Skip to main content

Bulk access

Downloading whole windows​

/v1/windows.csv returns a complete market window as one CSV — one request per market, no pagination:

curl -H "X-Api-Key: $KEY" -o btc-5m-1793331300.csv \
  "https://polybk.com/v1/windows.csv?symbol=btc&timeframe=5&window_start=1793331300"

Walking a day efficiently​

import requests, time

KEY, BASE = "pb_yourkey", "https://polybk.com"
H = {"X-Api-Key": KEY}

wins = requests.get(f"{BASE}/v1/windows",
                    headers=H,
                    params={"symbol": "btc", "timeframe": 5, "date": "2026-10-30"}).json()

for ws in wins["windows"]:
    csv = requests.get(f"{BASE}/v1/windows.csv",
                       headers=H,
                       params={"symbol": "btc", "timeframe": 5, "window_start": ws})
    open(f"btc-5m-{ws}.csv", "w").write(csv.text)
    time.sleep(0.2)  # stay comfortably inside 300 req/min

Archive / Parquet​

Pro and Archive customers receive the month-partitioned Parquet tree — every column, every stream, compressed. Query it directly without any API calls:

SELECT slug, min(yes_bid), max(yes_bid), count(*)
FROM read_parquet('curated/v1/symbol=btc/timeframe=5m/**/*.parquet',
                 hive_partitioning = true)
GROUP BY slug;

See data schema for the column reference.