L1 Ticks provides market microstructure data for Hyperliquid and Polymarket. Hyperliquid data is captured from L1 infrastructure and delivered in Zstd-compressed Apache Parquet; Polymarket raw data is captured from CLOB WebSocket feeds and delivered as hourly JSONL.ZST files.

Pipeline running — data updated daily

Institutional-Grade
Market Microstructure

Name: Hyperliquid & Polymarket Market Microstructure Data
Creator: L1 Ticks

Market microstructure data for Hyperliquid and Polymarket. Hyperliquid ships as Zstd Parquet; Polymarket ships as raw hourly CLOB WebSocket JSONL.ZST.

Get Free Sample → View Data Catalog

10 HL Perps

1 BTC HF Book

Raw PM WS Ticks

30B+ Rows Delivered

99.7% Uptime

2000+ Kaggle DLs

Daily Data Updates

What You Can Build

Real use cases. Not hypothetical demos.

Smart Money Detection

Track wallet-level order flow to identify informed traders. Cluster wallets by behavior, detect accumulation/distribution patterns before price moves.

⇄

Execution Quality Analysis

Maker/taker attribution with per-fill fees. Model slippage, measure fill rates, and benchmark execution quality across venues.

ML Signal Engineering

Block-sequenced features with causal ordering guaranteed. Build order flow imbalance, cancel ratios, and toxicity signals for price prediction models.

◈

Market Microstructure Research

20-level orderbook snapshots + funding rate dynamics across 10 assets. Study liquidity provision, OBI signals, and cross-asset correlations on a fully on-chain exchange.

▣

Raw Tick Replay

Replay compressed WebSocket tick files to rebuild your own bars, filters, joins, and event studies from the raw feed.

⇌

Orderbook Reconstruction

Replay the raw WebSocket tick stream to reconstruct market state, inspect liquidity changes, and audit bid-ask dynamics.

⌁

Custom Feature Engineering

Start from raw files instead of someone else's feature table. Build volatility, spread, depth, and imbalance features your way.

∑

Academic Audit Trails

Keep a raw, replayable archive for reproducible research on prediction-market microstructure and information aggregation.

Data Catalog

Every field verified. Every schema documented.

BTC ETH SOL HYPE SPX PAXG XRP LINK AAVE DOGE

Orders

Per-block ~2s

block_height Int64 timestamp_ms Int64 wallet String coin String action Enum status String (L1 native) is_buy Boolean price Float64 size Float64 order_type String tif String reduce_only Boolean order_id Int64 (100% populated) cloid String

1B+ events/day • 100% order_id coverage

Fills (L1)

Per-trade

block_height Int64 timestamp_ms Int64 wallet String coin String side String role Maker / Taker price Float64 size Float64 fee Float64 is_liquidation Boolean tid Int64 oid Int64

1.6M+ fills/day • wallet-level tracking

L2 Orderbook

60s snapshots

timestamp_ms Int64 coin String level 1–20 bid_price Float64 bid_size Float64 ask_price Float64 ask_size Float64 obi Float64

20 levels bid + ask • all 10 assets • OBI pre-computed

Funding + OI

5 min

timestamp_ms Int64 coin String open_interest Float64 funding_rate Float64 mark_price Float64 oracle_price Float64 premium Float64 day_volume Float64

1.4K+ readings/day • all 10 assets

Special Book Stream

BTC Microstructure Book

BTC only · sub-second

Standalone BTC high-frequency historical data for spread dynamics, queue depth, liquidity shocks, adverse selection, and short-horizon execution research. Binance context data is included as a bonus for cross-venue validation.

Queue dynamics Spread behavior Liquidity shocks Execution quality Binance bonus data

server_time Int64 timestamp_ms Int64 snapshot_id Int64 coin BTC level 1–20 bid_price Float64 bid_size Float64 bid_n Int32 ask_price Float64 ask_size Float64 ask_n Int32 obi Float64

BTC only • 20 levels bid + ask • order counts + OBI included

Raw WebSocket Firehose

Hourly JSONL.ZST

raw_line Exact CLOB WS JSON event_type book / price_change / other market Condition / market id asset_id Token id changes Raw delta array bids / asks Raw book levels when present

One raw format • one WebSocket message per JSONL line • compressed hourly

First-Party Capture

CLOB WS

Captured from Polymarket CLOB WebSocket subscriptions for active prediction-market tokens discovered through the market registry.

No outside scraper archives are packaged into the paid Polymarket product.

Source-controlled raw feed collection

UTC Hour Files

Replayable

folder ticks_ws/ file ticks_ws_YYYY-MM-DDTHH.jsonl.zst timezone UTC compression Zstandard

Simple retention and easy backtest window selection

Raw Delivery Only

No derived labels

No pre-computed ML features.
No supervised target bundle.
No opinionated feature engineering.
No secondary Parquet add-ons.

