FactorGuide

Python client for the FactorGuide decision intelligence oracle.

FactorGuide answers one question: which couplings in your system are load-bearing and which can you safely ignore? Send a coupling matrix from any domain — portfolios, Bayesian models, sensor networks, multi-agent systems — and get back zone classifications, optimal factorization strategy, and calibrated risk predictions.

Install

pip install factorguide

Quick start

from factorguide import FactorGuideClient

fg = FactorGuideClient()

# Full navigation: zone classification + block-diagonal strategy
result = fg.navigate(
    coupling={"correlation_matrix": [[1.0, 0.5], [0.5, 1.0]]},
    sample_size=200,
)

# What zone is the pair in?
pair = result["dipole_analysis"]["pair_classification"][0]
print(f"IC = {pair['ic']}, Zone = {pair['zone']}")

# Quick single-pair diagnostic
diag = fg.diagnose(i="x", j="y", coupling_value=0.3, sample_size=100)
print(f"Constitutive risk: {diag['risk_constitutive']:.1f}x")
print(f"Inductive risk: {diag['risk_inductive']:.1f}x")

# Report outcome to improve future predictions (free)
fg.report_outcome(
    prediction_hash=result["prediction"]["prediction_hash"],
    approach_taken="factorized",
    ess_ratio=0.85,
)

Payment

Trial tier: 15 free queries per wallet, no payment required.

Paid tiers: send stablecoins to the wallet address in the 402 response, then submit the transaction hash via the API. See factorguide.io for pricing.

Links

• Website: https://factorguide.io
• llms.txt: https://factorguide.io/llms.txt
• Health: https://factorguide.io/health
• Theory: Circulatory Fidelity framework

mcp-name: io.factorguide/factorguide