This sample is provided for evaluating QuantFeatures data products under the QuantFeatures Sample Evaluation License (see LICENSE.md). Redistribution, resale, and production or live-trading use are not permitted without a separate agreement.

QuantFeatures Bybit Hourly Features (Sample)

Research-ready, hourly features derived from licensed historical Bybit market data: trade activity and flow, trade distributions, top of book, funding, open interest, and liquidations. This sample lets you evaluate coverage, schema, and data quality before licensing the full product.

Attribute	Value
Venue	Bybit
Instruments	BCH, BNB, BTC, ETH, HYPE, XRP (quoted in USDT)
Markets (splits)	`perp` (perpetual)
Sample days	First UTC day of each month, 2020-06-01 through 2026-10-01
Interval	1 hour, 24 intervals per sample day
Feature version	`3.0.0`
Rows	150,960 across 17 files

Some instruments start later because the contract was not yet listed: BCH perp from 2021-01-01; BNB perp from 2021-07-01; ETH perp from 2020-11-01; HYPE perp from 2025-01-01; XRP perp from 2021-06-01.

Load

from datasets import load_dataset

perp = load_dataset("QuantFeatures/bybit-hourly-features", "TradeFlow", split="perp")

Or with pandas:

import pandas as pd
df = pd.read_parquet("hf://datasets/QuantFeatures/bybit-hourly-features/data/TradeFlow/perp.parquet")

Modules (configs)

Config	Family	Input	Splits	Columns	Rows
`TradeFlow`	Trade activity	Exchange trades	perp	45	8,880
`TradeTick`	Extended trade activity	Exchange trades	perp	33	8,880
`TradeCovar`	Trade relationships	Exchange trades	perp	30	8,880
`TradeCovarPlus`	Extended trade relationships	Exchange trades	perp	54	8,880
`TradeStats`	Trade distribution	Exchange trades	perp	24	8,880
`TradeGroup`	Trade grouping	Exchange trades	perp	123	8,880
`TradeGroupPlus`	Grouped trade distribution	Exchange trades	perp	124	8,880
`TradeLarge`	Trade tails	Exchange trades	perp	123	8,880
`TradeGradient`	Trade dynamics	Exchange trades	perp	42	8,880
`TradeImpact`	Trade dynamics	Exchange trades	perp	42	8,880
`QuoteBasic`	Top of book	Best-bid/offer quotes	perp	71	8,880
`FundingRate`	Funding	Derivative ticker feed	perp	33	8,880
`LiqFlow`	Liquidations	Liquidation feed	perp	15	8,880
`LiqStats`	Extended liquidations	Liquidation feed	perp	77	8,880
`OpenIntBasic`	Open interest	Derivative ticker feed	perp	33	8,880
`OpenIntPlus`	Extended open interest	Derivative ticker feed	perp	45	8,880
`OpenIntGradient`	Open-interest dynamics	Derivative ticker feed	perp	39	8,880

Full field definitions are in the Feature Catalog.

Record semantics

Each row is one instrument in one half-open UTC interval [start_time_utc, end_time_utc).
exch, type, base, and quote identify the instrument; rows are sorted by start_time_utc, then base.
An interval's information is complete only at end_time_utc. Use end_time_utc as the availability time to avoid look-ahead bias.
NaN means unavailable or not validly computable, never zero; there are no nulls. Integer counts are always defined. A missing row is not zero activity.
Sample days are not contiguous; stateful and long-lookback features can be less reliable early in each sample day.
Prices are in the quote currency (USDT), volumes in the base asset, and _bps fields in basis points.

See the Data specification and Coverage for details. manifest.json lists row counts, coverage, and a SHA-256 checksum for every file.

Intervals with limited or no data

Exchanges and data recorders have outages and maintenance windows. In these sample intervals the source feed had no records:

LiqFlow and LiqStats, 2020-06-01 to 2020-10-01 and 2020-12-01: no liquidation records exist for these days, so zero counts there mean no data, not no liquidations.

License

QuantFeatures Sample Evaluation License (LICENSE.md): you may use this sample internally to sample and review QuantFeatures data products. Redistribution, resale, production use, reconstructing underlying market data, and reverse-engineering the feature methods are not permitted without a separate agreement. No raw exchange or vendor data is included. Not investment advice.

Full data

Continuous history and real-time features for Hyperliquid and Binance are available on request:

Delivered datasets: we source the tick data and deliver features at intervals from 10 minutes to 3 hours.
Your own tick data: if you license Tardis data, we compute features at any interval from 30 seconds to 1 hour.

Real-time features run on your own infrastructure with your own data feeds.

Contact

Full history, other intervals or instruments, and commercial licensing: QuantFeatures@gmail.com

Downloads last month: 11

Total file size:

50.7 MB