Aleatoric Systems // Kinetic Stress

Deterministic Synthetic Market Data for DeFi & Quantitative Finance

Markets are aleatoric by nature - driven by chaotic, irreducibly random forces. Yet traditional backtesting relies on static historical data, overfitting to patterns that never repeat.

We solve this paradox by generating deterministic synthetic scenarios through Kinetic Stress - a proprietary framework rooted in Stochastic Resonance and Robust Control. Every generated dataset is fully reproducible via cryptographic manifest.

Test against millions of unseen market regimes, not just the one historical path that happened to occur. Ensure capital preservation where traditional models collapse.

MCP CONFIG
{
  "mcpServers": {
    "aleatoric": {
      "url": "https://mcp.aleatoric.systems",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}
aleatoric-engine
$ aleatoric generate \ --preset hyperliquid_perp_sol \ --duration 3600 --seed 42 Manifest locked: deterministic regime Feed: L2 books + trades streaming... Funding: curve calibrated for venue Output: ./artifacts/sol_perp.parquet
MODE: DETERMINISTIC SEED: 42
Playground

GENERATE MODELED DATA

Try a sample generation with mock data. Choose a preset, set parameters, and see the output format.

300s
OUTPUT PREVIEW Parquet Schema
timestamp side price size order_id
1703894400000bid104.2315.50x8a3f...
1703894400050ask104.258.20x9b4e...
1703894400100bid104.2222.00x7c5d...
1703894400150ask104.2612.80x6d2c...
1703894400200bid104.215.30x5e1b...

This is a static preview. Get your API key to generate real datasets.

Quick Start

GET STARTED IN 3 STEPS

From API key to first dataset in under 5 minutes.

01

Get Your API Key

Sign up for a free account or start your 14-day trial. No credit card required for the free tier.

Create Account
02

Configure Your Client

Add the MCP config to Claude Desktop, Cursor, or any MCP-compatible client.

{
  "mcpServers": {
    "aleatoric": {
      "url": "https://mcp.aleatoric.systems",
      "headers": {
        "Authorization": "Bearer $API_KEY"
      }
    }
  }
}
03

Generate Your First Dataset

Use natural language or direct API calls to generate modeled market data.

"Generate 1 hour of SOL perpetual
order book data with Kinetic Stress
funding regime at seed 42"
What Teams Say

TRUSTED BY QUANT TEAMS

"We ran Kinetic Stress scenarios on our liquidation engine using Aleatoric's funding regimes. Found an edge case that would have cost us $2M in prod."
Head of Risk DeFi Protocol
"Finally, reproducible backtests. Same manifest, same seed, same results every time. Our audit team loves it."
Quant Researcher Prop Trading Firm
"We replaced 3 months of historical data ETL with one API call. The modeled data is statistically indistinguishable from real venue feeds."
CTO Trading Infrastructure Startup
Pricing

SIMPLE, TRANSPARENT PRICING

14-day free trial on Starter. All plans include deterministic outputs and audit-ready artifacts.

14-DAY FREE TRIAL
STARTER
$249/mo
Try free for 14 days
  • 100k MCP pulls / month
  • 5M modeled events included
  • 5 presets
  • Parquet exports + manifests
  • Email support
Overages: $0.50/1k pulls
ENTERPRISE
CUSTOM
Full control
  • All Pro features +
  • Custom venue calibration
  • Dedicated infrastructure
  • Compliance documentation
  • Unlimited requests
  • SLA + priority support
Committed discounts available
FAQ

FREQUENTLY ASKED QUESTIONS

What's an MCP pull?

Each request to the Aleatoric API to generate data counts as one MCP pull. A typical backtest simulation might use 10-50 pulls depending on the duration and granularity of data needed. The modeled data is generated on-demand and delivered in your chosen format (Parquet, JSON, or CSV).

Can I use this without Claude or Cursor?

Yes. While Aleatoric is MCP-native and works seamlessly with Claude Desktop and Cursor, you can also use our standard HTTPS REST API directly. Any HTTP client or programming language can integrate with Aleatoric.

Is the modeled data realistic?

Yes. Our modeled market data is calibrated against real venue microstructure patterns. Order book dynamics, trade distributions, and funding rate behaviors are statistically modeled from actual exchange data. The key difference is reproducibility: same seed + same manifest = identical output every time.

What output formats are supported?

Parquet is the default and recommended format for performance and compatibility with data analysis tools. We also support JSON for web applications and CSV for spreadsheet workflows. All formats include the same schema and can be converted between each other.

Can I cancel anytime?

Yes. All plans are month-to-month with no long-term contracts required. You can cancel or downgrade at any time from your dashboard. Annual billing offers a 17% discount for teams that want to commit.