All systems operational · v0.1.2 · Apache-2.0 licence

385 risk functions.
One REST API.
Open source.

Financial and risk computation across 8 domains — Monte Carlo VaR, Greeks, stress testing, FRTB, IFRS 9 ECL and more. Open source and fully auditable, cross-validated against published references. Free to use.

Explore domains API docs
$ pip install pyvar-client
385 functions
8 domains
99.97% uptime
Numba JIT-accelerated
pyvar.com · Monte Carlo VaR demo
$ python3
>>> from pyvar import VaRClient
>>> client = VaRClient(api_key="pv_...")
>>> result = client.var.compute(
    portfolio_value=1_000_000,
    returns=returns,
    confidence=0.99,
    n_simulations=1_000
)

# ── Result ──────────────────────────
>>> print(result.var_abs)
28200.0 # £28,200 (99% VaR)
>>> print(result.cvar_abs)
35100.0 # £35,100 (CVaR/ES)
>>> print(result.runtime_ms)
2341 # 1k paths · 2.3s
risk domains

Eight domains.
385 functions.

Every function is accessible via the pyvar REST API. Submit a job, receive a task_id, poll for results. All domains share the same auth, the same client library, the same response format.

domain_01
📈
Market Risk
VaR, ES, Greeks, stress testing, P&L attribution, FRTB
71 functions
domain_02
🏦
Credit Risk
PD/LGD/EAD, IRB, CVA/XVA, IFRS 9 ECL, SA-CCR
55 functions
domain_03
💧
Liquidity Risk
LCR, NSFR, intraday, survival horizon, ILAAP
40 functions
domain_04
⚠️
Operational Risk
LDA, AMA, RCSA, KRI, scenario analysis, SMA
44 functions
domain_05
📊
Portfolio Analytics
MV optimisation, Sharpe, factor models, Black-Litterman
50 functions
domain_06
📋
Regulatory & Compliance
Basel IV, FRTB, MiFID II, EMIR, ICAAP/SREP
30 functions
domain_07
∿
Derivatives & Pricing
Black-Scholes, Heston, yield curves, SABR, fixed income
62 functions
domain_08
⚖️
ALM & Balance Sheet
Duration, NII sensitivity, EVE, IRRBB, FTP
33 functions
api reference

Sync by default.
Async when it counts.

384 of pyvar's 385 functions are synchronous — POST your params to that function's own endpoint, get the JSON result straight back. The one exception is large-scale Monte Carlo VaR (up to 100k paths, 2–10 seconds): POST to submit, GET to poll. No timeouts, no blocking, either way.

core endpoints
POST
/api/v1/{domain}/{function}
Call any function · 384 of 385 — returns the result directly
POST
/api/v1/var/compute
The one async exception · large Monte Carlo VaR · returns task_id
GET
/api/v1/var/result/{task_id}
Poll the Monte Carlo VaR result · pending → started → success
GET
/api/v1/domains
List all 8 domains and function catalogue
GET
/health
Platform status · no auth required
Base URL: https://www.pyvar.com
Auth: Authorization: Bearer pv_...
technology

Built for
performance.

Anthropic Claude
AI reasoning
claude-sonnet-5 / claude-opus-4-8
NumPy · Numba JIT
Monte Carlo engine
@njit parallel
Polars · PyArrow
Data ingestion
lazy scan
FastAPI · orjson
REST API
async
Celery · SQS FIFO
Job queue
AWS
ECS Fargate · EC2 Spot
Compute · EU region
eu-west-1 / us-east-1
Built by Fibtec Limited
pyvar.com is a Fibtec open-source project
Enterprise support, SLA guarantees and private deployment available via fibtec.co.uk
fibtec.co.uk ↗
positioning

Cost, speed, transparency.
An honest comparison, not a sales chart.

pyvar is scored with Fibtec's shared Iron Triangle efficiency model — the same cost/time/quality-gap geometry used across Fibtec's product suite, shown here inverted so a bigger triangle reads as a better outcome — positioned against traditional enterprise risk vendors (Bloomberg, MSCI, Murex, Moody's Analytics and similar). This is a directional, illustrative self-assessment — not an independent or audited benchmark, and no vendor's product was tested to produce it.

Cost efficiency, speed and transparency positioning chart Illustrative positioning of pyvar versus traditional enterprise risk vendors across cost efficiency, speed and transparency/auditability, each normalised 0 to 1 and inverted so bigger is better. pyvar: 1.00, 0.80, 1.00 (efficiency score 10.0/10). Traditional vendors: 0.30, 0.70, 0.40 (efficiency score 7.3/10). Cost efficiency Speed Transparency & auditability
pyvar — free / open tier
TES 10.0/10
Traditional enterprise vendors — typical
TES 7.3/10

Apache-2.0-licensed and free to enter; a Numba JIT-parallel Monte Carlo engine (100k paths in 2–10 seconds); and every formula publicly readable and cross-validated against the published Basel/FRTB reference documents — auditable correctness rather than trust in a brand.

Traditional vendors bring decades of regulatory acceptance and institutional trust this platform hasn't earned yet — this chart doesn't dispute that. Illustrative positioning, not measured units or a named-product test.

get started

Free to use. Open source.

pyvar.com is Apache-2.0 licensed. The API is free to use. Enter your email for a free-tier API key — no password, no credit card.

GitHub ↗ Enterprise support →