0xinsider

Read-only batch lookup of market flow for 1-25 raw provider condition_id values. Results preserve request order and duplicate inputs return duplicate rows. Uses the same V1 batch item quota as POST /api/v1/markets/flow/batch. Args: - condition_ids (string[], required): Raw provider condition IDs - timeframe (string, optional): "1h", "4h", "24h", or "7d" (default "24h") Returns: Ordered batch market-flow results with per-item success or error details.

Deprecated alias of batch_get_market_flow (#16312), kept live; it calls the deprecated POST /api/v1/markets/intel/batch path, whose envelope keeps object: market_intel_batch. Read-only batch lookup for 1-25 raw provider condition_id values. Results preserve request order and duplicate inputs return duplicate rows. Uses the same V1 batch item quota as POST /api/v1/markets/intel/batch. Args: - condition_ids (string[], required): Raw provider condition IDs - timeframe (string, optional): "1h", "4h", "24h", or "7d" (default "24h") Returns: Ordered batch market-intelligence results with per-item success or error details.

Read-only batch lookup for 1-25 trader wallet addresses or known usernames. Results preserve request order and duplicate inputs return duplicate rows. Uses the same V1 batch item quota as POST /api/v1/traders/batch. Args: - traders (string[], required): Wallet addresses or known usernames - expand (string[], optional): Heavy fields to include for every trader: "strategy", "categories", "quant_metrics", "trust" (per-item quant_metrics is omitted unless its computed row is strictly under six hours old; trust is per-field provenance for grade, P&L, and stats) Returns: Ordered batch trader results with per-item success or error details. Each successful item's trader carries data_quality: status, as_of and field_groups; fresh means tracked and clocked, not current, and unknown means undatable, never recent.

Browse large-trade-active titled markets for discovery surfaces. Returns grouped event clusters or standalone markets plus live facets. Cursor-paginated over visible discovery entries. Args: - limit (number, 1-48, default 24): Max discovery entries per page - cursor (string, optional): Pagination cursor from previous response - category (string, optional): Filter by market category (case-insensitive): a sport or esports bucket such as Basketball or Esports; any other category is rejected - status (string, optional): Filter by status: "active", "closed", or "all" - platform (string, optional): Explore is Polymarket-only; accepted for backward-compatibility, does not change the result set - sort (string, optional): Sort by "trending", "hot", "expiring", "large_trades" (deprecated spelling "whales"), "volume", or "newest" - q (string, optional): Keyword search against market titles, up to 64 characters before whitespace trimming Returns: Market discovery entries, facets, and pagination metadata.

One Polymarket combo: legs, conjunction result, all-time trade totals, the newest 100 taker fills, lifecycle totals and ingest coverage. Requires a Pro or Max API key. Unlike get_combo_fills it keeps executions priced above 0.95. Null lifecycle or missing leg facts are unknown, never zero. Args: - condition_id (string, required): Combo ID, 0x and 62 lowercase hex characters; not a binary market's condition_id Returns: combo, fills, lifecycle, coverage, access and meta. A malformed ID is a bad_request error and an unknown one not_found.

Live Polymarket combo (multi-leg) taker fills with exact decimal-string economics, wallet identity and grade, legs, conjunction result and ingest coverage. Requires a Pro or Max API key. The listing holds executions priced at most 0.95; get_combo has a combo's complete history. sort recent pages with next_cursor; largest and payout page with next_offset. A continuation can end on an empty page. Args: - result (string, optional): CSV of pending, won, lost; all clears the filter - legs (string, optional): CSV of 2, 3, 4, 5+; all clears the filter - min_stake (integer, optional, default 0): Minimum collateral in whole USDC - min_multiplier (integer, optional, default 0): 0 disables the filter; otherwise 2 to 1000000 shares per dollar at execution - grades (string, optional): CSV of S, A, B, C, D, F; all clears the filter; ungraded wallets never match - sports (string, optional): CSV of sport slugs such as nfl, nba, soccer, esports; all clears the filter; a combo matches when any leg is in a selected sport - sort (string, optional): recent (default, newest first), largest (collateral), or payout (buy-fill shares, sells excluded) - window (string, optional): 24h, 7d or all; defaults to all for recent and 24h for largest and payout - limit (number, 1-200, default 50): Fills per page - cursor (string, optional): next_cursor from the previous recent page - offset (number, 0-5000, optional): next_offset from the previous largest or payout page Returns: data (fills), has_more, next_cursor, next_offset, limit, sort, window, coverage and meta.

The stored count of traded Polymarket combos whose YES conjunction is still pending, and when it was computed. Requires a Pro or Max API key. The count refreshes at most every five minutes and is cached for 60 seconds. A null live value is unknown, never zero. Returns: live (count and computed_at, or null) and meta.

Read which V1 data surfaces the API serves for Polymarket, the one venue it covers. Every surface reports the status supported: grade, pnl, strategy, timeline, large_trades, whale_signal (the deprecated spelling of large_trades), suspicious_trades, insider_radar (the deprecated spelling of suspicious_trades), and market_snapshot. Use this before choosing trader, market, large-trade, or suspicious-trade routes. Args: none Returns: Static backend-owned coverage document under the single key polymarket.

Read one dated daily whale-activity report snapshot. Args: - date (string, required): UTC report date in YYYY-MM-DD format Returns: Report snapshot metadata and entries.

