Context and references.
Weekly Bitcoin exchange balance flow is the first thing to check each week: net outflows fit the structural pattern of BTC moving to self-custody and cold storage, net inflows can mean coins are being readied for sale. Glassnode and CryptoQuant both publish the canonical series; we do not quote a weekly figure here because the paid-tier readings could not be independently verified.
Over the longer run, the major providers' charts show exchange-held BTC trending down since around March 2020. The absolute level differs widely between providers, so read the multi-year direction within one provider rather than a single number.
Stablecoin flow is the corresponding companion metric, and here the numbers are public. Per DefiLlama, combined USDT and USDC supply rose about $0.7B in the week of 2026-05-04 to 05-11, fell about $0.8B the following week, and was up about $1.4B over the month to 2026-05-19 (about $266.8B) — dry powder roughly holding steady rather than building fast. Over the same stretch BTC slid from about $80,500 on 05-12 to about $76,800 on 05-19.
Whale wallet activity (wallets holding 1,000+ BTC) is the third lens. Glassnode publishes a whale supply-share series; we do not quote a reading because we could not verify it, and the caveats below explain why the first decimal of that metric is not worth much anyway.
Why two providers give you two different numbers.
Exchange balance is not an on-chain fact. The chain shows addresses; it does not label them. Glassnode and CryptoQuant each maintain their own map of which address clusters belong to which venue, built from deposit-address heuristics and in-house research, and neither publishes that map in full. So any weekly net-outflow figure you read is a labelled estimate, not a measurement. Pull the same week from the other provider and expect a different figure.
That is not a reason to distrust either one. It is a reason never to mix them inside a single series. If you have been reading CryptoQuant since January, keep reading CryptoQuant — otherwise a dramatic "trend break" can turn out to be nothing more than you switching dashboards. Re-labelling also rewrites history: when a provider re-clusters a large venue's wallets, the historical curve moves under you, and a chart you saved last quarter will no longer match the live one.
The same caveat governs whale supply share. A 1,000+ BTC threshold is an address-level cut, and exchange and custodian cold wallets sit far above it. Providers strip known custodial addresses out with differing levels of aggressiveness, which is why whale-share readings from two dashboards can differ by more than the week-over-week move being discussed. The direction over several weeks is the readable part. The first decimal place is not.
Crypto assets are volatile and not suitable for every investor. This page is editorial analysis, not financial advice.