Skip to content
Aleatoric Systems

Mission Briefing

The Operational Threat

Section titled “The Operational Threat”

Historical data is finite.

Strategies trained on static historical logs are inherently fragile. They overfit to past correlations that vanish during novel volatility events. When the market regime shifts—when “black swans” arrive—these agents fail catastrophically.

Aleatoric Systems provides the counter-measure: a deterministic market data engine that generates infinite, fully customizable “counterfactual days” of trading data on demand.

Capabilities

Section titled “Capabilities”

We do not just “replay” data. We simulate the physics of the market with reproducible manifests and seeds so every dataset can be regenerated exactly.

  • Stochastic Latency: Inject network jitter and packet loss to test execution resilience.
  • Adversarial Liquidity: Simulate hostile market makers that pull liquidity during stress events.
  • Funding Variance: Model stochastic funding rate shifts from venue-specific mechanics (HyperLiquid, Binance, OKX, Bybit, CME, SGX) via src/aleatoric/venues/*.
  • Deterministic Replays: Batch and stream parity tested in tests/test_determinism.py; manifests validated via /mcp/config/schema.

Integration

Section titled “Integration”

The engine is MCP-Native (Model Context Protocol). MCP tools expose config validation, preset discovery, funding simulation, normalization, and cache inspection as declared in mcp.json. Order submission is out of scope.