SIGNAL_BENCH DOCS

Transparent signals

Methodology

A concise overview of how Signal Bench computes and packages macro signals.

Coverage

Signal Bench currently tracks 148 public macro time series across 11 categories, computing 148 momentum signals (one per series, kept as diagnostic explainers) and 18 product-ready composites that fan in 4 to 10 inputs each.

WhatCount
Source series148
Daily-frequency series47
Composites (product_ready)18
Momentum signals (diagnostic)148
Research categories11

Every series traces back to FRED with the original provider attribution preserved. The catalog grows as new public sources are added — these numbers update automatically.

Normalization

Signal values are normalized into a bounded -1 to 1 score. Polarity determines whether higher, lower, or stable values are positive for interpretation.

Drivers

Composite explanations identify available inputs, coverage, direction, confidence, and the strongest drivers by absolute component score.

Relationships

Relationship analytics compare source series using monthly percent-change Pearson correlations, then surface positive, negative, and broken relationships for review.

Guardrails

Research product

Signal Bench is an analytics and research product. It does not provide personalized financial advice or trade recommendations.