Read the Max-only Edge Hunter API at https://0xinsider.com/sports/edge. Pro receives 403 forbidden with reason max_plan_required. One view selector offers board (default), full signals and cohort observations. All views accept kind sports|esports (omission covers both), sports, leagues, market_types, condition_ids, event_slugs, series_slugs and legacy category. Array selections accept 1..100 supplied values before deduplication, including comma-separated values inside a value; repeat REST query keys or supply MCP arrays. Lists match OR; axes match AND. Canonical sports and machine league identities normalize case-insensitively; market types and event/series slugs are exact case-sensitive provider strings. category=NFL selects the Football bucket; leagues=[nfl] selects NFL. Selection precedes the cap or pagination and appears as normalized selection beside the result fields; changed selections cannot share cursors or ETags. Every row carries nullable market_context with nullable league, series_slug, market_type and metadata_updated_at for authoritative filter discovery; metadata_updated_at is the local canonical writer clock, not provider observation time. Selected reads fail closed with 503 read_model_warming if frozen sidecar readiness is absent. Board defaults stay B grade floor, 30 rows and 48-hour horizon; min_grade S/A/B and limit clamp 1..30 are configurable, total is before the cap, with both current outcome quotes. Signals use min_grade default B, horizon default 12/clamp 1..48, limit default 20/clamp 1..100 and cursor, retaining token_id, all ranked/holder/directional/category-skill fields and clocks. Grade floor requires a holder at that grade or better without removing other holders. Observations require cohort wider_holder|in_play|emerging_pile, accept limit/cursor, and retain full observation rows, snapshot_as_of, degraded, source-generation/category funnel diagnostics and REST meta. With cohort wider_holder, admission most_held_48h opts in to the most-held selection instead: per sport, up to 8 wider_holder markets kicking off within 48 hours, most graded holders first, then volume, then soonest kickoff, at most 4 per event, recomputed every refresh and not held; every row's admitted_by is most_held_48h, the response echoes admission, and its funnel, coverage and degradation describe that selection. Required degradation_reasons names each cause with its canonical sport or null for a generation-wide cause. Degradation and coverage input_markets/capacity_limited_markets/emitted_observations/incomplete_holder_markets describe the whole all-sports generation and remain unchanged by selectors. coverage.selected_observations counts matching selected cohort rows before pagination; returned_observations counts this page. Intentional capacity_limited admission does not itself degrade a snapshot, and pagination cannot recover excluded markets. not_configured upcoming supply is not a provider failure. A selected funnel with no failure terminals does not prove the shared generation healthy. A legacy category value that cannot resolve to a sport retains its zero-work empty result: all six counts are zero and reasons are empty without reading the all-sports snapshot, so those values do not establish its health. Observation horizon stays 48h with 12h in-play kickoff lookback and 30s live snapshot ceiling; after one missed refresh, observations serve the last snapshot up to 45s longer with its age in meta.cache_age_s, and a second miss in a row returns 503; funnel accounting is not a league/type/ID selection census. All public views read warmed generations only; missing or stale readiness returns 503 read_model_warming with Retry-After 5. Advanced views use private,no-cache with REST/SDK ETag/304 helpers; board uses private,no-store and has no ETag/pagination. Invalid mode-specific parameters return 400. Price marks are market midpoints, not wallet entry prices; volume is shares. Null/db_only/degraded identify unavailable or weaker evidence. Analytics is not an order instruction.

Replay durable public large-trade events strictly after an opaque cursor, in commit order (a page never reaches past the oldest open write transaction, so a late-committing lower id is delivered later instead of skipped). Filters narrow the stream in place and are bound to the cursor: a cursor issued under other filters answers error.reason cursor_expired, so restart without a cursor. expand=trade adds the public trade read to every event, so a page needs no per-event get_whale_trade call. Args: - cursor (string, optional): Opaque event replay cursor - limit (number, 1-100, default 50) - trader (string, optional): wallet address, trd_ id or username - condition_id (string, optional): raw condition_id or mkt_ id - min_grade (string, optional): S, A, B, C, D or F - min_size (number, optional): minimum size in USD - expand (string, optional): "trade" Returns: Event replay window with retention and completeness metadata; each event carries trade when expanded.

The same game object list_games returns, for one event_slug: both sides with their provider ids and live scores, the UTC kickoff, the provider's own status, the esports series format, and every linked Polymarket market with its condition_id and outcome token ids. The slug is the identity the live_sports_updated webhook pulse carries, so a receiver can read the full game straight from a pulse. Args: - event_slug (string, required): the game's event slug, for example nfl-buf-nyj-2026-09-22. Matched case-insensitively Returns: the game object with the envelope's meta. status.state keeps postponed, cancelled, suspended and delayed distinct from a plain 'not live'. Each game's own coverage reports whether scores, competitor provider ids and the schedule were available: when competitors reads labels, do not join on names. A slug outside the published coverage answers 404, including a real Polymarket event in a sport this deployment does not serve; list_games names what is covered. Carries no price, no sharp-money split and no holder identity.

Deprecated alias of get_suspicious_trades (#16301), kept live; it calls the deprecated /api/v1/insider-radar path. Stored trades whose recorded suspicion score meets the live flag threshold. Cursor-paginated by suspicion score (highest first). Use mode=stable when a walk must stay on one published scoring generation; restart when it returns cursor_expired. Each suspicious trade includes: - suspicion_score (60-100) - severity (currently flag) - Sub-scores: size and fresh_wallet when recorded; timing and edge are null - Evidence JSON with the scorer's stored signals Args: - limit (number, 1-100, default 20): Max results per page - cursor (string, optional): Pagination cursor from previous response - min_suspicion (number, 0-100, default 0): Minimum score; the live floor of 60 also applies - severity (string, optional): "flag" selects live rows; "watch" returns none because no live watch policy exists - mode (string, optional): "live" (default) uses fresh cached pages; "stable" pins the walk to one published scoring generation Returns: List of suspicious trades with trader, market, scores, and evidence.

