments of 100 simulations each demonstrate that FXGP significantly outperforms existing approaches and generates profitable solutions with a high probability.

 

Teaching

Instructor, COMP 1510: An Introduction to Scientific Computing
Winter 2008 & Winter 2009

Department of Computer Science, Memorial University of Newfoundland, www.cs.mun.ca
St. John's, NL, Canada

Taught classes on Fortran and C programming to first year students, prepared and graded assignments and examinations, counseled and advised students, prepared lab tests.



Teaching Assistant and Marker for CSCI 3150 (now CSCI 4150): Artificial Intelligence
2000-2001

Faculty of Computer Science, Dalhousie University
www.cs.dal.ca
Halifax, NS, Canada

Marked assignments and provided feedback to students in third year Artificial Intelligence course; included analysis and testing of associated computer programs.



Lab Instructor, CSCI 1200: Computer Science for Non-majors
2000-2001

Faculty of Computer Science, Dalhousie University
www.cs.dal.ca
Halifax, NS, Canada

Instructed group tutorials on web design, HTML, and JavaScript; assisted students individually in the lab, resolved programming difficulties, and evaluated lab assignments.



Teaching Assistant and Marker for PHIL 2130: Logic: Deduction
1999

Department of Philosophy, Dalhousie University
philosophy.dal.ca
Halifax, NS, Canada

Tutored students on techniques of proof construction; marked final exams.