Reading the ledger instead of the chart
Charts show market outcomes — price and volume after the fact. On-chain analysis goes a layer deeper: because Bitcoin, Ethereum and most major networks are public ledgers, anyone can observe coins moving between wallets, into and out of exchanges, and across the holdings of the largest addresses. It's a data source traditional markets simply don't have — the settlement layer itself is public.
The metrics people actually watch
- Exchange inflows/outflows — inflows can increase potential sell-side supply; withdrawals to private wallets can suggest holding intent, though neither establishes intent by itself.
- Whale wallet activity — movements by the largest holders, watched as potential early signals of big flows.
- Active addresses — unique addresses transacting per period; a network activity pulse.
- MVRV — market cap divided by realized cap (the aggregate cost basis of all coins). High extremes have historically marked overheated markets; low extremes, undervalued ones.
- NUPL — net unrealized profit/loss across all holders; a cycle-position gauge.
- SOPR — whether coins being spent are moving at a profit (above 1) or loss (below 1).
- Hash rate — total mining power securing Bitcoin; a miner-confidence and network-health measure.
- Stablecoin supply — growth in USDT/USDC supply is often read as potential buying power accumulating on the sidelines.
Exchange flows: the most intuitive signal
The workhorse on-chain read is simple: where are the coins going?
- Large inflows to exchanges — holders positioning to sell; historically clusters before selling pressure.
- Large outflows to self-custody — accumulation and holding behavior.
- Stablecoins flowing onto exchanges — dry powder arriving; often read as pre-buying positioning.
The discipline is to read flows over days, not transactions. A single large transfer can be an exchange reshuffling its own wallets, a custodian rebalancing, or an OTC deal settling — none of which mean what a naive reading suggests.
Whale watching, with caveats
Tracking the largest wallets is popular for good reason — large-holder flows can move markets when the entity and context behind them are actually known. But the mapping from wallet to intention is loose: one entity can split holdings across hundreds of addresses, exchanges hold enormous balances in operational wallets, and a "new whale" address may just be an old whale rotating custody. Wallet-level conclusions are only as good as the clustering behind them — which is exactly what commercial on-chain platforms sell. (Coinyong's whale tracker takes a different, cleaner-data angle: on Hyperliquid, every trader's positions and PnL are natively public, so large-trader behavior can be ranked directly instead of inferred.)
Cycle gauges: MVRV and friends
Composite metrics like MVRV and NUPL compress the whole market's cost basis into one number, and their extreme readings have coincided with some historical cycle tops and bottoms. Two honest caveats: past thresholds are descriptive, not predictive — each cycle has reset expectations of "how extreme extreme gets" — and these are slow gauges, useful for cycle context and nearly useless for timing entries. They answer "where in the cycle might we be," not "should I buy today."
Where on-chain analysis falls short
- Exchange-internal trading is invisible. Coins on an exchange trade databases-to-databases; the chain only sees deposits and withdrawals.
- Stablecoin supply is policy, not just demand. Issuers mint and redeem for their own operational reasons.
- Clustering is interpretation. Different analytics platforms attribute the same addresses differently and reach different conclusions from identical raw data.
- Altcoin on-chain data is noisy. Thin activity plus wash trading makes small-cap on-chain signals largely untrustworthy.
How to actually use it
On-chain data earns its place as a cycle and flow check on daily-to-weekly horizons — a complement to, not a replacement for, chart structure and volume. When exchange flows, cycle gauges and market structure all lean the same way, that confluence provides much stronger context than any single metric; any one of them alone is a hypothesis, not a signal. For the rotation context on top, see Bitcoin dominance.