Deprecated alias of get_suspicious_trade (#16301), kept live; it calls the deprecated /api/v1/insider-radar/{id} path. Read one suspicious trade by ID. Accepts the rf_ ID returned by list responses or the raw numeric whale_alerts.id. Args: - id (string, required): Suspicious trade ID such as rf_123 or 123 Returns: One suspicious trade with trader, market, suspicion scores, and evidence.

List the largest current open positions from graded traders (Polymarket-only). Each row carries position value, open (unrealized) P&L, share count, entry/current price, outcome label, and trader/market context. Cursor-paginated by position value descending. Polymarket-only: the scanner filters platform = 'polymarket', so an unknown condition_id matches nothing and returns an empty list (never fabricated rows). Args: - limit (number, 1-100, default 20): Max results per page - cursor (string, optional): Opaque pagination cursor from previous response - min_size (number, optional): Minimum position value in USD - category (string, optional): Filter by provider-backed market category - min_grade (string, optional): Minimum trader grade (S, A, B, C, D, F) - condition_id (string, optional): Scope to one market (raw provider id or mkt_-prefixed) Returns: List of large positions with trader, market, value, P&L, and pagination metadata.

Read one large trade by ID. Accepts either the wt_ ID returned by large-trade list/history responses or the raw numeric id. Args: - id (string, required): Large trade ID such as wt_123 or 123 Returns: One large trade with trader info, market context, size, side, price, current review_score (signal_score is its deprecated twin, same value) and trader grade, point-in-time recorded_review_score (recorded_signal_score is its deprecated twin; null before 2026-08-03) and trader grade_at_trade with grade_at_trade_status (unknown before 2026-09-19), optional suspicion_score, and optional suspicion_track. A null suspicion track means the persisted legacy row has no track label.

Page the counterparty executions of one large trade past the first page get_large_trade embeds in counterparty_analysis. Reads the immutable snapshot that response names; a stale or changed snapshot fails with reason cursor_expired, so read get_large_trade again and restart. Args: - id (string, required): Large trade ID such as wt_123 or 123 - snapshot_id (string, required): counterparty_analysis.snapshot_id from get_large_trade - cursor (string, optional): executions_next_cursor from get_large_trade or from the previous page - limit (number, 1-100, default 100): Executions per page Returns: executions (each with its first makers page and makers_next_cursor), executions_next_cursor, execution counts, status and meta.

Page the maker wallets of one counterparty execution past the first page its execution embeds. Percentages keep the whole execution as their denominator across pages. Args: - id (string, required): Large trade ID such as wt_123 or 123 - execution_id (string, required): execution_id of the execution - snapshot_id (string, required): counterparty_analysis.snapshot_id from get_large_trade - cursor (string, optional): makers_next_cursor from the execution or next_cursor from the previous page - limit (number, 1-100, default 100): Makers per page Returns: items (makers), next_cursor, execution_id, snapshot_id and meta.

Get recent large trades from graded prediction market traders. Each trade includes a 0-1 review score, persisted suspicion facts, trader grade, and market context. Cursor-paginated, newest first. Args: - limit (number, 1-100, default 20): Max results per page - cursor (string, optional): Pagination cursor from previous response - min_size (number, default 5000): Minimum trade size in USD - category (string, optional): Filter by market category (case-insensitive) - min_grade (string, optional): Minimum trader grade (S, A, B, C, D, F) - suspicious_only (boolean, optional): Keep only persisted suspicion_score >= 60; false preserves the current result set - min_market_volume_share (number, 0-1, optional): Keep only trades whose market_volume_share is known and at least this; a fraction, so 0.01 is one percent of the market's traded volume - sort (string, optional): "recent" (default, newest first) or "market_volume_share" (biggest share of its market first, over the last 30 days, or from the from bound on the history tools; a cursor is bound to the order it was minted in) - since (string, optional): wt_ id of a trade a previous call returned; answers only the trades recorded after it, in commit order, so a trade recorded late still arrives. Poll with the first trade of the last answer that had trades and deduplicate on id; recent order only Returns: List of large trades with trader info, market context, size, side, price, current review_score (signal_score is its deprecated twin, same value) and trader grade, point-in-time recorded_review_score (recorded_signal_score is its deprecated twin; null before 2026-08-03) and trader grade_at_trade with grade_at_trade_status (unknown before 2026-09-19), optional suspicion_score, and optional suspicion_track. A null suspicion track means the persisted legacy row has no track label. The page carries data_quality beside data: status, as_of (the oldest clock the page rests on) and field_groups. fresh means tracked and clocked, not current: compare as_of with your own tolerance. unknown means this read cannot date that group, never that it is recent.

Replay historical large trades from the local record. Supports cursor pagination plus condition, trader, category, grade, platform, suspicion, market-share, and RFC3339 from/to filters, and a market_volume_share sort. Args: - limit (number, 1-100, default 20) - cursor (string, optional): Pagination cursor with wth_ prefix - min_size (number, optional): Minimum trade size in USD - condition_id (string, optional): Exact raw provider condition_id - trader (string, optional): Wallet, alias, or username - category (string, optional) - min_grade (string, optional): S, A, B, C, D, or F; filters on today's grade, not the grade at trade time - suspicious_only (boolean, optional): Keep only persisted suspicion_score >= 60 - platform (string, optional): polymarket or all - from/to (string, optional): RFC3339 traded_at bounds - min_market_volume_share (number, 0-1, optional): Keep only trades whose market_volume_share is known and at least this; a fraction, so 0.01 is one percent of the market's traded volume - sort (string, optional): "recent" (default, newest first) or "market_volume_share" (biggest share of its market first, over the last 30 days, or from the from bound on the history tools; a cursor is bound to the order it was minted in) Returns: Historical large trades with replay metadata, current review_score (signal_score is its deprecated twin, same value) and trader grade, point-in-time recorded_review_score (recorded_signal_score is its deprecated twin; null before 2026-08-03) and trader grade_at_trade with grade_at_trade_status (unknown before 2026-09-19), optional suspicion_score, and optional suspicion_track. A null suspicion track means the persisted legacy row has no track label. The page carries data_quality beside data: status, as_of (the oldest clock the page rests on) and field_groups. fresh means tracked and clocked, not current: compare as_of with your own tolerance. unknown means this read cannot date that group, never that it is recent.

