Up to 270× faster sensitivities, in the code you already run.
From the team that built XAD: AAD in your own pricing and risk code, every number checked against bumping before go-live.
Faster than bumping · Monte Carlo CVA on QuantLib, 90 inputs, 100,000 paths, XAD with code generation (commercial add-on) · QuantLibAAD, November 2026
Monte Carlo swaption on QuantLib · 45 inputs, 100,000 paths, XAD with code generation
European equity options on QuantLib · 98 inputs, XAD alone
The capital run from over two days of grid time to about an hour
AAD allowed by the MAR50 FAQ · tier-one bank
Sensitivity compute from about 8 hours to under 1, ahead of go-live
same grid · European bank
How we measured: QuantLib · Client results
Batch no longer finishing overnight? FRTB, SA-CVA and XVA, before and after go-live
Four ways to get there, each ending in a number you can check.
XAD
The AAD library for C++ and Python: every sensitivity for as little as 2× one valuation; up to 3× faster than one plain valuation with code generation. Open source under the AGPL; for use in your organisation, a commercial licence by annual subscription.
XADEvery sensitivity for as little as 2× one valuation
AAD enablement
A short, fixed-scope build of XAD into your pricers and risk engine, with tooling that speeds up the port and every sensitivity checked against your bumped numbers. The code is yours, and an XAD subscription keeps it current.
AAD enablementFirst results in weeks, not quarters
AAD training
The AAD course taken by quant teams at leading banks, taught by the team that built XAD, with a hands-on workshop on your own code.
AAD trainingA hands-on workshop on your own code
Xcelerit Verified
Evidence your numbers are right: every price and sensitivity that moved between two releases of your pricing library, each tied to its cause, for model validation; AAD sensitivities checked against bumping; code moved to accelerators checked against the original.
Xcelerit VerifiedEvery price and sensitivity that moved, tied to its cause
The code banks have already validated should keep up, without being rewritten.
Every new rule raises the compute bill: tens of thousands of sensitivities, every book, every day.
- Fast Every sensitivity for as little as 2 times the cost of one valuation with XAD's tape. With code generation, the whole run, every sensitivity included, is up to 3 times faster than one plain valuation of the original code.
- Verified Every sensitivity checked against your own bumped numbers, and code moved to accelerators against the original, before go-live and at every release.
- Open The code and 1,500+ unit tests are public, and the AAD benchmark results are published.
One path
Check
A free speed-up estimate: your code measured in two weeks on your hardware or a cloud equivalent, with a fixed quote. Or a validation plan for code that has to stay right on new hardware. Or start with the AAD course.
Build or verify
AAD enablement of your pricers, risk engine or QuantLib code, or your code moved to accelerators. Fixed scope, first results in weeks, not quarters, every number checked against its reference. The code is yours.
Keep it checked
An annual XAD subscription keeps the code current and checked again through every library release, and Xcelerit Verified checks it again on each new processor, compiler or library.
Start with a speed‑up estimate
Your code, measured free in two weeks on your hardware or a cloud equivalent, with a fixed quote for the build.
Free · Measured on your hardware or a cloud equivalent · A fixed quote for the build · NDA ready
