Research, made into software.
This is where our work comes together. Research frames the questions, experiments test the methods, and engineering turns useful ideas into tools for analysis and automation.
A workspace for company and market analysis.
QFI Terminal brings valuation, financial statements, factor grades and company relationships into one workspace. Start with an instrument and examine the data behind it.
Options, institutional positioning, earnings calls and macro data add context. QFI Intelligence works with the information available in the terminal.
Available at qfiterminal.com.
qfiterminal.com
A closer view of how markets trade.
DeepAlpha is a native desktop platform for index futures, with footprint charts, volume profiles, delta and cumulative delta. Live data, replay and simulation support the study of individual sessions.
The platform is built in Rust, with GPU rendering for detailed charts and market replay.
DeepAlpha is in development. Its product site follows that work.
deepalphacharts.com
From a tested procedure to a working system.
Alongside our user-facing products, we develop bots and algorithmic systems for research, testing and trading workflows. A model or a backtest is one component of that work.
- Research bots
- Connect data preparation, experiments and records so a research procedure can be run again and its outcome inspected.
- Trading algorithms
- Turn a strategy specification into explicit decisions, with data timing, execution assumptions and risk limits considered as part of the implementation.
- System behaviour
- Handle missing data, interruptions and unexpected conditions. Logs, replay and monitoring help explain what the software did and why.
The model is part of the system.
A useful product also needs reliable data, clear interfaces and repeatable behaviour. Some features use language or statistical models; others use explicit rules. The implementation follows the task.
Specify
Connect a user need to a defined task, the evidence behind it and the conditions in which it should work.
Implement
Build the data pipeline, logic and interface. Keep configurations and changes traceable, and test the behaviour of the whole system.
Learn from use
Inspect failures and practical limits. Feed those questions back into research and model evaluation.
New tools bring new questions.
Software puts methods in contact with real workflows. What we learn there helps decide what to investigate next.