> ## Documentation Index
> Fetch the complete documentation index at: https://docs.dune.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Hyperliquid HIP-4
> Curated datasets for Hyperliquid HIP-4 outcome (prediction) markets on Dune
HIP-4 is Hyperliquid's outcome-market standard: binary Yes/No prediction markets that trade on HyperCore order books next to Hyperliquid's perpetuals. Dune provides the full HIP-4 ledger — every fill leg, including the engine legs that mint, burn and settle outcome tokens — plus market metadata, hourly candles, hourly open interest, and daily account positions.
**Maintained by:** Dune · **Refresh:** hourly (`positions_daily` daily) · **Source:** Hyperliquid (HyperCore node fills and venue metadata) · **History:** from 2026-05-02
## Available Tables
One row per market: question, sides, event grouping, strike, expiry, and settlement
Every fill leg: trades plus split, merge, negate and settlement legs
Hourly OHLCV candles with VWAP, one series per outcome token
Hourly open interest and traded volume per market
Daily holdings per account and outcome token, valued at the end-of-day price
## Key concepts
* **Market = one binary contract.** `market_id` is Hyperliquid's outcome index. Every market has two outcome tokens, `yes_outcome_id` and `no_outcome_id`, in Hyperliquid's coin notation `#N`. `outcome_index` 0 is Yes and 1 is No. A Yes token pays 1 in the quote stablecoin if the market settles yes; a No token pays 1 if it settles no.
* **Events group mutually exclusive markets.** Where the venue groups markets into a question (for example, "Winner of English Premier League 2026/2027"), each member market carries the same `event_id` and `market_type = 'question_outcome'`. Standalone markets have a null `event_id`. An event can include a fallback outcome (`is_fallback`) that settles yes when none of the named outcomes happens.
* **Both sides have their own book.** The Yes and No tokens trade separately, and the book can also match a Yes buyer with a No buyer at prices that sum to 1. Prices are always the price of the token on that row, not the Yes price. On a No row, take `1 - price` for the implied Yes probability.
* **Quote stablecoin.** Markets are collateralised in USDC. The earliest markets used USDH. `*_usd` columns are denominated in the market's quote stablecoin.
## Notes
* `market_trades` has both legs of every match. Filter `is_taker AND fill_type = 'trade'` to count each trade once. Summing `amount_usd` over all rows double-counts volume.
* For volume at hourly or coarser grain, use `open_interest_hourly` (per market) or `ohlcv_hourly` (per outcome token). They already count each trade once and match the taker-filtered trades exactly, but scan far fewer rows.
* `ohlcv_hourly` has one candle per outcome token, so each market has two price series. To get the Yes series, filter `outcome_index = 0`.
* `open_interest_hourly` and `positions_daily` are built from the full token ledger (splits, merges, negations and settlements, not only trades). They exclude the venue's settlement counterparty (`is_system_account`).
* About a fifth of markets were listed before the venue published a market registry. Those markets have a null `template` and `title`. They are real markets, not an ingestion gap.
* HIP-4 is not yet part of the cross-venue [`prediction_markets`](/data-catalog/curated/prediction-markets/prediction_markets/overview) tables.
## Example query
```sql theme={null}
-- Top HIP-4 markets by traded volume, last 7 days
SELECT
m.market_id,
m.title,
m.event_title,
m.category,
SUM(o.volume_usd) AS volume_usd,
SUM(o.trade_count) AS num_trades
FROM hip4_hyperliquid.open_interest_hourly o
JOIN hip4_hyperliquid.market_details m
ON m.market_id = o.market_id
WHERE o.block_month >= DATE_TRUNC('month', CURRENT_DATE - INTERVAL '7' DAY)
AND o.hour >= NOW() - INTERVAL '7' DAY
GROUP BY 1, 2, 3, 4
ORDER BY volume_usd DESC
LIMIT 20
```