Research · 3 entries
Research
A working paper, a bachelor thesis and a paper replication, newest first, each with its data, method and result as tested.
- 01
Alpha search in crypto perpetual futures
Working paper, short public edition (10 pages, not peer-reviewed) · 2026
A counted search, July to October 2026, for a price-prediction edge in crypto perpetual futures that survives realistic costs. 1,871 trials were counted by 21 October (about 1,700 of them a written estimate of the unlogged early search), and every result is deflated for that count. From 12 October each new test was written into a hash-chained ledger, with its pass and kill bars, before it ran. Methods include walk-forward tests, combinatorial purged cross-validation (CPCV) with embargo, HAC t-statistics, the deflated Sharpe ratio, PBO, and placebo and label-permutation controls. One dollar-neutral, cross-sectional book that holds positions for about a week passed its look at a 2026 holdout: +23.0 bp/day net of costs and funding, HAC t 2.61, Sharpe ratio 3.45, worst drawdown 7.9% (backtest). A separate adversarial review that I ran then downgraded the pass, because part of the book's design had already seen that window; the paper reports the downgrade and its evidence. Backtest only; not investment advice.
- 02
Quantum Neural Networks: Theory and Application for Financial Time Series Forecasting
BSc thesis, University of Vienna (written within the Bachelor seminar, graded 1) · 2026
Bachelor thesis in mathematics, supervised by Priv.-Doz. Dr. Lukas Exl. It builds quantum neural networks in Qiskit, tunes them with Optuna and retrains them under simulated gate noise to forecast the returns and covariances of five assets; the case study grew out of a WUTIS team project I led. The pipeline is benchmarked on a 121-day held-out test window against classical baselines, and the thesis was graded 1.
- 03
Bass local-volatility model on SPX options: a replication of Conze and Henry-Labordère (2021)
Paper replication, MSc summer research project · 2026
Summer 2026 research project with Prof. Julio Backhoff-Veraguas, implementing “Bass Construction with Multi-Marginals: Lightspeed Computation in a New Local Volatility Model” on a public SPX option-chain sample: forward normalization from put–call parity, OTM quotes to marginals, fixed-point calibration, Asian and forward-start repricing, the local-vol representation, vol Greeks and vanilla (vega) hedges, and the Bass2 general-kernel operator (Theorem 6.1), checked against the Brownian case (scope: the paper through Theorem 6.1). Replacing 24-node Gauss–Hermite quadrature with an exact convolution of the piecewise-linear maps restores the martingale property (max |z| 17.8 → 1.8) and cuts the fixed point from 254/499 to 10/12 iterations. A faithful replication for research, not a production pricer.
Research only; not investment advice.