Learn: tokens
What can on-chain data really tell you?
On-chain data shows amounts moving between addresses on a public blockchain. From labelled exchange addresses you can read inflows, outflows, net flows and the largest day as facts. You cannot read why anyone moved coins, what price will do next, or who owns an unlabelled address. Labels are estimates, so "cannot tell" is often the honest answer.
How it works
Every transfer on a public blockchain is visible: the amount, the sending address, the receiving address and the time. Analytics firms add labels, guesses about which addresses belong to which exchange, fund or bridge. With labels, you can add up exchange inflows (coins sent to exchange addresses), outflows (coins sent from them) and the net flow.
People then tell stories about the numbers: "whales are moving coins to exchanges to sell", "outflows mean holders are taking coins to cold storage". The data cannot say any of that. A transfer to an exchange may be a sale, collateral for a loan, a market maker restocking, or an exchange moving coins between its own wallets.
The honest skill is sorting each claim into three boxes: true from the data, false from the data, or cannot tell from the data.
The exact rule the games grade by
The rule the game uses: flows show amounts moving between labelled addresses, so sums, net flows (inflow minus outflow), the largest day and a labelled exchange's balance change are facts, and claims about them are true or false. Why coins moved, what price does next, and who owns an unlabelled address are not in the data: cannot tell. An outflow to the same exchange's labelled cold wallet is an internal move, not a withdrawal, and labels themselves are estimates.
Step by step
- Read the claim and find the numbers it is about.
- If it is about sums, net flow or the largest day, add up the data and mark it true or false.
- If it is about a motive, such as "to sell", it is cannot tell, however plausible it sounds.
- If it is about the future price, it is cannot tell.
- If it names the owner of an unlabelled address, or calls an unlabelled outflow self-custody, it is cannot tell.
Worked example
A simulated week for one coin. Exchange inflows: 1,200, 900, 2,400, 1,100 and 800 coins over five days, 6,400 in total. Outflows: 1,000, 1,300, 1,500, 900 and 1,200, 5,900 in total. Net flow: 6,400 minus 5,900 = 500 coins into exchanges. There is also one transfer of 3,000 coins from an exchange's hot wallet to the same exchange's labelled cold wallet, and one of 2,000 coins from an exchange to an unlabelled address.
"Net flow into exchanges this week was positive." True: plus 500. "The largest inflow day was day 3." True: 2,400. "More coins left exchanges than arrived." False: 5,900 left and 6,400 arrived.
"The 3,000 coin transfer was a withdrawal by a whale." False: it moved between two wallets labelled as the same exchange, an internal move. "The 2,000 coins went to a holder's cold storage." Cannot tell: nobody knows who owns the unlabelled address. "Inflows mean selling is coming." Cannot tell: motives and future prices are not in the data.
Takeaway: Sums and net flows are facts. Motives, owners and price forecasts are not. Say cannot tell when the data cannot tell.
Common mistakes
- Reading motives into transfers. A transfer to an exchange is not a sale. It might be collateral, a market maker, a custodian or an internal move. The chain records amounts and addresses, not intentions.
- Treating labels as facts. Address labels are estimates by analytics firms and can be wrong or out of date. One entity can control many addresses, and one exchange address can hold coins for millions of people.
- Counting internal moves as flows. Exchanges shuffle coins between hot and cold wallets all the time. A huge transfer between two addresses of the same exchange is housekeeping, not a whale leaving.
- Trading on whale alerts. Automated alerts about large transfers are often followed by stories after the fact. Most of those transfers have ordinary explanations, and the alert cannot tell you which.
What this does not tell you, and what can go wrong
On-chain data is public but incomplete. Trades inside an exchange happen on its own books, not on chain, so a lot of activity never appears in flows at all. Activity on other chains, bridges and derivatives venues is also outside any single chain's view.
We know of no reliable evidence that exchange flow data on its own predicts prices, and stories built on it are often wrong. Treat flows as questions, not signals. For the limits of reading charts in general, see does technical analysis work in crypto.
On-chain data can be useful for checking claims: whether a project's treasury moved, whether an exchange's labelled reserves exist on a given day. See proof of reserves for what such checks can and cannot show.
Questions people ask
What are exchange inflows and outflows?
Exchange inflows are coins sent to addresses labelled as belonging to an exchange; outflows are coins sent from them. The net flow is inflows minus outflows. They depend on labels, which are estimates, and they miss everything that happens inside an exchange's own books.
Do whale transfers to exchanges mean a sell-off is coming?
Not necessarily. The data shows that coins moved, not why. A large transfer to an exchange might be a sale, collateral, a market maker restocking, or an internal move. Nothing in the transfer says what price will do next.
Can on-chain data show who owns a wallet?
Only through labels, which are guesses made by analytics firms from patterns and public information. Most addresses are unlabelled, one person can control many addresses, and an exchange address holds coins for many customers. Treat any claim about who owns an unlabelled address as unproven.
Is on-chain analysis useful at all?
Yes, for facts: how much moved, when, and between which labelled addresses. It can confirm that a treasury moved funds or that an address holds what it claims on a given day. It is weak for motives and forecasts, which is where most of the stories are.
Practise it
Reading about a rule is not the same as applying it. These games generate a fresh simulation each round, ask you to make the call first, and grade you against the same rule.
On-chain Flows
Exchange inflows, outflows and whale transfers. Say what can and cannot be inferred from them.