Discovery and markets API

Filter wallets by trading performance on one chain

Find the wallets on a chain matching stated bounds: minimum realised profit, volume, swaps, distinct tokens or win rate over 1d/1w/30d/1y, behaviour labels excludable.

GET/api/external/discovery/wallet-screener/{chain}

Get a free API key Try it in the API reference

What it returns

A WALLET screener: state the criterion (minimum realised profit, traded volume, swap count, distinct tokens, win rate; and behaviour labels to exclude) and get the wallets that meet it, ranked. The counterpart to screener, which screens TOKENS. Distinct from richest-wallets, realized-pnl-leaders and unrealized-pnl-leaders, which are three FIXED rankings with no caller-supplied bounds at all, and from wallet-labels/wallet-risk, which take a wallet and answer about it. WHAT THE LABELS MEAN, AND WHAT THEY DO NOT: behaviourLabels describes what a wallet DOES (every value observed across 24 labelled rows on four chains was a behaviour or holdings-size term — bot, scammer, sniper, smart trader, whale). It is NOT an identity: nothing here says a wallet is an exchange, a bridge or a treasury, and no identity term appeared once in the measured sample. category is the separate, structural axis and read null on every row measured through this query path — see its own description. The two are deliberately kept apart and are never merged into one field. NO TOTAL IS PUBLISHED, deliberately: the underlying source reports no count of matching wallets (its own count field was measured to be the page size, not the number of matches), so page forward with offset until a page comes back short rather than computing a page count. Every figure on every row belongs to the requested window, which is echoed in the response so a row can never be read window-less.

Chains

Solana SOL
BNB Chain BNB
Ethereum ETH
Stable STABLE
Base BASE
Robinhood Chain RHC
Arc ARC

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/discovery/wallet-screener/SOL"

Parameters
Parameter	Description
`window` query · string · optional	The trailing window every filter, every returned figure and the ranking apply to. All four were verified live against the source. CAVEAT ON `1y`, measured and published rather than smoothed over: on the two highest-profit rows sampled for that window, realizedProfitUsd came back within 0.0001% of volumeUsd with a winRatePct of 100 and 99.74 — an extreme reading that may be a genuine artefact of how the source aggregates a year. Sample size is two rows at the top of one ranking; treat 1y figures with that in mind.`1d1w30d1y`
`minRealizedProfitUsd` query · number · optional	Only wallets whose realised profit over `window` is at or above this, in USD. Verified to genuinely bind (every returned row satisfied it), not merely accepted.
`minVolumeUsd` query · number · optional	Only wallets whose traded notional over `window` is at or above this, in USD. Verified to genuinely bind.
`minSwaps` query · integer · optional	Only wallets with at least this many swaps over `window`. Verified to genuinely bind.
`minUniqueTokens` query · integer · optional	Only wallets that traded at least this many distinct tokens over `window`. Verified to genuinely bind.
`minWinRatePct` query · number · optional	Only wallets whose win rate over `window` is at or above this, 0-100. Verified to genuinely bind. Read it beside a profit figure, never alone: a high win rate is compatible with a large net loss (many small wins, a few catastrophic ones).
`excludeLabels` query · string · optional	A comma-separated list of behaviour labels to EXCLUDE, e.g. `BOT` or `BOT,SCAMMER`. Verified to genuinely filter, with the control and the treatment measured in a single request: the unfiltered top five wallets by realised profit all carried `BOT`, and the same query excluding it returned five entirely different wallets carrying none. Values observed live: `BOT`, `SCAMMER`, `SNIPER`, `WHALE`, `MEGA_WEALTHY`, `SMART_TRADER_TOKENS_OVER_TWO_DAYS_OLD`. An unrecognised value is a no-op, not an error. There is deliberately no include-labels counterpart: selecting FOR a label and receiving wallets carrying it establishes nothing a caller could not already ask for with a numeric bound.
`sortBy` query · string · optional	Which window figure the ranking is on. Every one of these was verified to genuinely sort (each returns a different, correctly-ordered top page), not merely accepted.`realizedProfitUsdvolumeUsdswapswinRateuniqueTokens`
`sortOrder` query · string · optional	Ranking direction. Defaults to `desc` — the largest value first.`ascdesc`
`limit` query · integer · optional	How many wallets to return. Capped at 100 — the source's own per-call ceiling (enforced after the API-key gate). Defaults to 20. limit ≤ 60 keeps the answer under 32 KB.
`offset` query · integer · optional	0-indexed row offset into the ranking. Page forward until a page comes back shorter than `limit`: no total is available to compute a page count from.
`chain` path · string · required	Every chain this API serves is supported here, each verified to return rows scoped to that chain. A chain outside the enum returns 400. CELO, ARB, OP and SCROLL return 400 here.

