Real BTC/USD price and on-chain fundamentals (CoinGecko, blockchain.info), a real bounded window of whale transactions scanned directly from recent Bitcoin blocks (blockchain.info rawblock API, no key), a generalized ESD anomaly detector applied to price and active addresses, and a lead-lag correlation test between whale activity and price returns, reported honestly whether or not it turns up a signal.
Red points are days flagged by a generalized ESD test on the residuals of price against its own 7-day moving average (Rosner 1983), not a fixed percentage threshold.
Unique active on-chain addresses per day (blockchain.info), the same ESD test applied independently.
Every block in the scan window was fully downloaded and every transaction's total output value checked; transactions moving ≥50 BTC in outputs are counted as whale transactions. This is a real, recent 2-day window, not a full year, scanning a year of blocks (roughly 52,000) block-by-block is not practical for a demo (see README).
Pearson correlation between daily whale BTC volume and price returns at different lags. Positive lag = whale activity that day correlates with a price return `lag` days later.
| Lag | Pearson r | p-value | n |
|---|
Full pipeline, data sources, and honest limitations: github.com/fborbon/blockchain-analytics.