Liquidity, order flow and microstructure metrics for crypto and RWA perpetuals. Point-in-time and research-ready.
Aperiodic turns hundreds of terabytes of exchange data into point-in-time metrics: book depth and liquidity, order flow, spreads and slippage, funding, basis and open interest. Quant teams backtest and research on them directly, 100–1000x faster than working from L2 data, without building the infrastructure.
aperiodic.io · Catalog · Indices · Markets · API docs · Pricing
Each index answers one question about a market, in one column you can read at a glance. Three families, nine indices:
| Family | Code | Index | The question | |
|---|---|---|---|---|
| Liquidity · what will it cost to trade? | CST |
Cost | What does crossing the spread cost? | Live |
DPT |
Depth | How much can you trade at the best price? | Live | |
IMP |
Impact | How far does $1M of buying move the price? | Live | |
| Flow · who is in control? | PRS |
Pressure | Who won this bar? | Soon |
CNV |
Conviction | Who has been winning, bar after bar? | Soon | |
| Positioning · what is being built? | OIB |
OI Build | Are positions being opened faster than usual? | Soon |
OID |
OI Divergence | Are positions moving more than price explains? | Soon | |
LEV |
Leverage | How big are open positions next to a day of trading? | Soon | |
CRW |
Crowding | Are traders paying to pile in? | Soon |
The Liquidity indices are live for every contract on aperiodic.io/indices, 15 minutes behind. Flow and Positioning are on their way.
Every dataset has a free preview slice (Binance BTC-USDT perpetual, 5m, May 2025). No account, no key.
pip install "aperiodic[polars]"from datetime import date
from aperiodic import get_metrics
df = get_metrics(
metric="flow",
exchange="binance-futures",
symbol="perpetual-BTC-USDT:USDT",
interval="5m",
timestamp="exchange",
start_date=date(2025, 5, 1),
end_date=date(2025, 5, 31),
preview=True,
)Or hand it to your agent. The MCP server works on the preview without a key, and signs you in with OAuth for the rest:
claude mcp add --transport http aperiodic https://aperiodic.io/mcpIt also works from claude.ai, Claude Desktop, ChatGPT and Cursor. Prefer the terminal? The CLI pulls the same slice as Parquet.
220 metrics in 19 datasets, from 1 minute to 1 day (15 and 30 seconds on Institutional).
| Datasets | |
|---|---|
| Trades | Flow (taker buy vs sell by trade size, with a toxicity score) · Impact (Kyle's λ, Amihud) · Slippage · Trade size · Run structure · Up/down ticks · Range · Returns and volatility |
| Order book | L1 price · L1 liquidity (spread, depth) · L1 imbalance · L2 liquidity and L2 imbalance at 5, 10, 20 and 25 levels |
| Derivatives | Funding · Basis · Open interest · Derivative price (mark, index, last) |
| Market data | Candlesticks · VWAP / TWAP |
Coverage. Every crypto perpetual on Binance, OKX and Hyperliquid, and 500+ RWA perpetuals on stocks, ETFs, commodities and FX (including Hyperliquid HIP-3 markets), under the same API and the same Atlas symbols.
Ways in. Parquet over REST · a live WebSocket stream · the Python client · the CLI · the MCP server · S3 and in-your-cloud delivery on Institutional.
Free, no key. Markets: every Binance, OKX and Hyperliquid perp, 15 minutes behind, with spread, depth, flow, funding, basis and Kyle's λ. Plus live funding rates for Binance, OKX and Hyperliquid.
Pre-flight. Test a trading agent on a recorded day of real order books before it trades real money: aperiodic.io/preflight.
Self-serve plans from $199 a month. Pricing · Talk to us about Institutional.
| Repository | |
|---|---|
client |
Python SDK: sync and async, concurrent Parquet downloads, Polars and pandas DataFrames |
cli |
Download any dataset as Parquet from the terminal; the preview needs no key |
atlas |
Securities master: one symbology for crypto and RWA contracts across exchanges, always up to date |
tparquet |
Parquet for time series: tick files at about half their ZSTD size, lossless, and still Parquet |
notebooks |
Notebooks to get started with microstructure data |
crypto-data-vendors |
16 crypto market data vendors compared on 40 criteria, every fact sourced and dated (we're one of the columns) |