Response fields

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

Response fields of Filter wallets by trading performance on one chain
Field	Type	Description
`window`	string	The window every figure on every row below belongs to — echoed from the request.
`wallets`	object[]	Ranked by `sortBy`/`sortOrder`. Empty when nothing on this chain matched every bound. A full page does NOT mean more wallets exist, and a short page is the only signal that they do not — there is no total.
`wallets[].wallet`	string	The wallet address.
`wallets[].firstTransactionAt`	integer \| null	Unix SECONDS of the wallet's first recorded transaction.
`wallets[].lastTransactionAt`	integer \| null	Unix SECONDS of the wallet's most recent recorded transaction.
`wallets[].behaviourLabels`	string[] \| null	What this wallet DOES, in the source's own vocabulary — bot, scammer, sniper, smart trader, whale and the like. NEVER an identity: nothing here says a wallet is an exchange, a bridge, a treasury or a market maker, and across 24 labelled rows on four chains no identity term appeared once. An EMPTY array is a real reading — the source classified this wallet and gave it no label, which is the majority reading once bots are excluded — and is deliberately different from `null`, which means the source did not report the field at all.
`wallets[].category`	string \| null	What this wallet structurally IS, in the source's own closed vocabulary (observed member: `NORMIE`; the vocabulary also names exchange, pool-authority, staking-vault, burn, locker and token-creator kinds). This is the identity axis, kept strictly apart from `behaviourLabels`. MEASURED CAVEAT: it read `null` on all 26 rows returned through this endpoint's own query path across four chains (SOL, ETH, BNB, BASE) — STABLE, RHC and ARC were not among those four — so expect `null` today — that is the source's own answer, not a field that went unchecked. Published verbatim and never normalised into a fixed enum, which would assert a category the source did not send.
`wallets[].scammerScore`	integer \| null	The source's own scam score. `null` means no score, which is NOT a score of zero — measured `null` on 3 of 5 rows and `90` on another in the same response. Higher is worse.
`wallets[].botScore`	integer \| null	The source's own automation score. Higher means more bot-like. `null` means no score, never zero.
`wallets[].volumeUsd`	number \| null	Total USD notional traded over `window`. A real `0` is a reading; `null` means the source said nothing.
`wallets[].realizedProfitUsd`	number \| null	Profit already taken over `window`, in USD. Can be negative. Read it beside `winRatePct`: the two disagree more often than they agree at the extremes.
`wallets[].averageProfitUsdPerTrade`	number \| null	Mean realised profit per trade over `window`, in USD.
`wallets[].realizedProfitPct`	number \| null	Realised profit over `window` as a PERCENT, not a fraction. Positive means profit.
`wallets[].avgHoldPeriodSec`	number \| null	Mean SECONDS between opening and closing a position over `window`.
`wallets[].swaps`	integer \| null	Swap count over `window`. A real `0` is a reading.
`wallets[].uniqueTokens`	integer \| null	Distinct tokens traded over `window`.
`wallets[].winRatePct`	number \| null	Share of profitable closed positions over `window`, 0-100. The source counts a closed position only when its profit or loss exceeds $1, so a wallet whose positions are all smaller than that reads 0 rather than null.
`wallets[].nativeTokenBalance`	number \| null	The wallet's balance of the chain's OWN native token (SOL, ETH, BNB …), in whole tokens — already scaled by the source, never raw base units. Verified decimalised on Solana, Ethereum, BNB and Base; not separately verified on the remaining chains.

Price per call

Solana, BNB Chain, Ethereum, Stable, Base, Robinhood Chain, Arc: 141 credits

Charged only when the call succeeds. Prices as of 2026-10-30; the current ones are always in the API reference.

Filter tokens by liquidity, market cap, volume, holders or recent listing
The same filter-and-rank shape asked about tokens instead of wallets.
Realized-profit trader ranking, Solana only
A fixed profit ranking where this one takes caller-supplied bounds.
One wallet's profit across tokens
The per-wallet detail behind a row this returns.
Most-called contracts on a chain, by window
The contracts on a whole chain that drew the most transactions over a chosen window (5m/1h/1d/7d/30d), with the gas each used. RHC only.
How big is the whole crypto market right now?
Total crypto market cap and 24h volume, the 24h change, active coins and markets, Bitcoin and other top-coin dominance, and the DeFi share. No chain, no token: the whole market.
How has the whole crypto market moved over time?
Total crypto market cap and 24h volume over the last day (hourly) or 7, 30, 90, 365 days or all history since 2013 (daily). No chain, no token: the whole market.

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