Get the top-ranked prediction market traders (grades S, A, B only). Sorted by composite score descending. Supports cursor pagination and optional filters. Args: - limit (number, 1-100, default 20): Max results per page - cursor (string, optional): Pagination cursor from previous response - category (string, optional): Filter by category slug - strategy (string, optional): Current styles: two_sided, category_focused, high_activity, diversified, mixed, unclassified. Historical filters remain accepted: accumulator, algo_trader, arbitrageur, directional, event_driven, market_maker, momentum, scalper, speculator, swing_trader Returns: Ranked list of traders with grade, score, P&L, volume, win rate, and strategy.

OHLC price candles for a market's outcome tokens from the stored Polymarket CLOB price history, the series the market chart draws, for open and resolved markets. The stored series is daily: a 1d candle usually holds one point, so its open, high, low and close are that day's close, and 1w aggregates the daily closes. No intraday detail is invented. Args: - condition_id (string, required): Raw provider condition_id or the mkt_-prefixed market id - resolution (string, optional): 1d (default, UTC day) or 1w (ISO week from Monday 00:00 UTC) - from (integer, optional): Exclusive lower bound, unix seconds - to (integer, optional): Inclusive upper bound, unix seconds; from must not be after to Returns: condition_id, resolution, and outcomes, each with its token_id and candles, plus meta.

One market's flow and top positions: the signed flow of every tracked large trade in the window (net_flow_usd and its direction, YES or NO), the large-trade count, gross buy and sell volumes, and the top five graded trader positions. The envelope object is market_flow. Use search_markets first to find the condition_id for a market. Args: - condition_id (string, required): Market condition ID (from search_markets results) - timeframe (string, optional): Lookback window: "1h", "4h", "24h", "7d" (default "24h") Returns: Market info, the flow under sharp_money (net_flow_usd, direction YES/NO, large_trade_count, volumes) with smart_money as its deprecated byte-identical alias, and the top 5 graded trader positions.

The S, A and B graded wallets holding one Polymarket market, from a complete provider holder scan: shares, Polymarket's currentValue for the leg, grade, win record in the market's category, and badges. A wallet holding both outcomes is listed once on its net side. Ordered by current value. The roster is at most 60 seconds old over a provider walk at most 180 seconds old; scan.fetched_at is its only timestamp. An incomplete scan is an error with a retry time, never a shorter list. C, D, F and ungraded wallets are never listed and are counted only in scan.wallet_count. Counts differ from get_positions by design: a live scan, the graded cohort, and no 100 USD size floor. Args: - condition_id (string, required): Raw provider condition_id or the mkt_-prefixed market id - outcome (string, optional): yes, no, or all (default) - min_grade (string, optional): S keeps S, A keeps S and A, B (default) keeps S, A and B - limit (number, 1-100, default 20): Holders per page - cursor (string, optional): next_cursor from the previous page, bound to the same market and filters Returns: data (holders), has_more, next_cursor, total, market, scan, totals and meta.

Deprecated alias of get_market_flow (#16312), kept live; it calls the deprecated /api/v1/market/{condition_id}/intel path, whose envelope keeps object: market_intel. Sharp money flow analysis for a specific prediction market. Shows net flow direction, whale trade count, buy/sell volumes, and top graded trader positions. Use search_markets first to find the condition_id for a market. Args: - condition_id (string, required): Market condition ID (from search_markets results) - timeframe (string, optional): Lookback window: "1h", "4h", "24h", "7d" (default "24h") Returns: Market info, sharp_money flow (net_flow_usd, direction YES/NO, volumes) with smart_money as its deprecated byte-identical alias, and top 5 trader positions.

Provider-first live market-card snapshot for one raw provider condition_id. Forwards IDs to the V1 handler without mkt_ rewriting. Args: - condition_id (string, required): Raw provider condition_id returned by search or explore - expand (string[], optional): "trust" adds per-field provenance (source, freshness, reconciliation, completeness) for current_price and spread_bps Returns: Market identity, outcomes, liquidity, sports context, and freshness states, plus trust when expanded.

Read one UTC monthly whale-activity report snapshot. Args: - month (string, required): UTC report month in YYYY-MM format Returns: Report snapshot metadata and entries.

