Design of algorithmic trader using the Deep Reinforcement learning - ON-191

Preferred Disciplines: Computer Science, Electrical, Statistic , Industrial & mechanical (PhD or Post-Doc)
Company: Anonymous
Project Length: 4-6 months (1 unit)
Desired start date: 2nd week of January 2019
Location: Toronto, ON
No. of Positions: 3
Preferences: Vector Institute, UofT, York Uni. Ryerson Uni., MaRS Discovery , MILA

About the Company: 

Our Team consist of Forex Traders, Computer engineers and business consultants

Project Description:

Suppose we have a model that can predict the sell/buy state of the market with an acceptable accuracy. Now, our goal is to design a RL agent that can place orders such that:

  1. The big orders would be split into smaller ones, so that the orders get filled with no slippage
  2. Considering the predictor mistakes, the agent should still be able to maximise profits in the long run
  3. It should maximise profit given the dynamic, limited action space (action space of the agent is relatively limited, as it has a certain budget and fixed risk factor).

Research Objectives:

  1. In a market X, train a RL agent to maximize profit by trading given the market state predictor.
  2. Update that RL agent such that it would consider risk factors and account balance


  • Deep Reinforcement learning  

Expertise and Skills Needed:

  • Programming: Python, Tensorflow, Pytorch, C++(optional)
  • Machine learning: Deep learning, Reinforcement learning

For more info or to apply to this applied research position, please

  1. Check your eligibility and find more information about open projects
  2. Interested students need to get the approval from their supervisor and send their CV along with a link to their supervisor’s university webpage by applying through the webform.