Lecture 6: Final presentations

Group presentations · Q&A · wrap-up

Prof. Dr. Andre Guettler Director of the Institute
Helmholtzstraße 22, Room 205
andre.guettler@uni-ulm.de
+49 731 50 31 030

Oliver Padmaperuma Doctoral Candidate
Helmholtzstraße 22, Room 203
oliver.padmaperuma@uni-ulm.de
+49 731 50 31 036

6.1 Welcome back

• 6.1 Welcome back
• 6.2 Logistics
• 6.3 Wrap-up & farewell

• Welcome to
• Course at a glance (1/2)
• Course at a glance (2/2)
• A quick look back over the term

Welcome to Finance Project — Asset Management

• This is a project course: there is no central exam to register for. Sign up on the course Moodle page by 15 April 2026 so you receive announcements and the data link.
• Submit the project by 30 June 2026 as a single zip — name pattern: Asset2026_surname1_surname2_surname3. Email it to oliver.padmaperuma@uni-ulm.de, CC andre.guettler@uni-ulm.de and your team-mates.
• Ask questions during or right after each session — that is the preferred channel.
• Admin / studies / exam-eligibility questions go to the registrar’s office (Studiensekretariat) at studiensekretariat@uni-ulm.de.
• Course-content questions outside class: email oliver.padmaperuma@uni-ulm.de, CC andre.guettler@uni-ulm.de.
• We also recommend the student advisory service.

Course at a glance (1/2)

Foundations

Week 1

15.04.2026

Course outline · Backtesting fundamentals

• Course aim & organisation
• Backtesting overview & case study
• In-sample tests (Welch & Goyal 2008)
• Out-of-sample (walk-forward, R²_OS)
• Useful predictors & p-hacking

Introduction to R

Week 2

22.04.2026

RStudio · variables · vectors · data frames · live coding

• Why R for empirical asset-management research
• RStudio and the script editor
• Variables, vectors, matrices, data frames, lists
• Functions and loops
• Data import and export

Assessing model accuracy & Ridge regression

Week 3

29.04.2026

Statistical learning · MSE · bias-variance · linear model selection · Ridge

• Statistical learning: Y = f(X) + ε
• Quality of fit and the train/test MSE distinction
• Bias-variance trade-off and overfitting
• OLS limits: prediction accuracy & interpretability
• Ridge regression and the L2 penalty

Lasso, cross-validation & Elastic Net

Week 4

06.05.2026

Sparse regularisation · resampling for honest test error · choosing λ

• Lasso: L1 penalty and exact-zero coefficients
• Cross-validation: validation set, LOOCV, K-fold
• Choosing the optimal λ for Lasso
• OLS post-Lasso for cleaner coefficient inference
• Elastic Net — combining Ridge and Lasso

Prediction markets, the Polymarket Quant Bench & your project

Week 5

13.05.2026

From Welch-Goyal to event-resolved binary contracts

• Prediction markets — definition and Polymarket as the canonical venue
• How prices form: liquidity, resolution, mechanics
• The Polymarket Quant Bench dataset (HuggingFace): access and schema
• First look at the data in R
• Your project: indicator design, back-test, deliverables, R toolbox

Course at a glance (2/2)

Final presentations

Week 13

01.07.2026

Group presentations · Q&A · wrap-up

• Presentation order and time budget
• Q&A rules
• Closing thoughts and feedback

A quick look back over the term

• L1: backtesting fundamentals — the IS-vs-OOS empirical discipline you’ve now applied yourselves.
• L2: hands-on R — the language you’ve used end-to-end.
• L3 + L4: the regularisation toolbox — Ridge, Lasso, Elastic Net, cross-validation.
• L5: prediction markets primer + the Polymarket Quant Bench dataset.
• Today: you present your strategies. Each group: ~20 minutes.

The course’s arc, in retrospect: we started with the empirical-finance discipline that out-of-sample testing is what matters (Lecture 1, Welch and Goyal). We added the R fluency to do it (Lecture 2). We added the modelling toolkit — bias-variance trade-off, Ridge, Lasso, cross-validation — that lets you pick a model honestly even with many candidate predictors (Lectures 3–4). We applied all of it to a real, messy dataset of resolved prediction markets (Lecture 5). Today you show what you built.

Even if your specific indicator turned out not to work, the intellectual habits — out-of-sample evaluation, walk-forward backtesting, transparent reporting of failures — generalise to any empirical research you do later. That meta-skill is what the course was really teaching; the prediction-market application was the venue, not the destination.

6.2 Logistics

• 6.1 Welcome back
• 6.2 Logistics
• 6.3 Wrap-up & farewell

• Presentation order
• What we expect to see in your slides
• Q&A rules

Presentation order

• Slot order will be drawn from a hat at the start of class so all groups are equally prepared.
• Each slot is 20 minutes presentation + 5 minutes Q&A = 25 minutes in total.
• Switch laptops between groups during the Q&A so we don’t lose time on cable changes.

Bring two laptops per group (primary + backup) and your slides on a USB stick, just in case.

