Přístupnostní navigace
E-application
Search Search Close
Master's Thesis
Author of thesis: Bc. Chiril Florescu
Acad. year: 2025/2026
Supervisor: Ing. et Ing. Zuzana Janková, Ph.D.
Reviewer: Ing. Jan Budík, Ph.D., MBA
This thesis develops a quantitative trading system for drawdown-averse retail investors — market participants whose primary objective is capital preservation, not continuous market exposure. The system defaults to cash and deploys long-equity exposure only when several independent mathematical signatures of a market bottom agree simultaneously. The design rests on three empirical regularities of equity markets. First, every developed equity index has eventually recovered from every drawdown it has experienced — long-equity downside is bounded in the limit of recovery. Second, daily returns are not forecastable but their volatility clusters strongly and persistently (Engle 1982; Bollerslev 1986) — so the system trades regime, not direction. Third, returns and volatility are negatively correlated (the leverage effect); the signature of a market bottom is therefore not a low price (which holds throughout the descent) but a joint condition — deep drawdown paired with volatility that has peaked and is turning down. This logic is implemented in the Crisis-Invariant Classifier system architecture: two transparent K of-N voting classifiers built from pre-registered, distribution-free invariants drive a four state position machine with a 15 % recovery-state stop-loss override. On a sealed out-of sample window from January 2020 to April 2026 (COVID, the 2022 rate shock, the 2023 long-bond episode), the strategy delivers a Sharpe ratio of 0.99 against buy-and-hold's 0.73 and a maximum drawdown of −17 % against −34 %. Drawdown is reduced in 11 of 12 historical anchor crises and in all 11 international equity indices tested without parameter retuning. The contribution is a measurement-consensus discipline that exchanges full-sample compounding for material drawdown reduction.
algorithmic trading · crisis-invariant classifier · market-bottom detection · recovery timing · conditional volatility · crisis regime · K-of-N voting classifier · state-machine trading · per-window backtesting · methodological pre-registration · maximum drawdown reduction · sealed out-of-sample evaluation · capital preservation · leverage effect · volatility autocorrelation · GARCH · phase-space trajectory · bootstrap confidence intervals
Date of defence
15.06.2026
Result of the defence
Not defended (thesis was not successfully defended)
Grading
F
Process of defence
Student ve své prezentaci seznámil komisi s cíli, řešením a výsledky, ke kterým v závěrečné práci dospěl. Komise se poté seznámila s posudky a hodnocením vedoucího práce a oponenta. Vedoucí práce hodnotil práci stupněm „F“, oponent hodnotil práci stupněm „F“. Otázky z posudku vedoucího student zodpověděl v plném rozsahu, otázky z posudku oponenta zodpověděl v plném rozsahu. Otázky členů komise: 1. prof. Ing. Alena Kocmanová, Ph.D.: Definujte co je v tomto případě agenta. Co je přínosem Vaší diplomové práce? Jak bude plánovaná strategie reagovat na neočekávané krize či změny? Vyzkoušel jste strategii na nějaké společnosti? 2. Mgr. Štěpán Konečný, Ph.D.: Kdo by mohl strategii implementovat? 3. doc. Ing. Marie Pavláková Dočekalová, Ph.D.: Jak jste testoval model? Na základě přednesené prezentace a odpovědí na otázky položené v diskusi komise konstatovala, že nedostatky v závěrečné práci jsou závažné a rozhodla, že student práci neobhájil.
Language of thesis
English
Faculty
Fakulta podnikatelská
Department
Institute of Economics
Study programme
International Economics and Business (MGR-MEO)
Composition of Committee
prof. Ing. Alena Kocmanová, Ph.D. (předseda) doc. Ing. Marie Pavláková Dočekalová, Ph.D. (místopředseda) Mgr. Štěpán Konečný, Ph.D. (člen) Ing. Markéta Kruntorádová, Ph.D. (člen) Ing. Roman Ptáček, Ph.D. (člen)
Supervisor’s reportIng. et Ing. Zuzana Janková, Ph.D.
Grade proposed by supervisor: E
Reviewer’s reportIng. Jan Budík, Ph.D., MBA
Grade proposed by reviewer: C
Responsibility: Mgr. et Mgr. Hana Odstrčilová