Get today's published Pick of the Day selections and their recorded facts. Pro opens five selections in total: the explicitly designated free selection plus the first four non-free selections in publication order. Max opens every available daily selection, up to twenty. Read entitled `picks` in `publication_order` order; use stable decimal-string `pick_id` for identity, not deprecated `pick_rank`. `is_free_selection` designates account access; `supersedes_pick_id` is deprecated and always null. `locked_picks` lists extra unresolved slots with only pick_rank and required_tier=max; game identity, identifying clocks and backed facts are omitted. Each full pick carries the matchup, backed side (pick_outcome_label), recorded publication price and return on the flat $1,000 stake (`stake_usd`, `return_usd`; `return_per_100` keeps the literal $100 basis), available holder facts, top grade, CLV capture status and backend-provided thesis. `scheduled_picks` contains entitled scheduled slots with release_at and kickoff; unauthorized slots appear in identity-free locked_picks instead. No market identity is revealed before publication. Withdrawn picks are omitted from current and dated recommendations, the archive and its statistics, even after market settlement. The public ledger retains every published commitment under its original identity and proof bytes. When no active pick is available, the tool returns reason pick_not_released and retry_at. Schedule the next read from the returned retry guidance or an entitled scheduled slot's release_at instead of polling. An entitled published pick whose holder proof is not readable appears in `proof_pending_picks` (pick_rank, release_at, kickoff, retry_at), while `picks` contains readable picks. The tool returns read_model_warming only when no entitled published pick has readable proof. If only Max-locked selections have published, a successful state=none response carries picks=[], pick_count=0, locked_picks and an upgrade action. For automated entry, require a present, matching, unexpired entry_authorization. Honor its returned max_entry_price and expires_at, quote a current executable book for the actual stake and retain any stricter local limits. Fees are separate; authorization guarantees neither liquidity nor execution. Args: none Returns: The published pick set for the current product day (America/New_York), or a pick_not_released error with retry_at.

Get the Pick of the Day track record: each published pick that has not been withdrawn, with its outcome (win, loss, void or pending), recorded publication price, unit score and CLV fields, plus the rolling hit rate (wins / decided; void and pending excluded) and flat-$1,000-stake profit and ROI (`hit_rate.stake_usd` names the stake). The actual API key tier controls unresolved access: Pro opens five selections in total, the designated free selection plus the first four non-free selections in publication order; Max opens every available daily selection, up to twenty. Every resolved outcome (win/loss/void) is public. Unentitled pending rows retain only slot/date/outcome and required_tier with backed_side_locked=true; matchup, category, league, image, publication clock and backed facts are omitted. The archive retains the included historical picks, losses included. Withdrawn recommendations are omitted from archive rows and every aggregate, including the cumulative series, even after market settlement. The public ledger retains every published commitment under its original pick_id and proof bytes. Args: none Returns: picks[] newest first with outcome and CLV fields, plus the aggregate record (hit_rate, decided, wins, losses, voids, profit, roi, total risked).

Read one published pick's sealed, opened or uncommitted ledger entry by its stable decimal-string pick_id. commitment_version explicitly selects version 1 legacy proofs or version 2 rank-free proofs. Never infer a proof format from dates or reorder canonical bytes. Pending proofs withhold the backed side and nonce. This tool requires the usual authenticated MCP tools/call transport; the corresponding REST read is public.

Deprecated alias of get_coverage (#16315), kept live; it calls the deprecated /api/v1/platforms path, which serves the same body. Read which V1 data surfaces the API serves for Polymarket, the one venue it covers. Every surface reports the status supported: grade, pnl, strategy, timeline, large_trades, whale_signal (the deprecated spelling of large_trades), suspicious_trades, insider_radar (the deprecated spelling of suspicious_trades), and market_snapshot. Use this before choosing trader, market, large-trade, or suspicious-trade routes. Args: none Returns: Static backend-owned coverage document under the single key polymarket.

Get stored Polymarket fills available for one tracked trader in one market, newest first, with server-computed running_amount and running_avg_price. The REST identity accepts a 0x wallet, username, trd_-prefixed trader id, or bare integer traders.id with precedence wallet, trd_, integer, then username. Only warm-tier traders, the wallets we sync most often, are tracked. Args: - condition_id (string, required): Market condition ID - address (string, optional): Any supported REST trader identity, including a bare integer traders.id - trader_id (number, optional): Convenience alias for a bare integer traders.id; provide exactly one of address or trader_id - limit (number, 1-100, default 20): Max events per page - cursor (string, optional): Pagination cursor from previous response Returns: List of position timeline events with pagination metadata.

Get stored Polymarket fills available for one internal trader id in one market, newest first, using the public V1 /api/v1/traders/{id}/position-timeline route. Args: - condition_id (string, required): Market condition ID - trader_id (number, required): Internal traders.id alias - limit (number, 1-100, default 20) - cursor (string, optional) Returns: List of position timeline events with pagination metadata.

List current prediction-market positions from the public positions-board feed. Returns trader, market, value, side, share, P&L, and freshness context. Cursor-paginated by current value descending. Args: - limit (number, 1-100, default 20): Max results per page - cursor (string, optional): Pagination cursor from previous response - min_size (number, optional): Minimum current value in USD; defaults to 100, or to 0 when wallet is present; send 0 for every reconciled position - category (string, optional): Filter by provider-backed market category - condition_id (string, optional): Scope to one market: raw provider condition_id or mkt_-prefixed market id - wallet (string[], optional): Scope to one wallet or a book of wallets (up to 25); each value is a wallet address, known username, or trd_-prefixed trader id - min_grade (string, optional): Minimum trader grade (S, A, B, C, D, F) - side (string, optional): Filter by binary side: "yes" or "no" - consistency (string, optional): live (default) or snapshot. snapshot requires wallet and freezes up to 500 positions and 2 MB for up to five minutes; repeat the same filters and consistency on every page. A new first page replaces the API key's prior snapshot; a changed filter errors, and an expired or replaced cursor returns cursor_expired Returns: List of current positions with pagination metadata. The page carries data_quality beside data: status, as_of (the oldest clock the page rests on) and field_groups. fresh means tracked and clocked, not current: compare as_of with your own tolerance. unknown means this read cannot date that group, never that it is recent. With consistency=snapshot, the page also carries snapshot: as_of, expires_at, row_count.

