Test ideas without losing track of the data.
Market research software has to handle continuous data, repeat experiments, and make it clear which information was available at each point in a test. Otherwise, a result can look convincing for the wrong reason.
What I built
AstraQuant is a local research system that collects public exchange data, stores it for analysis, and tests ideas with simulated trading. It separates data collection, experiment records, simulated execution, and risk controls so the work can be inspected and repeated.
Technologies used
Python, SQLite, automated data collection, data pipelines, and a paper-only simulation engine.
What this demonstrates
The project shows how I design software that handles demanding data workflows: capture information reliably, process it consistently, automate repeatable tasks, and make failures easier to investigate.
Technical details for engineers
Receipt-time capture records when public market data reached the system. Checksums help validate stored order-book data. Causal replay keeps later information out of earlier decisions. A latency- and depth-aware paper broker models simulated execution, independent risk controls limit experiments, and an experiment ledger records provenance. Verified snapshots support recovery.
- Immutable receipt-time data capture
- Checksum-validated order books and causal replay
- Independent risk controls and paper execution
- Experiment provenance and snapshot recovery
Verified outcome and limits
AstraQuant is paper-only. It does not place live orders or access trading accounts. The documented engineering work does not establish predictive accuracy or trading profitability.
Let’s discuss your project.
Whether you need a customer-facing application, a data pipeline, or help investigating a difficult technical problem, tell me what you are trying to solve.
Start Your Project