Data Sources

Where all data comes from, how it's collected, and why each source matters.

Overview

Six independent data feeds supply the BTC Regime scoring model. All data is stored in PostgreSQL and refreshed on a fixed cron schedule via Vercel. Everything flows through the Coinglass API except stablecoins (DefiLlama + CoinGecko) and macro indicators (FMP + FRED).

Source	API	Pillar fed	Frequency
Market Sentiment (Fear & Greed)	Coinglass	Price & Structure	Daily
Bitcoin Spot ETF Flows	Coinglass	Liquidity	Daily
Stablecoin Supply	DefiLlama + CoinGecko	Liquidity	Daily
Exchange Netflows	Coinglass on-chain	Liquidity	Every 2 hours
Derivatives (Funding, OI, Liquidations)	Coinglass	Derivatives + Volatility	Every 4 hours
Macro Indicators (DXY, US10Y, SPX, VIX)	FMP + FRED	Macro Context overlay	Daily

Required env vars: COINGLASS_API_KEY, FMP_API_KEY, FRED_API_KEY. DefiLlama and CoinGecko are free with no key required.

1. Market Sentiment: Fear & Greed Index

Source

Coinglass API : cg.indicators.fearGreed(). Requires COINGLASS_API_KEY.

Data collected

Field	Description	Range
fear_greed_value	Composite sentiment index (latest entry from data_list)	0 (Extreme Fear) to 100 (Extreme Greed)
fear_greed_classification	Text label derived from the value	e.g. “Fear”, “Greed”

Why it matters: This is the sole input for the Price & Structure pillar (30% of the final score). Moderate greed (61–75) signals healthy bull momentum and scores highest (+5). Extreme greed (91+) is treated as a contrarian warning (−2). Extreme fear (0–10) is bearish (−6).

Update schedule

Every 2 hours via /api/sync/daily.

2. Bitcoin Spot ETF Flows

Source

Coinglass ETF API: cg.etf.btcFlowHistory() for daily flows, cg.etf.btcAum() for AUM history. Aggregated across all major ETFs. Requires COINGLASS_API_KEY.

Data collected

Field	Description	Unit
date	Trading date (timestamps are in milliseconds from API)	YYYY-MM-DD
total_net	Total net flow across all ETFs for the day	Millions USD
aum_usd	Total AUM across all ETFs (from btcAum or btcList fallback)	USD

How it's used: The model sums the last 3 trading days of total_net (absolute USD millions) as the etf3dSum input. Thresholds are fixed dollar amounts: >$500M strongly bullish (+6), <−$500M strongly bearish (−6). Feeds Component B of the Liquidity pillar (55% weight within pillar).

Update schedule

Every 2 hours via /api/sync/daily. ETF data reflects the previous trading day's close.

3. Stablecoin Supply

Sources

Primary (mcap history): DefiLlama : fetchStablecoinMcapHistory(). Free, no key required.
Secondary (total market cap for dominance): CoinGecko : /api/v3/global. Free, no key required. 2s rate limit enforced.

Data collected

Field	Description	Unit
market_cap_billions	Total stablecoin market cap (all major coins)	Billions USD
supply_change_percent	7-day % change in total supply	Percentage
dominance_percent	Stablecoin mcap / total crypto mcap × 100	Percentage

Why it matters: Stablecoin supply is crypto's “dry powder.” Rising supply means new capital is entering the ecosystem. Contraction signals capital exiting. The supply_change_percent (7-day change) feeds Component A of the Liquidity pillar (25% weight within pillar).

Update schedule

Every 2 hours via /api/sync/daily.

4. Exchange Netflows

Source

Coinglass on-chain API: client.onchain.balanceChart() for historical balance, client.onchain.balanceList() for per-exchange breakdown. Requires COINGLASS_API_KEY. Results cached for 3 hours (1.5× the sync interval) to survive one failed sync.

Data collected

Field	Description	Unit
exchange_netflow	Net BTC value on exchanges (positive = inflow/distribution, negative = outflow/accumulation)	Millions USD

Why it matters: When large holders move BTC off exchanges they are accumulating (bullish). High inflows signal holders preparing to sell (bearish). Feeds Component C of the Liquidity pillar (20% weight within pillar).