Unified whale-activity report snapshot selector. One route for all three granularities; dispatches to the same per-granularity cap and date window as the granularity-specific report tools. Args: - granularity (string, required): "daily", "weekly", or "monthly" - period (string, required): Period token for the granularity. daily: UTC date YYYY-MM-DD. weekly: ISO week YYYY-WW, or a from,to YYYY-MM-DD pair. monthly: UTC month YYYY-MM. Returns: Report snapshot metadata and entries (source_range, snapshot.status, completeness, reconciliation, report).

Ranked sharp-money flow discovery across prediction markets. Use this before you know a condition_id to find where graded traders are net buying or selling. Cursor-paginated, ranked by absolute net flow descending. Args: - timeframe (string, optional): Lookback window: "1h", "4h", "24h", "7d" (default "24h") - limit (number, 1-100, default 20): Max results per page - cursor (string, optional): Pagination cursor with smf_ prefix - category (string, optional): Filter by provider-backed market category - platform (string, optional): "polymarket" or "all" (default "all") - min_grade (string, optional): Minimum trader grade S, A, B, C, D, or F (default "B" = S/A/B) - direction (string, optional): Filter by net flow direction: "YES" or "NO" Returns: Ranked markets with sharp_money (net_flow_usd, direction, whale_trade_count, buy/sell volumes) and market identity. smart_money is a deprecated byte-identical alias of sharp_money.

Deprecated alias of get_sharp_money_flows (#16308), kept live; it calls the deprecated /api/v1/markets/smart-money-flows path. Ranked sharp-money flow discovery across prediction markets. Use this before you know a condition_id to find where graded traders are net buying or selling. Cursor-paginated, ranked by absolute net flow descending. Args: - timeframe (string, optional): Lookback window: "1h", "4h", "24h", "7d" (default "24h") - limit (number, 1-100, default 20): Max results per page - cursor (string, optional): Pagination cursor with smf_ prefix - category (string, optional): Filter by provider-backed market category - platform (string, optional): "polymarket" or "all" (default "all") - min_grade (string, optional): Minimum trader grade S, A, B, C, D, or F (default "B" = S/A/B) - direction (string, optional): Filter by net flow direction: "YES" or "NO" Returns: Ranked markets with sharp_money (net_flow_usd, direction, whale_trade_count, buy/sell volumes) and market identity. smart_money is a deprecated byte-identical alias of sharp_money.

Read one suspicious trade by ID. Accepts the rf_ ID returned by list responses or the raw numeric whale_alerts.id. Args: - id (string, required): Suspicious trade ID such as rf_123 or 123 Returns: One suspicious trade with trader, market, suspicion scores, and evidence.

Stored trades whose recorded suspicion score meets the live flag threshold. Cursor-paginated by suspicion score (highest first). Use mode=stable when a walk must stay on one published scoring generation; restart when it returns cursor_expired. Each suspicious trade includes: - suspicion_score (60-100) - severity (currently flag) - Sub-scores: size and fresh_wallet when recorded; timing and edge are null - Evidence JSON with the scorer's stored signals Args: - limit (number, 1-100, default 20): Max results per page - cursor (string, optional): Pagination cursor from previous response - min_suspicion (number, 0-100, default 0): Minimum score; the live floor of 60 also applies - severity (string, optional): "flag" selects live rows; "watch" returns none because no live watch policy exists - mode (string, optional): "live" (default) uses fresh cached pages; "stable" pins the walk to one published scoring generation Returns: List of suspicious trades with trader, market, scores, and evidence.

Look up a prediction market trader by wallet address or username. Returns grade (S through F), P&L breakdown, win rate, and volume. Optional expand fields for heavier data: - strategy: Trading strategy classification - categories: Per-category performance breakdown - quant_metrics: curated advanced metrics (copy_score and smart_score 0-100, sharpe_30d, sharpe_7d, profit_factor, edge_consistency, sharpe_percentile, pf_percentile, consistency_percentile); emitted only for a computed row strictly under six hours old, otherwise omitted; each present value is a number or null - trust: Per-field provenance (source, freshness, reconciliation, completeness) for grade, P&L, and stats A wallet address with no trader yet returns sync_status "unknown" (not an error). A username or trader id that matches no trader is a not_found error. Args: - address (string, required): Ethereum wallet address (0x...) or trader username - expand (string[], optional): Heavy fields to include: "strategy", "categories", "quant_metrics", "trust" - max_age_s (integer, optional, 0 or more): Freshness ceiling in seconds. The call answers only when data_quality.status is fresh and data_quality.as_of is no older than this; otherwise it fails with reason freshness_ceiling_unsatisfied (HTTP 409). Nothing is refreshed to meet it Returns: Trader profile with grade, P&L, stats, and optional expanded fields. The trader carries data_quality: status, as_of (the oldest clock the body rests on) and field_groups. fresh means tracked and clocked, not current: compare as_of with your own tolerance. unknown means this read cannot date that group, never that it is recent.

One wallet's win record in every canonical category it has a settled market in: wins, decided (wins plus losses) and win_rate per category, busiest first. The Esports record also lists each esports game it has a settled market in. The basis is every settled market at any size, so it differs from get_trader's categories expansion, which applies a 20 USD notional floor and an entry-price rule; neither corrects the other. A category with fewer decided markets than min_decided_for_win_rate keeps its counts with win_rate null and status not_enough_data. An absent category means no record, never a 0% record. Args: - address (string, required): Wallet address (0x...), username, or trd_-prefixed trader id - category (string, optional): One canonical category through the shared rollup: soccer and EPL reach Soccer, every esports label reaches Esports, football is American football. A game is not a filter: ask for esports and read games Returns: records, min_decided_for_win_rate, basis, computed_at and meta. A trader with no settled market returns empty records; an unknown address is a not_found error.

