Crypto trading bot data API

A bot needs clean inputs from many places: candles from one venue, funding from another, liquidations from a third, options from somewhere else. Each has its own keys, limits and formats. f1ow gives you all of them behind one REST API and one WebSocket, with one key and consistent JSON.

Building blocks for strategiesCopy Anchor Link

Every tool page has the full parameter list, the request in cURL, TypeScript and Python, and a real response.

Backtest, then go liveCopy Anchor Link

Pull history with market_history, which takes up to 1,000 rows per call. Pass nextEndMs back as end to page backwards. Funding and OI history go back years at daily resolution:

import os, requests

H = {"Authorization": f"Bearer {os.environ['F1OW_API_KEY']}"}
rows, end = [], None
while len(rows) < 3000:
    params = {"metric": "funding", "interval": "1d", "limit": 1000, **({"end": end} if end else {})}
    page = requests.get("https://mcp.f1ow.com/v1/market/history/BTC", headers=H, params=params).json()
    rows = page["rows"] + rows  # each page is oldest-first; earlier pages come next
    end = page.get("nextEndMs")
    if not end:
        break

The field names in the response are documented on the tool page and typed in the OpenAPI spec.

Push instead of pollCopy Anchor Link

Polling every symbol every minute wastes calls. Subscribe over WebSocket instead. The server re-runs the tool and pushes only when the result changes, and only pushes are metered.

{ "jsonrpc": "2.0", "id": 1, "method": "subscriptions/subscribe",
  "params": { "name": "derivatives_open_interest", "arguments": { "symbol": "BTC" }, "intervalSeconds": 60 } }

Read-only by designCopy Anchor Link

f1ow never trades. Your bot keeps its own exchange keys and order logic; f1ow is the market-data and research layer. Snapshots carry generatedAt and coverage notes, so check freshness before acting.

For agents: this page as Markdown · all docs