Your team owns the replay, cleaning, bars, features, and labels.

Researcher-controlled transforms and labeling

Load in One Line

Hyperliquid loads as Parquet. Polymarket replays as raw compressed JSONL.

orders_sample.py

import polars as pl

# Load 163M+ order events with 100% order_id coverage
orders = pl.read_parquet("orders/*.parquet")
btc = orders.filter(pl.col("coin") == "BTC")

print(btc.head(5))

shape: (5, 14)

block_height	timestamp_ms	wallet	coin	action	status	is_buy	price	size	order_type	tif	reduce_only	order_id	cloid
i64	i64	str	str	str	str	bool	f64	f64	str	str	bool	i64	str
0	1774396843090	0xce71..5a42	BTC	Rejected	badAloPxRejected	false	70525.0	0.12282	Limit	Alo	false	360336907743	0x0000..
0	1774396843090	0xf36c..e1a4	PAXG	Filled	filled	false	4548.0	0.56	Limit	Alo	false	360336907286	0x1490..
0	1774396843090	0x5483..a6b2	SOL	PlaceOrder	open	true	90.282	88.34	Limit	Alo	false	359298802930	0xc71b..

polymarket_raw_ticks.py

import io, json
from pathlib import Path
import zstandard as zstd

path = Path("ticks_ws/ticks_ws_2026-05-13T13.jsonl.zst")

# Stream one raw Polymarket WebSocket message without loading the full hour
with path.open("rb") as fh:
  reader = zstd.ZstdDecompressor().stream_reader(fh)
  for line in io.TextIOWrapper(reader, encoding="utf-8"):
    msg = json.loads(line)
    print(msg.get("event_type", msg.get("type")), msg.keys())
    break

raw JSONL line

event_type	market	asset_id	payload
str	str	str	raw JSON
price_change	0x4a8f..c2e1	7132..9481	{"changes":[{"price":"0.45","size":"200"}]}
book	0x7b2d..a9f3	4890..2217	{"bids":[...],"asks":[...]}

Delivery & Folder Structure

Private Google Drive delivery. UTC-organized files with source-specific formats.

Standard Hyperliquid Package

10-Market Order Flow Delivery

BTC · ETH · SOL · HYPE · SPX · PAXG · XRP · LINK · AAVE · DOGE

The standard package ships as one clean folder with stream-first layout. Each stream is partitioned by UTC date and time bucket, with a coin column covering all 10 supported Hyperliquid markets.

l1ticks_hyperliquid_YYYY-MM-DD_to_YYYY-MM-DD/
├── hl_orders/
│   └── YYYY-MM-DD/HH-MM.parquet
├── hl_fills/
│   └── YYYY-MM-DD/HH-MM.parquet
├── hl_book/
│   └── YYYY-MM-DD/HH-MM-SS.parquet
└── hl_funding/
    └── YYYY-MM-DD/HH.parquet

All 10 markets in one schema Zstd-compressed Parquet No JSON parsing

Standalone Research Pack

BTC High-Frequency Historical Delivery

BTC only · Binance bonus data

The BTC high-frequency package is delivered separately for microstructure research. The core product is BTC historical book data; Binance folders are included as bonus cross-venue context.

btc_high_frequency_historical/
├── hl_book/
│   └── YYYY-MM-DD/HH-MM-SS.parquet
├── hl_orders/
│   └── YYYY-MM-DD/HH-MM.parquet
├── hl_fills/
│   └── YYYY-MM-DD/HH-MM.parquet
├── hl_funding/
│   └── YYYY-MM-DD/HH.parquet
├── hl_candles/
│   └── YYYY-MM-DD.parquet
├── hl_snapshots/
│   └── YYYY-MM-DD.parquet
├── binance_futures_bbo/
│   └── YYYY-MM-DD/HH-MM.parquet
├── binance_spot_bbo/
│   └── YYYY-MM-DD/HH-MM.parquet
├── binance_futures_trades/
│   └── YYYY-MM-DD/HH-MM.parquet
└── binance_futures_liquidations/
    └── YYYY-MM-DD/HH-MM.parquet

BTC microstructure core Binance context included Monthly refresh access

Polymarket Raw Delivery

Single-Format WebSocket Tick Archive

JSONL.ZST only

Polymarket is sold as raw WebSocket files only. Each file is an hourly UTC archive containing exact CLOB WebSocket payloads, one message per line, compressed with Zstandard. No derived bars, labels, feature tables, outside scraper archives, or mixed Parquet formats are included in the paid raw package.

polymarket_raw/
└── ticks_ws/
    ├── ticks_ws_2026-05-13T00.jsonl.zst
    ├── ticks_ws_2026-05-13T01.jsonl.zst
    ├── ticks_ws_2026-05-13T02.jsonl.zst
    └── ticks_ws_YYYY-MM-DDTHH.jsonl.zst

Raw CLOB WebSocket UTC hourly files JSONL.ZST only No outside archives

Why This Data

Built for quants, not dashboards.

