Hyperliquid hourly positioning data
gold_positioning_1h is hourly Hyperliquid open-interest and positioning data, derived from the full two-sided perp ledger: total open interest, long-vs-short account skew, per-side concentration, whale-cohort net position, and the taker flow behind each change. Download the history as Parquet or query it through the Tessera API.
Use it for: Tracking open-interest trends, gauging crowded or concentrated positioning, and spotting squeezes and capitulations from whale and taker-flow shifts.
- Grain
- 1h
- Columns
- 29
- Partition
- coin × month
- History
- October 2025 →
- Price
- Free, trailing month
Timing · time marks the
snapshot
A day of BTC, bar by bar
2026-11-16 00:00–23:00 UTC. Read straight out of the Parquet a free key downloads, at the dataset's own grain.
- Bars in the day
- 24
- Open interest (BTC)
- 17,743 – 19,302
- Accounts at close
- 39,615
- Whale accounts
- 3,962
The hours behind it
The first four hours of 2026-11-16, exactly as the Parquet returns them. The file holds all 24 hourly bars of the day for BTC.
Download it, or read it
One call returns a presigned URL per day object for the open month, and a single URL for a closed one. The file is Parquet: partition columns, one row per hour, every column in the dictionary below.
- REST reference for the route, the error shapes and the pagination.
- Python and Rust clients stream it into Polars or DuckDB without a local copy.
- MCP serves this schema through
describe_dataset, so an agent can read the columns before it queries.
HTTP
curl -s -H "Authorization: Bearer $TESSERA_API_KEY" \
https://tesseralytics.dev/v1/datasets/gold_positioning_1h/BTC/2026-11/downloadPython
import tessera
client = tessera.TesseraClient() # reads $TESSERA_API_KEY
df = client.read("gold_positioning_1h", "BTC", "2026-11")
print(df.head())Data dictionary
29 columns, every one documented. The same table is served by describe_dataset to an MCP client and
by /v1/catalog/gold_positioning_1h over HTTP.
Core aggregates
snapshot · as of timestampAttribution: what changed since the previous snapshot
window ending at timestampHead-count deltas versus the prior hour: how many wallets opened, closed, or flipped sides. This is the raw churn behind the positioning change, which is how you tell a quiet drift apart from a violent reshuffle.