Skip to main content
Pull Hyperliquid HIP-4 outcome-market data under /v1/hyperliquid/hip4/*: outcome and question discovery, side-level order books, trades, candles, and open interest. Check each data type’s history start in Venue coverage.
HIP-4 mark_price and mid_price values are implied probabilities in the [0, 1] range. Do not treat them as USD prices, and never compare them with a dollar price for BTC, ETH, or a Spot pair.

Discover an outcome first

Each outcome groups a Yes side and a No side. List outcomes before you pull a book or trades:
This representative response omits meta.next_cursor; paginate only with the cursor returned by the actual response.
  • An outcome groups two sides: side 0 is Yes (coin #0), side 1 is No (coin #1). Pass that side coin, URL-encoded (%230), as the symbol on order-book and trade routes.
  • target_price and expiry define the binary; is_settled and status say whether it resolved. Re-check them with /v1/hyperliquid/hip4/outcomes/{outcome_id} or /v1/hyperliquid/hip4/outcomes/by-slug/{slug} to see when an outcome settles.
  • Use /v1/hyperliquid/hip4/questions and /v1/hyperliquid/hip4/questions/{question_id} when you start from question metadata instead of an outcome side.
Store the outcome identifier, side, and venue family with every record, so probability series can be traced later.

Request parameters

Outcome discovery (/v1/hyperliquid/hip4/outcomes): Side-level order-book routes take a HIP-4 coin id as {symbol}, for example %230 for the Yes side. Snapshot requests use timestamp and an optional depth (maximum 20). History requests use start, end, cursor, limit (default 100, maximum 1000), and an optional depth (maximum 20). Trade routes use their own start, end, cursor, and limit (default 100, maximum 1000) and do not take depth. Question discovery (/v1/hyperliquid/hip4/questions) uses the same bounded-list pattern as outcomes. Use /v1/hyperliquid/hip4/questions/{question_id} for one question, and /v1/hyperliquid/hip4/outcomes/by-slug/{slug} when the public question URL slug is the identifier you have. Exact parameters for every route are in the HIP-4 endpoint reference.

Response fields

Each outcome in the data array: Each side_specs entry has side (0 for Yes, 1 for No), name, coin (the #-prefixed id you pass as the symbol), and asset_id. The single-outcome route (/outcomes/{outcome_id}) also returns aggregated_oi with per-side open-interest contracts.

Side-level data

HIP-4 order books and trades use the same fields as the core Order books and Trades schemas (see the HIP-4 tab on each page); only the symbol form differs. Candles and Open interest carry the probability history and outcome-side positioning.
A side with no stored book, such as a side whose outcome has settled, returns 404 with error_code not_found.

Live and historical HIP-4

HIP-4 live WebSocket delivery covers hip4_trades, hip4_l4_diffs, and hip4_l4_orders. hip4_orderbook and hip4_open_interest replay stored history but do not stream live; read the current book and open interest over REST. All five HIP-4 channels support historical replay: the order book, trades, and open interest use timed replay, and the two L4 channels use checkpoint-anchored bulk replay from 2026-05-02. HIP-4 has no funding or liquidation route and no candle stream; its candles are available over REST as probability-history series. Outcome-side open interest begins May 2, 2026 at roughly 10-second resolution. Pass the outcome-side coin as the symbol: WebSocket subscriptions take it as returned, such as #0, while REST routes accept 0, #0, or %230 and report #0 in meta.symbol.
Connection and reconnect handling are in the WebSocket tab; WebSocket channels lists every family.

Export in bulk

HIP-4 side-level data exports under the standard schemas, delivered as Parquet with ZSTD compression: l2_orderbook ($3/GB, $5 minimum), l4_orderbook ($4/GB, $12.50 minimum), trades ($4/GB, $7.50 minimum), and oi ($0.50/GB, $2.50 minimum). Each file keeps the outcome and side context. Build a selection in the Data Catalog; columns and coverage keys are on Export schemas.

Next step

Check a probability series for gaps before it feeds a model with Data quality.
Last modified on October 6, 2026