Tech Arsenal & Toolchain
The production tools behind the research, with the trade-offs that decided each one.
Quantitative Modeling & Core Analytics
Python3.12
Model research, prototyping, orchestration glue
95% track record
Rationale: Ecosystem breadth outweighs raw speed for iterative modeling.
Polars1.9
Tick-level data transforms and joins
88% track record
Rationale: Selected over Pandas for 15x memory efficiency in tick-level simulations.
NumPy2.1
Vectorized numerical kernels
92% track record
Rationale: Baseline computational standard for contiguous memory matrices.
C++C++20
Backtesting engine hot path
78% track record
Rationale: Python overhead was unacceptable inside microsecond tick fill loops.
Data Pipeline & Server Infrastructure
Docker27
Reproducible research environments
85% track record
Rationale: Eliminates environment drift between desk laptop and remote compute nodes.
Airflow2.10
Nightly pipeline orchestration
80% track record
Rationale: Chosen over cron for automated retries, backfills, and DAG visibility.
PostgreSQL16
Research warehouse & time-series features
90% track record
Rationale: Strong window functions and relational auditability.