One structured context object for a trader: the full get_trader profile plus position_summary (sync coverage and realized and unrealized P&L rollups with an as_of open-position clock), data_as_of and a freshness_note on what the point-in-time snapshot covers. position_summary is absent when the trader is not tracked locally or native net economics is unavailable. A wallet address with no trader yet returns sync_status "unknown" on the nested trader; a username or trader id that matches no trader is a not_found error. Args: - address (string, required): Wallet address (0x...), username, or trd_-prefixed trader id Returns: trader, position_summary, data_as_of, freshness_note and meta.

Read export source-range, completeness, volume reconciliation, row-count estimate, and large-export policy for one trader. Metadata only; does not start an export job. Args: - address (string, required): Trader wallet address, known username-style lookup, or trd_-prefixed trader ID emitted by the API. Bare integer database IDs are not accepted Returns: Trader export snapshot metadata.

The grade a trader was visibly recorded with at a past instant, from prospective grade observations: what the wallet's grade was when a trade or decision happened, never a reconstruction. status is graded, ungraded or unknown; use grade only when status is graded. Before available_from (the first recorded observation), after deletion, or in a grade-transition gap, status is unknown and grade is null. observation carries observed_at, published_by and model provenance, which older observations leave null. Never substitute a current grade. Args: - address (string, required): Wallet address (0x...), username, or trd_-prefixed trader id - as_of (string, required): RFC3339 instant, such as the trade or decision time; future instants are refused Returns: status, grade, available_from, observation and meta.

Get a trader's daily P&L time series and pre-derived stats from the precomputed daily_pnl read model. Returns entries (daily cumulative P&L), period stats (all/90d/30d/7d), monthly aggregation, per-year totals, and the drawdown series. Reads the refreshed read model, not a per-request equity replay. A resolved trader with no daily P&L returns an empty structured object (HTTP 200); an unknown address returns not found. Covers all Polymarket markets, not only sports and esports: daily_change is Polymarket's own wallet-wide P&L figure, which cannot be split per market. Args: - address (string, required): Wallet address (0x...), username, or trd_-prefixed trader id - from / to (string, optional): Inclusive UTC calendar dates (YYYY-MM-DD) bounding the daily series (entries and drawdown); stats, monthly and year_totals stay defined over the full history - sections (string[], optional): Which sections to return: "entries", "stats", "monthly", "year_totals", "drawdown"; omit for all five. A section left out is absent, and an unrecognized name is an error, never ignored Returns: Structured trader P&L object (entries, stats, monthly, year_totals, drawdown). A request that sent from, to or sections also carries view: the bounds applied, the sections present, entries_total, entries_in_window and the pre-window anchor.

List wallets ranked by trailing-window realized P&L (Polymarket-only discovery). Each row carries the window P&L, volume, distinct markets, grade, hot-streak tier, and a shape-only daily P&L series (per-day deltas from the provider curve; dates ascending, gaps are gaps — NOT zero-filled, and not guaranteed to sum to the window P&L). Cursor-paginated by window P&L descending. A cold read model returns a 503 retry-after (warming), never a 500 or fabricated ranking. Args: - limit (number, 1-50, default 20): Max results per page - cursor (string, optional): Opaque pagination cursor from previous response - window (string, optional): Trailing window: "7d" (default) or "30d" Returns: Ranked trending wallets with window stats and pagination metadata.

Read-only lookup of one webhook endpoint owned by the authenticated API key user. This does not create, update, verify, rotate, or delete webhook endpoints. Args: - id (number, required): Webhook endpoint id Returns: Webhook endpoint configuration record.

Read one weekly whale-activity report snapshot. Pass either from/to UTC dates or an ISO YYYY-WW week token. Args: - week (string, optional): YYYY-WW selector - from/to (string, optional): UTC dates in YYYY-MM-DD format Returns: Report snapshot metadata and entries.

Deprecated alias of get_large_trade (#16304), kept live; it calls the deprecated /api/v1/whale-trades/{id} path, whose envelope keeps object: whale_trade. Read one whale trade by ID. Accepts either the wt_ ID returned by whale-trade list/history responses or the raw numeric whale_alerts.id. Args: - id (string, required): Whale trade ID such as wt_123 or 123 Returns: One whale trade with trader info, market context, size, side, price, current review_score (signal_score is its deprecated twin, same value) and trader grade, point-in-time recorded_review_score (recorded_signal_score is its deprecated twin; null before 2026-08-03) and trader grade_at_trade with grade_at_trade_status (unknown before 2026-09-19), optional suspicion_score, and optional suspicion_track. A null suspicion track means the persisted legacy row has no track label.

