CS7646 Spring 2017

Overview

You are on the page for information specific to the Spring 2017 session of this course. Go here (Machine_Learning_for_Trading_Course) for overall course policies.

2017 Spring Schedule

• [[1]]

Assignments & Grading

Projects: Projects are worth 70% of the final grade. There are a total of 90 points possible in this section. Your total projects grade is calculated as min(70, sum(project grades)). We strongly recommend that you focus on the first 7 projects and skip the last one. You can earn a perfect score if you do this. However, if you bomb one of the other projects you can make up the points with VICIOUS RABBIT 2000. Note however that this last project is extremely hard. It has so far only been attempted by PhD students without full success.

• [MC1-Project-1: Assess portfolio] 5% (easy)
• [MC1-Project-2: Optimize a portfolio] 5% (easy)
• [MC2-Project-1: Build a market simulator] 10% (moderate)
• [MC3-Project-1: Implement and assess a regression learner using decision trees and random forests] 15% (hard)
• [MC3-Project-2: Q-learning maze navigation] 10% (easy)
• [MC3-Project-3: Implement a "manual" quant strategy, then do it with decision tree classification] 15% (very hard)
• [MC3-Project-4: Q-learning trader] 10% (moderate)
• [MC3-Project-5: Q-learning VICIOUS RABBIT 2000] 15% (extremely hard)

Homework:

• [MC3-Homework-1: Generate datasets that defeat learners] 5%

Exams:

• Midterm Study Guide
• Midterm 11%
• Final 11%

• Class participation: 3%

Thresholds:

• A: 90% and above
• B: 80% and above
• C: 70% and above
• D: 60% and above
• F: below 60%

The projects linked to below are from previous semesters. We keep them here so you can peek ahead, but please keep in mind that they will be revised.

• [MC2-Project-2: Implement bollinger bands, and create a simple trading strategy]
• [MC3-Homework-1: Create a Finance midterm question]
• [MC3-Project-1]
• [MC3-Project-2]
• [MC3-Project-3]