Every stock trade settles behind closed doors — outside quarterly filings and delayed reports, you’re guessing what other investors are doing. Public blockchains flip that: every transaction ever made is visible to anyone, forever. On-chain analysis is the craft of reading that open ledger for market intelligence. It’s a genuine informational edge that traditional markets don’t offer — and, like every edge, it’s oversold by people who profit from your attention. Here are five metrics actually worth learning, each with the catch that comes attached.

First: what on-chain data is — and isn’t

A public blockchain records amounts, addresses and timestamps — pseudonymously. You can see that some address moved 5,000 BTC to an exchange; you cannot see whose address it is. Analytics firms close part of that gap by labeling known wallets (exchanges, miners, funds, ETFs) and then aggregating the labeled ledger into metrics. So every on-chain number is two layers deep: raw truth (the ledger) plus estimation (the labels). Keep that in mind — it explains most of the failure modes below.

1. Exchange netflow

What it is: coins moving into exchange wallets minus coins moving out. How to read it: sustained inflows suggest holders positioning to sell — you don’t move coins to an exchange to admire them. Sustained outflows suggest accumulation: coins heading to cold storage don’t get sold next week. How to misread it: exchanges constantly reshuffle funds between their own wallets, and label updates lag reality — a scary “inflow spike” can be one exchange doing internal housekeeping. Use netflow as a trend over weeks; never trade a single print.

2. MVRV

What it is: market value divided by realized value. Realized cap re-prices every coin at the value of its last on-chain move — an estimate of the network’s aggregate cost basis — and MVRV compares today’s market cap to that basis. How to read it: historically, MVRV above roughly 3 marked overheated zones (the average holder sitting on ~200% unrealized profit, with every incentive to sell); below 1 meant the average holder was underwater — territory that has historically coincided with late-stage bear markets. How to misread it: those thresholds are backward-looking heuristics distilled from a handful of cycles, and they drift as the asset matures. “Historically” is doing heavy lifting in this paragraph — treat MVRV as a temperature gauge, not a trigger.

3. SOPR

What it is: the Spent Output Profit Ratio — when coins move, are they being sold above or below their acquisition price? How to read it: SOPR above 1 means sellers are realizing profit; below 1 means they’re realizing losses. Sustained sub-1 stretches are the signature of capitulation — people exiting at a loss because they can’t take it anymore. In bull markets, SOPR repeatedly bouncing off 1 (holders refusing to sell at a loss on dips) has marked healthy corrections rather than trend breaks. How to misread it: one ancient whale moving coins distorts the series for days, and intraday SOPR is mostly noise. Smooth it, or be fooled by it.

4. Long-term holder supply

What it is: the share of supply dormant for 155+ days — the threshold past which coins statistically rarely move. How to read it: rising long-term holder supply during and after declines means convicted holders are absorbing what tourists sell — accumulation. LTH supply falling into a rally means old hands are distributing to new buyers, a pattern that has preceded cycle tops. How to misread it: coins “age into” the 155-day bucket mechanically, so LTH supply rises on a lag even without fresh conviction; and large custody migrations (ETF coins moving into institutional wallets) can imitate either signal.

5. Active addresses

What it is: the count of unique addresses sending or receiving per day — the closest thing a network has to a “daily active users” number. How to read it: price grinding higher while active addresses flatline suggests a speculation-led move with no growing usage underneath; price plus rising activity is the healthier combination. Over years, it doubles as an adoption curve. How to misread it: addresses are not users — one person runs dozens, bots run thousands, and layer-2 batching moves real activity off the main chain where this metric counts. Compare like-for-like over long windows only.

The honest limits of the method

Before the limits, one habit that compensates for most of them: never read a metric alone. A single indicator is a story; three pointing the same way are evidence. Exchange outflows plus rising long-term-holder supply plus SOPR resetting at 1 is a coherent accumulation picture — any one of them by itself is a coincidence waiting to be traded on. Analysts call this confluence, and it’s the difference between using on-chain data and being used by it. With that said:

  • Labels lag. Wallet identification is detective work; the discovery of a new exchange wallet rewrites history retroactively.
  • Short-term price is set in derivatives. Perpetual futures and their liquidation cascades dominate daily moves, and none of that happens on-chain. The six forces in what moves crypto prices put these metrics in their proper place — context, not oracle.
  • Activity is migrating. Layer-2s and ETF custody move volume off the visible mainnet rails, so mainnet metrics understate a growing slice of reality.

Where to look, free

You don’t need a paid terminal to start. DefiLlama tracks TVL, stablecoin flows and chain activity for free. Dune hosts thousands of community-built dashboards. Glassnode and CryptoQuant both publish free chart tiers covering the metrics above. Pair any of them with our own price, dominance and sentiment pages — a metric without market context is a Rorschach test.

Where to go next

Combine these signals with the bigger picture in what moves crypto prices, learn the cycle backdrop in crypto bull and bear markets, and keep the glossary bookmarked — every term above has a one-sentence entry there.

This guide is educational only and is not financial advice. On-chain metrics are estimates, not guarantees, and you can lose everything you put into crypto — read our full disclaimer.