Deprecated alias of get_large_trades (#16304), kept live; it calls the deprecated /api/v1/whale-trades path. Get recent large trades from graded prediction market traders. Each trade includes a 0-1 review score, persisted suspicion facts, trader grade, and market context. Cursor-paginated, newest first. Args: - limit (number, 1-100, default 20): Max results per page - cursor (string, optional): Pagination cursor from previous response - min_size (number, default 5000): Minimum trade size in USD - category (string, optional): Filter by market category (case-insensitive) - min_grade (string, optional): Minimum trader grade (S, A, B, C, D, F) - suspicious_only (boolean, optional): Keep only persisted suspicion_score >= 60; false preserves the current result set - min_market_volume_share (number, 0-1, optional): Keep only trades whose market_volume_share is known and at least this; a fraction, so 0.01 is one percent of the market's traded volume - sort (string, optional): "recent" (default, newest first) or "market_volume_share" (biggest share of its market first, over the last 30 days, or from the from bound on the history tools; a cursor is bound to the order it was minted in) - since (string, optional): wt_ id of a trade a previous call returned; answers only the trades recorded after it, in commit order, so a trade recorded late still arrives. Poll with the first trade of the last answer that had trades and deduplicate on id; recent order only Returns: List of whale trades with trader info, market context, size, side, price, current review_score (signal_score is its deprecated twin, same value) and trader grade, point-in-time recorded_review_score (recorded_signal_score is its deprecated twin; null before 2026-08-03) and trader grade_at_trade with grade_at_trade_status (unknown before 2026-09-19), optional suspicion_score, and optional suspicion_track. A null suspicion track means the persisted legacy row has no track label. The page carries data_quality beside data: status, as_of (the oldest clock the page rests on) and field_groups. fresh means tracked and clocked, not current: compare as_of with your own tolerance. unknown means this read cannot date that group, never that it is recent.

Deprecated alias of get_large_trades_history (#16304), kept live; it calls the deprecated /api/v1/whale-trades/history path. Replay historical whale trades from local whale_alerts rows. Supports cursor pagination plus condition, trader, category, grade, platform, suspicion, market-share, and RFC3339 from/to filters, and a market_volume_share sort. Args: - limit (number, 1-100, default 20) - cursor (string, optional): Pagination cursor with wth_ prefix - min_size (number, optional): Minimum trade size in USD - condition_id (string, optional): Exact raw provider condition_id - trader (string, optional): Wallet, alias, or username - category (string, optional) - min_grade (string, optional): S, A, B, C, D, or F; filters on today's grade, not the grade at trade time - suspicious_only (boolean, optional): Keep only persisted suspicion_score >= 60 - platform (string, optional): polymarket or all - from/to (string, optional): RFC3339 traded_at bounds - min_market_volume_share (number, 0-1, optional): Keep only trades whose market_volume_share is known and at least this; a fraction, so 0.01 is one percent of the market's traded volume - sort (string, optional): "recent" (default, newest first) or "market_volume_share" (biggest share of its market first, over the last 30 days, or from the from bound on the history tools; a cursor is bound to the order it was minted in) Returns: Historical whale trades with replay metadata, current review_score (signal_score is its deprecated twin, same value) and trader grade, point-in-time recorded_review_score (recorded_signal_score is its deprecated twin; null before 2026-08-03) and trader grade_at_trade with grade_at_trade_status (unknown before 2026-09-19), optional suspicion_score, and optional suspicion_track. A null suspicion track means the persisted legacy row has no track label. The page carries data_quality beside data: status, as_of (the oldest clock the page rests on) and field_groups. fresh means tracked and clocked, not current: compare as_of with your own tolerance. unknown means this read cannot date that group, never that it is recent.

The fixture list: one coherent game per row, so an agent never has to assemble a game from market searches and team-name matching. Each row carries both sides with their provider ids and live scores, the UTC kickoff, the provider's own status, the esports series format, and every linked Polymarket market with its condition_id and outcome token ids. Ordered by kickoff, then event_slug; games whose kickoff the provider has not published sort last. Args: - sport (string, optional): canonical bucket, case-insensitive, with - and _ read as a space (table-tennis and Table Tennis are one bucket). Omitted means every covered sport - league (string, optional): league tag as coverage.leagues spells it (nfl, epl, cs2), case-insensitive - status (string, optional): scheduled, live, paused, ended, postponed, cancelled, suspended, delayed, or unknown - starts_after / starts_before (string, optional): RFC 3339 kickoff bounds; a game with no published kickoff is excluded whenever either bound is set - limit (number, 1-100, default 20) - cursor (string, optional): opaque gms_v1_ cursor from next_cursor. It pins the page position, not a snapshot: the catalog is live, so a game added or removed between pages moves with it Returns: data, has_more, next_cursor, as_of, coverage and meta. Read coverage before reading an empty page: it names the sports and leagues this deployment serves and any scope whose source was unavailable for this read, so an unserved sport is never confused with a quiet night. A sport or status this deployment does not serve returns an empty page, not an error. status.state keeps postponed, cancelled, suspended and delayed distinct from a plain 'not live', and a kickoff in the past is never read as live on its own. Scores are the provider's own strings, because a set score and a run total are not both integers. Carries no price, no sharp-money split and no holder identity; those stay on get_market_snapshot, get_sharp_money_flows and get_edge_hunter.

Read-only list of webhook endpoints owned by the authenticated API key user. This does not create, update, verify, rotate, or delete webhook endpoints. Args: none Returns: List of webhook endpoint configuration records.

Search 0xinsider's editorial content by keyword. Returns matching learn articles, glossary entries, comparisons, research, and trading strategies with their canonical URLs. Args: - q (string, required): Search query, 1-256 characters before whitespace trimming - limit (number, 1-50, default 10): Max results Returns: List of matching content items with content_id, kind, slug, title, excerpt, and url.

Search prediction markets by keyword. Returns matching markets with status, category, and platform. Use this to find condition_id values needed by get_market_flow. Supports cursor pagination: pass next_cursor from the previous page with the same q, status and category to read the next page; stop when has_more is false. Args: - q (string, required): Search query - limit (number, 1-100, default 20): Max results per page - cursor (string, optional): Pagination cursor from the previous response's next_cursor - status (string, optional): Filter by status: "active", "closed", or "all" (default "all") - category (string, optional): Filter by category Returns: data (matching markets with id, condition_id, title, slug, category, platform, status), has_more, next_cursor.