Tokens API

Trading flow by labelled wallet group, over time

Buy/sell USD volume and wallet/transaction counts per labelled wallet cohort (developers, snipers, smart money, KOLs), bucketed over time; Solana only.

GET/api/external/token-analysis/cohort-flow/{chain}/{mint}

Get a free API key Try it in the API reference

Chains

Solana SOL

Request

Send your key in the x-dex-data-key header. Replace each <placeholder> with a real value.

curl -H "x-dex-data-key: $DEXDATA_API_KEY" \
  "https://api.dexdata.io/api/external/token-analysis/cohort-flow/SOL/<mint>?interval=1m&from=<from>&to=<to>"

Parameters
Parameter	Description
`interval` query · string · required	Bucket width. A curated subset of the underlying time_frame enum — one representative width per its own max-range class (minutes/hours/day-and-up) — `1m` for intraday/launch-forensics reads, `1h` for a medium trend, `1D` for a long one.`1m1h1D`
`from` query · integer · required	Window start, Unix SECONDS. Inclusive.
`namedWallets` query · boolean · optional	Opt in to naming some of the wallets behind each bucket, instead of only counting them. COSTS AN EXTRA UPSTREAM CALL, which is why it is off by default — leave it off if you only need the aggregates. READ THIS BEFORE USING IT: at most 10 wallets are named per bucket, across all cohorts together, so a cohort reporting 270 buying wallets in a bucket can come back with three names. This answers "name me some of them", never "name the cohort". Each point carries `namedWalletsComplete` saying whether the shortfall is proven, and the selection rule behind the ten is not documented — do not read the list as a ranking.
`to` query · integer · required	Window end, Unix SECONDS. Inclusive. The declared ceiling per interval is 60 days at `1m`, 90 at `1h`, 365 at `1D` — a wider window returns 400 naming the real ceiling rather than a silently truncated series. Even within that ceiling the response can still come back `truncated: true` (a real, regularly observed page limit, independent of the day-range check) — split a longer read into multiple calls.
`chain` path · string · required	Solana only; every other chain returns 400. No other wired source reports trading flow bucketed by a labelled wallet cohort for a token on any chain.
`mint` path · string · required	The token mint address.

Response fields

A JSON response. Every field is always present; null means the value could not be established, never zero.

Response fields of Trading flow by labelled wallet group, over time
Field	Type	Description
`interval`	string
`cohorts`	object[]	One entry per labelled wallet cohort with any trading activity in the window. OVERLAPPING BY DESIGN — a wallet tagged both dev and sniper is counted in BOTH cohorts' numbers, so never sum across entries. A cohort with zero activity across the whole window is omitted entirely, never published with an empty series.
`cohorts[].label`	string	This wallet cohort's own tag name. Today exactly four are possible — dev, sniper, smart_trader, kol — a closed vocabulary confirmed against the underlying schema, not discovered by sampling. bundler and insider are real concepts elsewhere on this surface but this endpoint can never return either.
`cohorts[].points`	object[]	Ascending by `at`.
`cohorts[].points[].at`	number
`cohorts[].points[].volumeBuyUsd`	number \| null
`cohorts[].points[].volumeSellUsd`	number \| null
`cohorts[].points[].walletBuyCount`	integer \| null
`cohorts[].points[].walletSellCount`	integer \| null
`cohorts[].points[].txBuyCount`	integer \| null
`cohorts[].points[].txSellCount`	integer \| null
`cohorts[].points[].namedWallets`	object[] \| null	Present only when `namedWallets=true` was requested. Wallets named in this bucket that carry this cohort's tag. AT MOST 10 PER BUCKET, counted across ALL cohorts together — and the same bucket routinely holds hundreds. Compare this list against this point's own `walletBuyCount`/`walletSellCount`: ten names beside a count of 270 is the normal case, not an error. `namedWalletsComplete` says whether the gap is proven. THE SELECTION RULE IS NOT DOCUMENTED and has not been measured — these are not "the largest", not "the first", and not a ranking. Treat the order as arbitrary. `null` means the per-wallet detail could not be read for this bucket; an empty array means nobody in this cohort was named. Different answers.
`cohorts[].points[].namedWallets[].address`	string
`cohorts[].points[].namedWallets[].volumeBuyUsd`	number \| null
`cohorts[].points[].namedWallets[].volumeSellUsd`	number \| null
`cohorts[].points[].namedWallets[].txBuyCount`	integer \| null
`cohorts[].points[].namedWallets[].txSellCount`	integer \| null
`cohorts[].points[].namedWalletsComplete`	boolean \| null	Present only when `namedWallets=true` was requested. `true` — the bucket came back under the upstream per-bucket ceiling, so nothing was withheld. `false` — PROVEN incomplete: this cohort's own `walletBuyCount`/`walletSellCount` exceeds the names given. `null` — the bucket sits exactly on the ceiling and the counts do not prove a shortfall, so it is undetermined. This is derived from the counts in this same response, not from any upstream "more data" flag — that flag reads false on a capped bucket and must not be trusted here.
`truncated`	boolean	true when more buckets exist past what this response covers. A real, regularly observed truncation, independent of the day-range ceiling above — narrow the window or coarsen the interval and re-request.

Start building with DexData today

Sign in, create a key and make your first call in a minute. Free credits every month, and you only pay for calls that succeed.

Get a free API key