CURRENTLY IN DEVELOPMENT

Test your strategies
against crashes
that haven't happened yet.

Generate, store, analyze, and visualize synthetic market scenarios — calibrated to real assets, powered by AI. Via browser, Python SDK, or REST API.

Quant researchers Risk managers ML teams Fintech builders

✓ You're on the list — we'll reach out before launch.

Free tier at launch. No credit card required.

Monte Carlo — AAPL · Heston Model

252 steps · 1000 paths

Sharpe Ratio 1.34

Max Drawdown −23.4%

VaR 95% −$8.20

3 models GBM · Heston · Bates

~25 tickers At launch

5 scenarios Named market regimes

3 interfaces Cloud · SDK · API

Your strategy survived 2008.
But did it survive 2008?

Historical data gives you one path through every crisis. One speed. One depth. One recovery shape. If your strategy survives that sequence, you don't know if it's genuinely robust — or just lucky on that particular accident.

The alternative is a quant spending days manually fitting stochastic models, calibrating parameters, and writing simulation code from scratch. We do it in one click or one API call.

Historical data alone One path per crisis. Biased toward recent regimes. Sparse for tail events by definition.

Manual model fitting Days of parameter calibration. Custom simulation code. Results that disappear when the session ends.

✓

QPaths platform Thousands of statistically rigorous variations. Auto-calibrated to real assets. Stored, analyzed, and visualized — in one step.

Who it's for

Developers & Quant Researchers

Test strategies against scenarios that haven't happened yet

Historical data is one sequence of events. Generate thousands of statistically realistic variations — different crash speeds, depths, recovery shapes — and know if your strategy survives the regime, not just the accident.

→ Strategy backtesting · ML training data · Factor stress tests

Risk Managers & Fintech Founders

Explore tail risk without a quant or a Bloomberg terminal

Simulate flash crashes, rate shocks, and bear markets on demand via the browser. No code required. Built-in Sharpe, VaR, and drawdown analysis — computed natively, no export needed.

→ VaR modeling · Scenario planning · Stress testing

ML Teams & Fintech Startups

Generate training data for rare market regimes

Real data is sparse for tail events by definition. Build robo-advisors, risk engines, and portfolio tools on top of synthetic data that passes rigorous statistical validity tests — not just random noise.

→ Model training · Regime exposure · Embedded simulation

Platform capabilities

∿

Engine

Three stochastic models

GBM for baselines. Heston for stochastic volatility and fat tails. Bates adds jump diffusion for flash crash behavior.

⚡

Scenarios

Named market regimes

Pre-built: low_vol_bull, flash_crash, bear_market, rate_shock, high_vol_choppy. Ready to use without parameter tuning.

✦

NL Prompt

Natural language scenarios

Describe any market condition in plain text — "a 2008-style crisis with faster recovery" — and the AI translates it to calibrated parameters automatically.

◎

Calibration

Auto-calibrated to real assets

Pass any ticker and get paths that mirror its real statistical behavior — volatility clustering, tail risk, mean reversion. No manual tuning.

▤

Storage

Persistent dataset library

Every generated dataset is stored with a unique ID. Named, tagged, searchable. Retrieve, don't regenerate. Your work accumulates across sessions.

∑

Analysis

Native financial metrics

Sharpe ratio, VaR, max drawdown, realised volatility, and a statistical quality score — all computed natively. No export, no pandas, no manual work.

◈

Visualization

Interactive fan charts

Fan chart with percentile bands, return distribution histogram, single path explorer, and side-by-side scenario comparison with metric diffs.

{}

Access

Three ways to access

Browser GUI for non-technical users. Python SDK for developers. Raw REST API for full control. Same platform, same data, same capabilities.

How it works

Sign up and get your API key

From signup to first dataset in under 5 minutes. Free tier included, no credit card required.

Generate a scenario dataset

Pick a named scenario, describe one in plain English, or pass a ticker for auto-calibration. The platform generates, runs analysis, and stores the dataset in one step.

Retrieve, analyze, compare

Datasets persist with a unique ID. Pull built-in metrics — Sharpe, VaR, drawdown — or retrieve and compare two scenarios side by side. No regeneration needed.

Your interface, your workflow

Browser GUI for no-code exploration. Python SDK for programmatic access. Raw REST API for full control. Same platform, same data, all three ways.

workflow.py

from qpaths import Client

client = Client(api_key="...")

# Generate and store a dataset
ds = client.generate(
    scenario="flash_crash",
    asset="AAPL",       # auto-calibrate
    paths=1000
)

# Retrieve later — no regeneration
ds = client.datasets.get("ds_abc123")

# Native analysis
ds.sharpe()
ds.var(confidence=0.95)
ds.max_drawdown()
ds.summary()          # all metrics at once

# Compare two scenarios
client.datasets.compare(
    "ds_abc123", "ds_xyz456"
)

# Drop into pandas
df = ds.to_dataframe()

Be first on the platform.

We're building. Join the waitlist and help shape what we ship.