Early access is open. Free sample datasets are live now. Download a sample

Data product · Early access

On-Chain Data API: Exchange Flows and Network Metrics

How many bitcoin left exchange wallets yesterday, and is that number moving the same way as price or against it?

On-chain metrics are planned for later. They need our own blockchain nodes and indexing, so they're not offered at launch. You can reserve early access now; the requests tell us which chains and metrics to build first.

Coverage

What you get

Item Detail
Product code CR-ONCHAIN
Coverage Major chains, starting with bitcoin; metrics by asset
Granularity 10m, 1h or 1d intervals
History Back to each chain's early blocks where the metric allows
Delivery REST time series
Update speed After each interval closes

Everything here is read from the blockchain itself, not from exchange order books. That's what makes it different from our price data. One metric, one value, per interval. Exchange flow metrics rely on knowing which addresses belong to which exchange, and that labelling is an estimate. We'll publish how we label wallets alongside the data.

Schema

Schema

Field Type Required Description
asset string yes e.g. BTC
network string yes Chain, e.g. bitcoin
metric string yes Metric code, e.g. exchange_netflow
timestamp string yes Interval start, ISO 8601 UTC
interval string yes 10m, 1h, 1d
value number yes Metric value
entity string optional Exchange or holder group, e.g. binance
computed_at string optional When the value was calculated, for point-in-time backtests

Example record:

{
  "asset": "BTC",
  "network": "bitcoin",
  "metric": "exchange_netflow",
  "timestamp": "2026-11-18T00:00:00Z",
  "interval": "1d",
  "value": -4218.37,
  "entity": "all_exchanges",
  "computed_at": "2026-11-19T00:12:41Z"
}

The value is an example. A negative exchange_netflow means more coins left exchange wallets than arrived.

Sample data

Sample data

There's no free on-chain sample yet, because we haven't built the indexer. Request the metrics and date range you'd test with, and we'll send a file when the first chain is ready.

Use cases

Use cases

Exchange outflows as supply signal. Coins moving off exchanges are often going to cold storage, which some traders read as less supply for sale. Plot a 7-day sum of exchange_netflow against BTC price and check how often the two agree.

Inflow spikes before selling. Large positive value readings for a single entity can show coins arriving at one exchange ahead of a sale. Watch the 10m interval around big price drops.

Point-in-time backtests. On-chain metrics get revised when wallet labels improve. The computed_at field lets you filter to values that existed on a given date, so your backtest doesn't see labels it couldn't have known.

Network activity. Metrics like active addresses track usage of the chain. Compare their trend with price over months, not days; daily readings are noisy.

Code

Code examples

Example: daily BTC exchange netflow with curl.

curl "https://api.marketdata.co/v1/crypto/onchain?asset=BTC&metric=exchange_netflow&interval=1d&start=2026-01-01" \
  -H "Authorization: Bearer $MARKETDATA_API_KEY"

Example: Python, a point-in-time 7-day netflow series.

import os
import requests
import pandas as pd

resp = requests.get(
    "https://api.marketdata.co/v1/crypto/onchain",
    params={"asset": "BTC", "metric": "exchange_netflow", "interval": "1d",
            "start": "2026-01-01", "end": "2026-11-18"},
    headers={"Authorization": f"Bearer {os.environ['MARKETDATA_API_KEY']}"},
)
resp.raise_for_status()
flow = pd.DataFrame(resp.json()["results"])
flow["timestamp"] = pd.to_datetime(flow["timestamp"])
flow["computed_at"] = pd.to_datetime(flow["computed_at"])

# keep only values computed within two days of the interval start, as a live trader would have seen them
live = flow[flow["computed_at"] <= flow["timestamp"] + pd.Timedelta(days=2)]
print(live.set_index("timestamp")["value"].rolling(7).sum().tail())

Pricing

Pricing

On-chain data is not offered at launch and has no price yet. It won't be sold as a dataset at first.

Vendors we studied charge $35 to $375 a month for entry plans that include on-chain data, median $42. Commercial use at the specialist sellers starts around $584 a month.

Side by side

How we compare

7 of the 50 providers we studied sell on-chain metrics, and a couple of specialists do it in depth. We won't match their metric count early on. What we plan to do differently: on-chain series in the same API and key as your crypto prices and funding rates, a computed_at on every value for honest backtests, a published wallet-labelling method, and history you can download and keep.

FAQ

Frequently asked questions

Which chains will you cover?

We plan to start with Bitcoin, then Ethereum. Further chains depend on early access demand.

How do you know which wallets belong to exchanges?

Through address clustering and labels, which are estimates. We'll publish the method, and computed_at shows when labels changed a value.

What time zone are timestamps in?

UTC. Each timestamp is the start of its interval, so a 1d row covers 00:00 to 24:00 UTC.

Is it real-time?

Close to it. The shortest interval planned is 10m, and each value appears after its interval ends and blocks are confirmed.

Do values change after publishing?

Some do, when wallet labels improve. That's why we keep computed_at, so you can see the revision history.

What format does it come in?

JSON time series from the REST API. File downloads may follow.

Keep reading