Hi all — I maintain EPSILON, an open-source backtesting simulator (GitHub - DresdenGman/EPSILON-trading-simulator: Open-source quantitative decision lab: test market assumptions, preserve negative results, and reproduce evidence. · GitHub, 171 stars), and I wanted to share a teaching resource built on top of it: the Assumption Clinic.
It’s a 25-minute small-group worksheet in which students inspect the hidden assumptions behind a single recorded backtest result — no coding or brokerage account required. The worksheet is built around one fixed public case: baseline −7.75%, 0 of 5 predeclared perturbation checks passing, verdict “rejected.” The failure is the teaching material: students write down a rejection rule before seeing the numbers, list the assumptions the result depends on, and finish by writing what the evidence actually supports (and what it doesn’t).
I piloted it in April 2026 in two 40-minute sessions with 26 students at my school’s math club and collected 5 feedback forms (first batch). Candid framing: it’s an untested pilot resource, not an evaluated teaching intervention — the feedback was critical and useful (why only five perturbations? why no live recomputation?), and those limitations are now named in the worksheet itself.
Links:
- Worksheet + facilitator sheet (OER Commons): EPSILON Assumption Clinic — a 25-minute worksheet for inspecting backtest assumptions | OER Commons
- Archived version (Zenodo DOI): EPSILON Assumption Clinic: a 25-minute worksheet for inspecting backtest assumptions, with public-data demonstration D1 | Zenodo
- Source repo: GitHub - DresdenGman/EPSILON-trading-simulator: Open-source quantitative decision lab: test market assumptions, preserve negative results, and reproduce evidence. · GitHub
If you teach anything involving backtesting — or if you’ve tried getting students to take a published result apart — I’d love to hear how you’d use (or break) this. Happy to share the facilitator sheet and the 5-question feedback form.