About Veqctor
Veqctor helps quants prepare for the quantum era, with a clean, no-regret roadmap and SR 11-7 aligned model kits.
What it does
Assess, build and advance your R&D roadmap
Nothing to upload: no code or data leaves your bank.
Accelerate model development and validation
Choose a model, download its kit: Python prototype, benchmarks and documentation.
Explore quantum technology, hardware and circuits
Circuit kits coming soon.
The no-regret principle
Quantum may be years away, but the classical models it will be measured against are needed today. Use the wait to accelerate your R&D roadmap: you will hit the ground running when quantum arrives, and still be ahead of the game if it does not.
- TodayYour current model, priced and documented
- BenchmarksClosed forms and reference values, errors reported
- EfficiencyVariance reduction, convergence study, AAD Greeks
- Model riskRicher models compared, findings closed
- Quantum readinessReduced case on a simulator, resource estimate, go/no-go
Example for Monte Carlo pricing. Roadmaps differ by use case.
How the validated kits are built
Independent references
Computed once with QuantLib, cross-checked against closed forms and shipped as data. The prototype never imports QuantLib, so the comparison is a real check, and you do not need QuantLib to run anything.
Pre-registered criteria
Every test has its acceptance criterion fixed before the code is written: error bounds, standard errors, convergence rates, no-arbitrage shapes, limits. Criteria are never loosened to make a test pass.
Tested as you would run it
Each kit is installed and tested in a clean environment before release, and every test shipped to users runs again automatically at each change of the library, on two Python versions.
Documentation you can file
Model documentation aligned with SR 11-7 and the ECB guide, pre-filled with theory, assumptions, limits and test results, in Word (native equations) and LaTeX. Shaded sections are left for your institution.
The conventions nobody tests. Day counts, holiday calendars and schedules ship with every kit, checked against QuantLib on thousands of cases, every holiday from 2000 to 2050. Building them surfaced an error in QuantLib's French calendar: Ascension and Whit Monday fixed on set dates.
What it does not do
- It does not touch your systems. Veqctor does not run your models or connect to anything inside your firm. Kits run on your machine.
- It does not take your code or data. You describe your model family in generic terms; nothing is uploaded.
- It does not replace your validation. Kits are tested before release, but your team reviews and adapts them to its data, standards and policy, and owns any regulatory sign-off.
- It does not promise quantum advantage. If quantum computing is not the right answer for your use case, the verdict says so.
- It does not sell hardware. Veqctor is vendor-neutral, with no sponsored placements.
- It does not cover post-quantum cryptography beyond a pointer. That migration is urgent, but it is a security programme, not a modelling one.
Who it is for
Model validators, quant R&D, front-office quants and structurers, risk managers, innovation teams, and managers planning skills and budget. No quantum background is needed.
How it works
The verdict, roadmap and custom variants are generated by an AI model (Claude, by Anthropic) guided by our reference base: technology mapping, roadmaps and model cards. Treat them as a well-informed first draft. Validated kits involve no AI: they are written and tested once.
Veqctor is free. With your consent, your answers and the conversation are stored for 12 months to improve it; downloading a validated kit stores nothing about you. Do not share confidential information. See the privacy notice.
Contact
Feedback, an error in a kit or in the mapping, or a model you would like in the library: write to romaind@phalanx-capital.eu.
Veqctor is published by Phalanx Capital, 60 rue François Ier, 75008 Paris. How we handle code, data and access is described on the security page.