Quantitative researcher & developer building AI-driven, multi-horizon trading systems and the tooling that supports them.
My work blends market micro-structure modelling, reinforcement learning, stochastic calculus and rigorous risk management to squeeze insight out of noisy markets.
I believe in grounding practical applications in solid theory. Here are some of the frameworks I've developed:
- Extended Geometric Information Framework Overview: A conceptual model blending ideas from physics, differential geometry, and game theory to describe agents in dynamic environments.
- A Full Market Dynamics Model (Draft): A formal mathematical system for modeling market environments with hierarchical rules, time-varying graphs, and multi-layered geometric surfaces.
These will both be available under the TMRW repo within the next month(July/2025)
| Repo | TL;DR |
|---|---|
| TMRW part 1 | A research-grade framework that chains Risk AI, Stat AI, Strategy AI (RL) and Forecast AI pods from sub-second market-making to macro-thematic positioning. |
| TMRW part 2 | Live-trading playground for a startup: state-detection logic, tactic selection and execution notes for medium-frequency crypto & equity strategies. |
| Knowledgebase | 890-commit compendium of books, papers & notebooks across quant finance, insurance, maths and ML — currently being reorganised into step-by-step walkthroughs. |
| GoldenDice | Micro toolkit that Monte-Carlo simulates Brownian, CIR and OU paths for pricing and stress-testing. Great sandbox for stochastic-process intuition. |
| QPM | A collection of Jupyter notebooks that prototype mean–variance optimisation, actuarial cash-flows and other portfolio-math basics. |
 |
 |
 |
|
The Strategy AI layer leans on the classical Avellaneda–Stoikov market-making model for optimal bid/ask placement.
Diagrams pulled straight from TMRW/util so they stay up-to-date with the code base.
Python | Jupyter | R | C++ | TensorFlow | PyTorch | CCXT | ARIMA/Prophet | Monte-Carlo sim | Reinforcement Learning (A2C, PPO)
…and a healthy dose of ☕ & 📊.
In other words
Languages: Python, R, C++, SQL
Machine Learning: TensorFlow, Keras, Scikit-learn, LightGBM, XGBoost, HMMlearn
Data Science: Pandas, NumPy, SciPy, Statsmodels
Data Visualization: Matplotlib, Seaborn
The handle is the IATA route I fly to visit my partner — Copenhagen → London.
LinkedIn
Email: mark@brezina.dk
Feel free to reach out if you'd like to collaborate or chat about quantitative finance and AI!