Back to Documentation

SIPmath

Information about the SIPmath 3.0 standard and how it's used in Liquic

What is SIPmath?

SIPmath (Stochastic Information Packet Math) is a standardized format for representing and sharing probability distributions and stochastic models. SIPmath 3.0 provides a JSON-based specification that enables interoperability between different simulation and risk analysis tools.

Liquic uses SIPmath 3.0 to export probability distributions from prediction markets and economic data, allowing users to import these distributions into other SIPmath-compatible tools for further analysis, simulation, and risk modeling.

Probability Management Organization

The Probability Management Organization maintains and develops the SIPmath standard. The organization promotes the use of probability distributions as a fundamental data type, similar to how numbers and strings are used in traditional computing. They provide tools, specifications, and educational resources for working with stochastic information packets.

The organization's mission is to make probability distributions as easy to work with as traditional data types, enabling better risk analysis, decision-making, and communication of uncertainty across industries.

SIPmath in Liquic

Liquic generates SIPmath 3.0 compliant models from market-implied probability distributions. These models can be exported and used in other SIPmath-compatible tools for portfolio analysis, risk modeling, and scenario planning.

API Endpoints

Belief Oracle Query

Export belief distributions as SIPmath 3.0 models:

/api/beliefs/oracle/query?prob=0.5&asset=BTC&format=sipmath

Returns a full SIPmath 3.0 sipModel JSON with Metalog_2_0 distributions and HDR_2_0 RNG.

Joint Distribution Model

Build joint distributions from multiple data sources:

/api/sipmath/joint?assets=HL:BTC,HL:ETH&transform=pct_change

Creates a SIPmath 3.0 model with Gaussian copula for correlation structure.

Export Format

SIPmath exports from Liquic include:

  • Puesdo Random Number Generators: High-dimensional random number generators for sampling
  • Stochastic Information Packs: Source information, timestamps, and fitting diagnostics
  • Copulas: For joint distributions with correlation structure
  • Global Variables: For coherence and consistency