PC Labs

PC Labs are the hands-on core of the course: one project per week, done in groups, in Python, on real financial data.

Each lab is written as a short case study — you are dropped into a realistic professional situation with a concrete problem to solve. Read the case, download the data, and work through the assignment in a Jupyter notebook.

How it works:

• Submission: one Jupyter notebook per group, by email to the instructor and the TA. Email title (and notebook title): PCLab#N - Group X - Name1 Name2 Name3.
• Deadline: Friday of the lab’s week, midnight.
• Grading rewards four things: submitting on time, the quality of your code (comments, readability, use of functions), the structure of your notebook (well organized — explain what you are doing and why), and completing the tasks.
• Model solutions: after each deadline, a complete worked solution — code, output, charts, and commentary — is published on the lab’s page.
• Not every lab counts: your lowest lab score is dropped — so one bad week won’t sink you (full grading scheme on Course Info). The Week 6 Hackathon counts as a mandatory PC Lab; Lab 8 is entirely optional.

Beyond the grade: these projects make a genuinely good portfolio for CVs and job interviews.

How we present results

The model solutions follow the same exhibit conventions used in research papers — and points are earned by following them in your notebooks too:

1. One message per exhibit. Decide what a figure should prove before coding it; the title states the message, the axes carry the evidence.
2. Same entity, same color, everywhere. A reader should recognize Amazon (or “the strategy”, or “positive tweets”) by its color without re-reading the legend. Never rely on a library’s default color cycle.
3. No chart junk. Remove the top and right spines; keep a light horizontal grid only where it helps reading; never use a second y-axis.
4. Legends below, framed, centered — not floating over your data. Benchmarks (the S&P 500) are always black.
5. Annotate what matters. Heatmap cells carry their numbers; key points get labeled; reference lines (zero, β = 1, the 45° line) are drawn and named.
6. Numbers have a house format. Coefficients and test statistics: 3 decimals. Percentages and summary statistics: 2. Observations: integers. Everywhere, consistently.
7. Tables are booktabs. Horizontal rules only — no vertical lines, no colored stripes; numbers right-aligned.
8. Every exhibit is self-contained. A figure or table plus its surrounding text must be understandable without running the code: say what is shown, on which sample, and what to conclude.