Update schedule

Every 2 hours via /api/sync/exchange-flows.

5. Derivatives: Funding Rate, Open Interest & Liquidations

Source

Coinglass Hobbyist Plan : all three feeds come from Coinglass, which aggregates data from Binance, OKX, Bybit, and other exchanges. Direct exchange API access is not used. Requires COINGLASS_API_KEY.

cg.futures.fundingRateExchangeList() : Binance BTCUSDT perpetual funding rate

cg.futures.oiExchangeList() : aggregate OI across all exchanges (first entry: “All”)

cg.futures.liquidationAggregatedHistory() : 7-day daily breakdown from Binance, OKX, Bybit

Data collected

Field	Description	Unit
funding_rate_percent	Current 8h perpetual funding rate. API value × 100 is stored (0.01 from API = 0.01% stored)	% per 8h
open_interest_usd	Total notional open interest across all exchanges	USD
oi_change_24h_pct	24-hour % change in OI (computed from stored history, not fetched directly)	Percentage
liquidations_24h	Total liquidations in last 24 hours (Binance + OKX + Bybit)	Millions USD
avg_liquidations_7d	7-day average daily liquidations (from 7-day history)	Millions USD

Hobbyist plan limitation: The Top Trader Long/Short Ratio endpoint is locked on the Hobbyist plan. This component was planned but excluded. The Derivatives pillar currently uses three components: Funding Rate (40%), OI Change (30%), and Liquidation Cascade Ratio (30%). Historical z-scores are also unavailable via API and are built from accumulated database history instead.

Why it matters: Funding rate is contrarian: high positive funding means longs are overpaying, signaling overcrowded positioning and reversal risk. OI and liquidation data reveal market stress. These feeds drive both the Derivatives pillar (25%) and the Volatility pillar (15%).

Update schedule

Every 4 hours via /api/sync/derivatives.

6. Macro Indicators

Sources

SPX, VIX, forex pairs: Financial Modeling Prep (FMP) : getHistoricalPrices(). Requires FMP_API_KEY. Free tier: 250 req/day.

US 10-Year yield: FRED (Federal Reserve) : series DGS10. Requires FRED_API_KEY. Free, 120 req/min.

DXY: Synthesized from 5 FMP forex pairs + 1 FRED series using the ICE DXY formula (see below).

Data collected

Field	Description	Source
dxy	US Dollar Index (synthesized from forex rates)	FMP + FRED (computed)
us10y	10-year Treasury yield. Forward-filled on non-trading days.	FRED DGS10
spx	S&P 500 daily close	FMP ^GSPC
vix	CBOE Volatility Index. Forward-filled on non-trading days.	FMP ^VIX

DXY synthesis formula (ICE)

50.14348112 × EURUSD^−0.576 × USDJPY^0.136
              × GBPUSD^−0.119 × USDCAD^0.091
              × USDSEK^0.042 × USDCHF^0.036

EUR/GBP fetched from FMP. SEK via FRED DEXSDUS. JPY, CAD, CHF from FMP forex endpoints. 400 calendar days of history fetched to support the 200-day moving average.

Important: Macro data feeds the Macro Context overlay only, not the BTC regime score. It adjusts position sizing (100% / 75% / 50%) but does not change regime direction. See Macro Context for details.

Update schedule

Via /api/sync/daily (every 2 hours). Macro data from FMP and FRED updates after US market close; earlier runs will re-fetch the same values until new data is published.

Automated Sync Schedule

Schedule	Endpoint	Data synced
Every 2 hours	/api/sync/daily	Fear & Greed (Coinglass), ETF flows (Coinglass), stablecoins (DefiLlama + CoinGecko), macro (FMP + FRED)
Every 2 hours	/api/sync/exchange-flows	Exchange netflows (Coinglass on-chain)
Every 4 hours	/api/sync/derivatives	Funding rate, OI, liquidations (Coinglass)

← Back to Docs Scoring Formulas →