Python Quickstart with pandas
Details
Docs navigation
- Getting started: Quickstart, Authentication, Rate limits, Errors and retries
- Guides: Symbol formats, Timestamps and time zones, Pagination, Split and dividend adjustments, WebSocket streaming, Dataset downloads, Excel and Google Sheets, Changelog
- API reference: Stock bars, Stock trades and quotes, Snapshots and movers, Tickers, exchanges, calendar, Splits and dividends, Fundamentals and SEC filings, Options chain and Greeks, Options EOD, bars, OI, Forex, Crypto, Crypto derivatives, Futures bars and continuous, Settlements and COT, Indices, Economic data, Technical indicators
- SDKs: Python (you are here), JavaScript
Python plus pandas is how most of our early-access sign-ups plan to use the data, so this page gets you from an empty file to a DataFrame of stock data in Python in a few minutes. It uses plain requests, because an official client is still on the roadmap, and the whole helper fits in about 30 lines you can paste into a notebook.
Early access. The API is planned and not live yet. You can practise every pandas step below today on the free CSVs from /sample-data.
Details
Steps
1. Install two packages
pip install requests pandas pyarrow
pyarrow is optional; it lets you save to Parquet.
2. Store your key
export MARKETDATA_API_KEY="paste-your-key-here"
3. Write a small helper
Example, saved as md.py:
import os
import requests
import pandas as pd
API = "https://api.marketdata.co/v1"
SESSION = requests.Session()
SESSION.headers["Authorization"] = f"Bearer {os.environ['MARKETDATA_API_KEY']}"
def get_df(path: str, **params) -> pd.DataFrame:
"""Call an endpoint, follow every page, return one DataFrame."""
params.setdefault("limit", 50000)
frames = []
while True:
r = SESSION.get(f"{API}{path}", params=params, timeout=60)
if not r.ok:
err = r.json().get("error", {})
raise RuntimeError(f"{r.status_code} {err.get('code')}: {err.get('message')}")
body = r.json()
frames.append(pd.DataFrame(body["results"]))
if not body.get("next_cursor"):
break
params["cursor"] = body["next_cursor"]
df = pd.concat(frames, ignore_index=True)
for col in ("timestamp", "date"):
if col in df:
df[col] = pd.to_datetime(df[col], utc=(col == "timestamp"))
return df
4. Pull daily bars
Example (Python):
from md import get_df
msft = get_df("/stocks/bars", symbol="MSFT", interval="1d", start="2020-01-01")
msft = msft.set_index("date")
msft["ret"] = msft["adjusted_close"].pct_change()
print(msft[["close", "adjusted_close", "ret"]].tail())
Use adjusted_close for returns that include dividends, close for split-adjusted prices only. See adjustments.
5. Pull minute bars and resample
Example (Python):
spy = get_df("/stocks/bars", symbol="SPY", interval="1m",
start="2026-11-18T14:30:00Z", end="2026-11-18T21:00:00Z")
spy = spy.set_index("timestamp").tz_convert("America/New_York")
five = spy.resample("5min", label="left", closed="left").agg(
{"open": "first", "high": "max", "low": "min", "close": "last", "volume": "sum"}
).dropna()
Bars are stamped at the start, so resample with label="left" and closed="left" to keep the same convention.
6. Many symbols at once
Example (Python):
symbols = ["AAPL", "MSFT", "NVDA", "AMZN"]
closes = pd.concat(
{s: get_df("/stocks/bars", symbol=s, interval="1d", start="2024-01-01").set_index("date")["adjusted_close"]
for s in symbols},
axis=1,
)
print(closes.pct_change().corr().round(2))
7. Save it
closes.to_parquet("closes.parquet")
Sample data
Try it now on a real sample
The free SEC fundamentals sample loads the same way. Download us-company-financials.zip from /sample-data and run:
Works today on the free sample.
import pandas as pd
fin = pd.read_csv("us-company-financials.csv", parse_dates=["period_end", "filed_date"])
rev = fin[fin["metric"] == "revenue"].pivot_table(index="period_end", columns="symbol", values="value")
print(rev.tail())
The file covers AAPL, MSFT, NVDA, AMZN, GOOGL, META, TSLA and JPM since 2023. Source: SEC EDGAR.
Watch out
Notes and gotchas
- Keep
timestampin UTC for joins and resampling; convert only for display. Dailydatecolumns have no time zone. pd.DataFrame(body["results"])on an empty list gives an empty frame with no columns. Checkdf.emptybefore indexing.- Large numbers such as
volumearrive as JSON integers and load asint64. Fundamentalsvaluefields can exceed 100 billion; that still fits. - The helper reads every page into memory. For tick data, write each page to disk instead (see pagination).
- Respect the rate limits. The loop in step 6 makes one call per symbol; on the Free plan add
time.sleep(12)between calls.
Keep reading
Next steps
Details
Was this page useful?
[Yes] [No]
Want an official Python client sooner? Tell us on the custom data form.