lisson.constantin

Teaching

Fall Term 2026 · EBS Universität

• BSc Data Science (Labs)→
  Course homepage

Fall Term 2024 · EBS Universität

BSc Data Science (Lecture slides)

• Linear Regression→
• Classification→
• Multiple Testing & Resampling→
• Model Selection & Regularization→
• Tree-Based Methods→
• Unsupervised Learning→

Explorables

• Meaningless Price Path

  A freshly generated meaningless price path, based on a Brownian Motion with drift — the model that says markets are efficient and your stock picks are pure luck. Consider it an invitation to meditate on how false pattern recognition may lead one astray.

  A simulated stock price chart.

  click the chart for a fresh path

• 字 Radix — Learn characters without brute force→
  An app Claude and I built for learning Chinese and Japanese characters without spaced repetition and without rote visual memorization. Built for my own study, shared because it might help someone.
• Game of Life→
  Conway’s cellular automaton — a few rules about birth and death on a grid, out of which gliders, oscillators, and deterministic chaos emerge. Cells are coloured by how crowded their neighbourhood is.
  An animated Game of Life grid.
• Shrinkage — Lasso, Ridge, and Elastic Net→
  An interactive version of Tibshirani’s “Lasso Picture” to illustrate what the penalty parameter does in regularized regression and how shrinkage and variable selection are special cases of the elastic net idea.
  Two diagrams of elliptical loss contours, one meeting a lasso constraint diamond at its corner and one meeting a ridge constraint circle off-axis.
• Block Bootstrap — alternative histories for time series→
  Resample time series in 30-day blocks and the structure that matters — cross-asset correlation, volatility clustering — survives into every alternative history. Built up visually from the plain bootstrap, with the method’s limitations demonstrated as honestly as its powers.
  Three strips comparing the original sample, a block-bootstrap resample of contiguous chunks, and an i.i.d. resample scrambled into confetti.
• CLT — the Central Limit Theorem on a Galton board→
  Drop balls through a quincunx and watch a bell curve assemble itself, with a live histogram, kernel density estimate, normal fit, and a Kolmogorov–Smirnov test. A second mode shows the same bell emerging no matter how lumpy or skewed the underlying distribution is.
  An animated Galton board assembling a bell curve, beside its empirical, kernel density, and normal CDFs.

Writing

• Using Prospect Theory to Determine Investor Risk Aversion→
  In: Liermann, V., Stegmann, C. (eds). The Impact of Digital Transformation and FinTech on the Finance Professional. Palgrave Macmillan · 2019
• Robust Filtering for Financial Time Series→
  Master thesis · SSRN · 2017
• The Components of Private Debt Performance→
  Giuzio, M., Gintschel, A., Paterlini, S. The Journal of Alternative Investments, 20(4), 21–35 · 2018 · Research support

Elsewhere

Find me on LinkedIn, book a time to talk, or write to public@lisson.co.