Methodology & Data Sources
Everything on FEELS Analytics is built from publicly verifiable data. This page explains where each number comes from, how indicators are computed, and how often they update.
Data pipeline
An automated pipeline refreshes all indicator series once a day at ~04:00 UTC. Each feed is fetched independently: if a provider is down or returns bad data, that feed keeps its last good values instead of being overwritten — charts never silently show corrupted data. An automated watchdog monitors every feed's freshness and flags anything that stops updating.
Sources
Every series is built from established, publicly verifiable providers — on-chain aggregators, exchange APIs and official economic data (including the US Federal Reserve's FRED database). No black-box or self-reported numbers.
The full source-by-source breakdown — which provider powers each indicator and how often it updates — is available to registered users.
Sign in free to viewHow indicators are computed
Some indicators are served as published by their source (e.g. Fear & Greed, DVOL, COT). Most are computed by us from the base series above: moving averages and bands (Mayer Multiple, Bull Market Support Band, Pi Cycle), z-scores and normalizations (MVRV Z-Score bands, Bubble Risk, Risk Band), log-regression and power-law fits over the full price history (Rainbow, Power Law Corridor, Log Regression Bounds), ratios of on-chain aggregates (NVT, Thermocap Multiple, Hash Price), and rolling correlations against macro series. Where a provider only serves a truncated history, we merge new data into our own preserved full-history series.
Composites
FEELS Score is a 0–100 cycle-risk composite blending eight indicators: Risk Band, NUPL, log-regression deviation, Mayer Multiple, Fear & Greed, MVRV Z-Score, Puell Multiple and short-term Bubble Risk. Each input is normalized against its historical range; weights are re-normalized whenever an input is temporarily unavailable. 0 means maximum historical opportunity, 100 means maximum risk.
Market Cycle Index blends MVRV Z-Score, Puell Multiple and NUPL, each min-max normalized over its historical extremes, into a single 0–100 cycle gauge.
Accuracy & limitations
On-chain and macro series are revised by their providers from time to time; TradFi series (FRED) pause on weekends and US holidays; a few feeds arrive with a 1–3 day natural lag. Values can differ slightly from other platforms because of source choice, UTC alignment and smoothing. Nothing here is financial advice — these are historical statistics, not predictions.