Reading Smart-Money Flow On-Chain: Who Is Buying and Who Already Left
2026-10-08
Price tells you what happened. Flow tells you who did it. The Smart-Money Flow tool looks at the biggest traders in a token's sampled window and asks whether they are accumulating or handing tokens to latecomers, in the spirit of the trader and holder tabs on block explorers.
What is measured
- Buy pressure: buy dollars divided by buy plus sell dollars across top traders. 50% is balanced.
- Net buyers and net sellers: traders whose buys exceed their sells, and the reverse.
- Exited sellers: net sellers who no longer hold the token.
- Profit-takers and accumulators: the five largest net sellers and net buyers by dollars.
- Trend: compares volume in the last third of the window with the first third. More than 1.5 times is accelerating; under half is fading.
The verdict rule
Accumulation needs buy pressure of 60% or more and at least as many net buyers as sellers. Distribution needs 40% or less and at least as many sellers as buyers. Anything else is mixed. Using both a dollar measure and a head-count prevents one giant trade deciding the verdict.
An invented example
Top traders spent $8,100 buying and $1,100 selling: 88% buy pressure, two net buyers, one net seller who fully exited. Verdict: accumulation. If instead two wallets sold $13,000 combined against $500 of buying from a third, pressure falls under 10% and the verdict is distribution.
Limits
Only sampled top traders are included, so small-wallet behaviour is invisible. Labels such as "smart" describe size, not skill. Dollar values use the trade records available and can be incomplete. Use this as a prompt to look closer, not as a trade signal.
API: /api/lab/smart-flow?chain=robinhood&address=0x....
Pricing and limits
Runs are metered. Hindsight tools cost $0.002 a run, Token Lab holder tools $0.01, and Social Pulse or the AI analyst $0.05, taken from account credits. Paid plans include a monthly allowance and then fall back to credits. Anonymous visitors get a few free hindsight samples a day, and any result someone has already computed is free to view for a few minutes. Daily and monthly spend caps apply to every account.
The worked example uses an invented token to show how the score moves; it is not a real project. Run the tool on a real contract to see observed data. Holder data is a sample, linked wallets are observed transfers rather than proof of one owner, and nothing here is financial advice.