CS 7646: Machine Learning for Trading
Review and Retrospective
Disclaimer
The views and opinions expressed in this post are solely my own and do not reflect those of Georgia Tech, the OMSCS program, or any affiliated instructors, TAs, or staff.
Instructor: Tucker Balch & David Joyner
Semester: Spring 2023
Overall Rating: 7.3/10 👍
✅ Pros
- Clear rubric and generous project weighting (71 % of the grade) lets you recover from weak exam scores
- Hands‑on, autograded projects
- Plentiful TA office hours
❌ Cons
- Lectures use some dated libraries (e.g., Q‑learning code in Python 2 style), requiring refactors
- Autograder is unforgiving
🕒 Time Commitment
Plan on 15 – 25 hours per week. The early projects are manageable, but workload spikes with Projects 7–8 and exam prep.
📝 Grade Breakdown
| Component | Weight | Details |
|---|---|---|
| Projects (8 total) | 71% | P1: Martingale (3%) |
| P2: Optimize (3%) | ||
| P3: Assess Learners (15%) | ||
| P4: Defeat Learners (5%) | ||
| P5: Marketsim (8%) | ||
| P6: Indicator Evaluation (7%) | ||
| P7: Q‑Learning Robot (10%) | ||
| P8: Strategy Evaluation (20%) | ||
| Exams (2) | 25% | 12.5% each, proctored via Honorlock |
| Quizzes (~10) | 2% | Weekly multiple‑choice mini‑quizzes |
| Course Surveys | 2% | Start, mid, and end‑of‑course |
✍️ Assignments
P1 Martingale
Simulate a gambler’s ruin strategy.
P2 Optimize Something
Find the max‑Sharpe allocation.
P3 Assess Learners
Evaluate KNN vs. decision trees on price data.
P4 Defeat Learners
Craft an adaptive benchmark to break weak models.
P5 Marketsim
Build a vectorized back‑tester.
P6 Indicator Evaluation
Author and rank technical indicators.
P7 Q‑Learning Robot
Train an RL agent to trade.
P8 Strategy Evaluation
End‑to‑end strategy, report, and video demo.
📖 Exams
Two closed‑book exams delivered via Canvas + Honorlock. Questions are a mix of multiple‑choice and short problems; key formulas must be memorized.
🗂 Final Project
Project 8 (20%) is the finale. In this project you create and defend a full trading strategy.
📚 Course Content
| Mini‑Course | Focus | Representative Topics |
|---|---|---|
| Manipulating Financial Data in Python | Data wrangling | Pandas, NumPy, vectorization |
| Computational Investing | Portfolio theory | Technical indicators, marketsim |
| Machine Learning Algorithms for Trading | ML & RL | Regression trees, ensemble learning, Q‑learning |
💬 Participation & Interaction
TA office hours rotate across time zones. Peer code snippets are banned expect conceptual help only.
💭 Final Thoughts
Projects are Python heavy and leverage basic libraries for data manipulation such as Pandas and Numpy. If you want a showcase of ML + finance concepts, the eight‑projects and lectures provide a solid introduction.