Three pieces of practical preparation:

• Time-box the practice presentation strictly. A 20-minute deck practiced once-through usually runs 25–30 minutes on the day (nerves slow you down). Practice twice; the second time should hit 18 minutes.
• Bring backup hardware. Laptops crash, projectors don’t recognise USB-C, the file you renamed last night didn’t sync to OneDrive. The two-laptop + USB stick policy is for when one of these inevitably happens.
• Know who’s saying what. A 3-person group has 20 minutes to share — typically 5 minutes per person plus an introduction and a wrap-up. Pre-assign sections; don’t decide on the day.

What we expect to see in your slides

1. The question / hypothesis — what predictive idea are you testing?
2. Data & universe — which Polymarket markets did you keep? Which did you exclude, and why?
3. Indicators — describe at least 5; explain why each is plausibly informative.
4. Strategy & back-test — signal construction, walk-forward results, key metrics, one figure that sells it.
5. Robustness & limitations — where does it break? What couldn’t you check?
6. Reflection — what did you learn that you’ll bring to your Master’s thesis?

The six-section structure mirrors the report’s structure but compressed for verbal delivery. A few tips per section:

• (1) Question — single slide, single sentence. The audience should know in 30 seconds what you tested.
• (2) Data & universe — a table of inclusion/exclusion criteria; one slide. State the resulting N markets.
• (3) Indicators — one row per indicator with the formula and a one-line motivation. Don’t show every code chunk; the marker reads the Rmd separately.
• (4) Strategy & backtest — the killer chart of cumulative-returns or hit-rate vs benchmark. This is the slide most of the audience will remember.
• (5) Robustness & limitations — three or four bullets. Honest reporting beats triumphant-but-fragile claims. “In the politics-only subsample our hit rate dropped from 58 % to 52 %; we discuss why in the report” is a great kind of bullet.
• (6) Reflection — what’s the next experiment you’d run? What does this make you want to do for your thesis? This is what we want to hear.

Q&A rules

• Q&A is for the group, not the lecturers — we’ll let questions flow before stepping in.
• Other groups: keep it constructive — frame critiques as “I’d be interested to see…” rather than “you should have…”.
• Keep answers short; if you don’t know, say so. The honesty is recorded as a positive.

The Q&A is graded as part of the presentation — a confident, calibrated handle on questions adds points; defensiveness or evasive answers subtract. Three failure modes to avoid:

• Defensive overreach — “Why didn’t you cluster standard errors?” is not an attack; it’s a probe of your methodological awareness. The right answer is either “we did, here’s where in the report” or “we considered it but chose X for Y reason” — not “the result holds anyway”.
• Inventing answers — making up justifications for choices you can’t actually defend is much worse than saying “we didn’t think about that”. Markers can almost always tell when a justification is post-hoc.
• Talking too long — short, focused answers leave room for follow-ups. A two-minute monologue per question burns the Q&A and leaves the audience disengaged.

For other groups: constructive Q&A makes the session better for everyone. “I’d be interested to see how this performs on sports markets specifically” is a useful question; “you should have used time-series CV” is the same observation phrased better.

6.3 Wrap-up & farewell

• 6.1 Welcome back
• 6.2 Logistics
• 6.3 Wrap-up & farewell

• Grading & feedback
• A short word of thanks
• Good luck out there
• References

Grading & feedback

• Project report: 50% — Rmd + knitted PDF, 10–15 pages.
• Final presentation: 50% — what you do today.
• We’ll publish individual feedback within two weeks of today.
• For the course-wide debrief we’ll send a short anonymous survey via Moodle — please fill it in. It directly shapes the next iteration.

A short word of thanks

• Thank you for engaging with a non-traditional asset class — prediction markets are messy on purpose.
• Special thanks to the groups who flagged issues with the dataset during the project phase: that flows back into the data dump for next year’s cohort.
• Stay in touch — we’re happy to talk about Master’s thesis topics in the same area, including extending today’s work to Kalshi, NFL futures, or macro-conditional event pricing.

The thesis-extension paths mentioned are concrete project shapes for anyone who wants to push the work further:

• Kalshi — US-regulated prediction-market exchange with a different (and arguably better-designed) market structure than Polymarket. Comparing your Polymarket strategy on Kalshi data is a natural extension; the dataset is publicly available via Kalshi’s API.
• NFL futures — sports betting markets are a richer environment for some indicator types (player news, injury reports, weather). Bookmaker odds (Pinnacle, Betfair) are accessible.
• Macro-conditional event pricing — combining prediction markets with macro state (interest-rate environment, equity-market regime). Less explored academically; could be thesis-quality.

The institute is happy to supervise theses in this area; reach out by email if you want to discuss extending today’s work into a Master’s project. Several recent institute publications grew out of project-class kernels — there’s a real path from “course project” to “publishable working paper” if you find a thread worth pulling.

Good luck out there

One last reminder

• Make sure your zip submission landed in oliver.padmaperuma@uni-ulm.de’s inbox by 30 June 2026, 18:00.
• Push your code to a private repo (GitHub, GitLab, or Cloudstore) — useful for thesis interviews.
• The most underrated skill we can hand off: be the person who opens the data and runs glimpse() first. Make it a habit.

References