Facilitator guide: Module 1, Alternative data in finance
A teacher-facing companion to Module 1. The module is the full lesson; this guide adds the facilitation layer: a preparation checklist, timing cues, guidance for the discussion questions, and notes on what to emphasize and where students get stuck. Slides and a printable version are below.
View slides (PDF) Download slides (.pptx) View printable guide (PDF)
At a glance
- Length: 75 minutes. A 50 minute and a 90 minute variant are noted under Teaching notes.
- Level: undergraduate juniors and seniors, MBA, or graduate finance. No data or coding background required.
- Access: the no-code sandbox at sandbox.altfndata.com only. No API key is needed. Students self-register with a work or school email and are approved automatically.
- Goal of the session: students leave able to define alternative data, name where auction and resale pricing fits, and run their own first query in the sandbox.
Before class
- Register your own sandbox account and run both demo queries end to end, so you have known-good results on screen.
- Open the coverage browser on "watches" once beforehand; note the brand and vendor counts you see so you can point to real numbers live.
- Load the slides and confirm the projector or screen share shows the sandbox clearly. The SQL editor text should be readable from the back of the room.
- Optional: assign the Evaluation Guide at docs.altfndata.com/use-cases.html as a pre-read so students arrive with the vocabulary.
- Have students register their sandbox accounts before class if you can, to save the 5 minutes of setup.
Timed agenda with cues
| Time | Segment | What to do | Watch for |
|---|---|---|---|
| 0 to 10 | What is alternative data | Contrast traditional data (filings, price feeds, estimates) with alternative data (satellite, web traffic, card panels, transactional pricing). Ask for examples before you give yours. | Students conflating "alternative" with "informal." Stress that it is simply data outside the traditional trio. |
| 10 to 20 | Where auction and resale data fits | Introduce the coverage (watches, jewelry, art, wine, cars, and more) and that every record has a pre-sale estimate, a realized price, a vendor, and a brand. | Keep it concrete: show a single record's fields rather than listing categories abstractly. |
| 20 to 30 | Sandbox orientation | Students register and open the sandbox together. Walk the SQL editor, coverage browser, and data dictionary tabs. | Registration email typos. Have a backup shared account ready in case one student cannot get in. |
| 30 to 55 | Guided demo | Run the demo below live, narrating each query and result. Change the brand filter and re-run to show the comparison. | Do not rush the first query. Let students read the columns and find the highest and lowest price themselves. |
| 55 to 70 | Student exploration | In pairs, students run a demo query against a different brand or category and note what changes. | Pairs that pick a brand with very few rows. Nudge them to check the row count first. |
| 70 to 75 | Wrap-up and homework | Restate the one-sentence takeaway and hand out the homework. | Leave 2 minutes for the homework logistics, not zero. |
Guidance for the discussion questions
Use these as the "what to listen for" behind each question in the module. They are talking points, not a graded key.
- Alternative vs traditional data, and why demand has grown. Listen for: alternative data sits outside filings, price feeds, and analyst estimates; demand grew because those traditional sources are widely available and slow, so investors seek an edge in data that is harder to get and closer to real behavior.
- Risks of uneven or gappy coverage. Listen for: a signal that looks like a trend may just be changing coverage. Connect this directly to the recency caveat that recurs across the course.
- Why sale_date matters for point-in-time correctness. Listen for: using the actual sale date, not the ingestion date, keeps a backtest honest, since you must only use what was knowable at the time. The wrong date leaks future information.
- What actual-paid data tells you that a preference survey does not. Listen for: revealed preference. What someone paid is a stronger signal of demand than what they say they prefer.
- Why capture unsold lots, not only sales. Listen for: without the unsold lots you cannot compute sell-through, and a prices-only view is survivorship-biased toward what happened to sell.
- What auction data tells an analyst that filings do not. Listen for: brand-level, near-real-time demand behavior; filings are consolidated, lagged, and accrual-based.
- Ethics and privacy of public auction records. Listen for: these are public sale records rather than private consumer data, which lowers the privacy concern, but coverage and representativeness are still fair game to question.
Teaching notes
- Most common misconception: students treat a
designer LIKE '%Rolex%'result as the whole market for a brand. Show the vendor-count query from the module's Going deeper section so they see how few auction houses a result can rest on. - Second misconception: confusing the pre-sale high estimate with the realized price. Put the two fields side by side on one record and say plainly which is a forecast and which is what was paid.
- 50 minute variant: drop the paired student exploration and shorten the opening discussion to 6 minutes; keep the live demo intact, since it is the load-bearing part.
- 90 minute variant: after the demo, have pairs each pick a different category and briefly report back, then run the vendor-concentration query as a class to set up the coverage-bias theme.
- If the sandbox is slow or blocked on the room network: fall back to the pre-run screenshots you captured in preparation, and have students run the queries themselves as the homework instead.
For the full lesson content, queries, and homework, see Module 1. Questions or a class API key: info@altfndata.com.