100% Order Lifecycle Tracking

Every order carries an exchange-assigned order_id. Deterministic place-to-fill linkage with no fuzzy matching. Track the full lifecycle: placement, rejection, fill, cancellation.

Wallet-Level Order Flow

Track individual accounts across executions. See who's placing, canceling, and getting filled.

Maker / Taker Attribution

Every fill tagged with execution role. Know who initiated each trade at the protocol level.

L1 Native Status Stream

14 granular status enums direct from the exchange engine. Distinguish accepted orders from ALO rejections, partial fills from full fills, triggered stops from manual cancels.

Parquet-Native

Zstd-compressed, columnar storage. One line of Python to load. No JSON parsing, no CSV headers.

Cross-Asset Coverage

Crypto majors, DeFi protocols, memecoins, and index perps are fully available on decentralized order books. This provides an incredibly rare dataset.

Deterministic Clock Alignment

Data is snapped to strict UTC clock boundaries with zero drift. Delivered in mathematically perfect daily grids (e.g., exactly 1,440 order files/day at 60s intervals).

Zero Midnight Leakage

State persistence routing guarantees a 23:59:59 payload never contaminates the next day's partition, saving your quants hours of manual resampling.

Single Raw Format

Polymarket delivery is intentionally one format: hourly JSONL.ZST files containing exact WebSocket messages.

First-Party Capture

The paid raw package is sourced from our CLOB WebSocket recorder, not bundled from outside scraper archives.

Replayable Tick Archive

One JSON object per line makes it straightforward to replay the stream, reconstruct books, and build your own bars.

No Mixed Historical Formats

We avoid mixing old Parquet archives into the Polymarket product so customers receive one predictable layout.

No Opinionated Labels

We do not bundle resolved target labels or modeling assumptions into the raw product. Your research team defines truth and labels.

No Pre-Computed ML Features

Raw-only delivery avoids stale or black-box feature engineering. Build spread, depth, imbalance, and volatility features yourself.

UTC Hour Boundaries

Hourly files make retention, auditing, and backtest window selection easy without inventing another partitioning scheme.

Audit-Ready Research Base

Keep raw source files as the reproducible base for future experiments, transformations, and academic audit trails.

Frequently Asked Questions

Everything you need to know about the data.

What is L1 Ticks?

L1 Ticks provides market microstructure data for Hyperliquid and Polymarket. Hyperliquid is captured from L1 infrastructure and delivered as Zstd-compressed Parquet. Polymarket is captured from CLOB WebSocket feeds and sold as raw hourly JSONL.ZST tick files.

How is the data captured?

For Hyperliquid, we operate L1 node infrastructure that reads native order status streams in real time. For Polymarket, the raw product comes from our CLOB WebSocket recorder and active market discovery. The paid Polymarket package does not include outside scraper archives or derived feature tables.

What format is the data in?

Hyperliquid packages ship as Apache Parquet files with Zstd level 3 compression. The Polymarket raw package uses one format only: hourly ticks_ws_YYYY-MM-DDTHH.jsonl.zst files, with one raw WebSocket payload per line.

What markets are covered?

Hyperliquid: BTC, ETH, SOL, HYPE, SPX, PAXG, XRP, LINK, AAVE, and DOGE perpetual futures — spanning crypto majors, DeFi, TradFi-bridge, and meme sectors.

Polymarket: Raw CLOB WebSocket tick archives for active prediction markets.

What is the BTC high-frequency historical package?

It is a standalone BTC research package centered on high-frequency L2 book history. It is designed for market microstructure work: queue depth, spread behavior, liquidity shocks, short-horizon alpha features, and execution-quality analysis. Binance futures and spot context data is included as bonus data for cross-venue comparison, but the core product is the BTC high-frequency book.

Is there a free sample?

Yes. Free public Kaggle samples are available for BTC high-frequency microstructure and 10-perp Hyperliquid order flow. Polymarket raw tick samples use the same replay format as the paid raw delivery.

Pricing

Start free. Scale when ready.

Free

Public Kaggle samples for BTC high-frequency microstructure and 10-perp Hyperliquid order flow.

BTC HF sample: book, orders, fills, funding
10-perp sample: orders, fills, book, funding
Raw UTC-partitioned Parquet files
No sign-up or Telegram required

BTC HF Sample → 10 Perps Sample →

Starter

$69 / month

Daily Parquet delivery for all 10 Hyperliquid perps. 30-day rolling access from your start date.

Daily Parquet delivery for 30 days
All 10 perpetual markets
Orders, fills, book, funding streams
Auto-renews for continuous access

Get Started →

Institutional-GradeMarket Microstructure