PuckAPI for Python

Python client for the PuckAPI NHL data API. It has no required dependencies (standard library only) and works on Python 3.9 and newer. Pandas is optional.

Install

pip install puckapi
pip install "puckapi[pandas]"   # only if you want DataFrames

Set your key

Get a free key at https://puckapi.com/signup. Free accounts get 500 credits a month.

export PUCKAPI_API_KEY=your_key_here

from puckapi import Client

client = Client()                      # reads PUCKAPI_API_KEY
client = Client(api_key="your_key")    # or pass it in

Examples

1. Closing odds for a game

get_odds returns one row per bookmaker and snapshot. Ask for the closing snapshot and load it into pandas.

from puckapi import Client, to_dataframe

client = Client()

odds = client.get_odds(game_id="2026020080", snapshot_type="closing")
df = to_dataframe(odds, key="odds")
print(df[["bookmaker", "mlHome", "mlAway", "total"]])
print("credits left:", client.last_credits_remaining)

2. Who played the most minutes (shift data)

get_shifts returns every shift for every skater in a game, with start and end times.

from puckapi import Client

client = Client()

shifts = client.get_shifts(game_id="2026020080", team="TOR")
busiest = sorted(shifts["players"], key=lambda p: p["shift_seconds"], reverse=True)[:5]
for p in busiest:
    print(p["name"], p["shift_count"], "shifts,", p["shift_seconds"] // 60, "min")

3. A player's recent games

Find a player id with search_players, then pull the game log.

from puckapi import Client

client = Client()

match = client.search_players("Matthews", limit=1)["players"][0]
log = client.get_player_game_log(match["id"], season="20252026", limit=5)
print(match["name"], log)

Credits

Every method returns the data object from the response. After each call, client.last_credits_used and client.last_credits_remaining hold the counters from the X-Credits-Used and X-Credits-Remaining headers. Methods that take estimate_only=True return the price of a call without running or charging it.

Pandas

from puckapi import to_dataframe

df = to_dataframe(data)                # finds the main list in the response
df = to_dataframe(data, key="rows")    # or name the list yourself

Nested fields become dotted columns. Install pandas separately or with pip install "puckapi[pandas]".

Errors

Every error is an APIError subclass, with the server's message and the HTTP status.

from puckapi import Client, APIError, InsufficientCreditsError, RateLimitError

client = Client()
try:
    data = client.list_teams()
except InsufficientCreditsError:
    print("Out of credits, top up at https://puckapi.com")
except RateLimitError as exc:
    print("Slow down, retry after", exc.retry_after)
except APIError as exc:
    print(exc.status_code, exc.message)

The client retries 429 and 5xx responses and connection failures up to 3 times with backoff, and honors Retry-After. Set Client(max_retries=0) to turn that off.

If you see CERTIFICATE_VERIFY_FAILED on a python.org install for macOS, run the "Install Certificates.command" script that came with Python, or pip install certifi (the client uses it when present).

Options

Client(
    api_key=None,                          # defaults to $PUCKAPI_API_KEY
    base_url="https://mcp.puckapi.com",           # point at another host
    timeout=60.0,                          # seconds per request
    max_retries=3,
)

client.call("tool_name", {"arg": "value"}) calls any tool by name, which helps when the API adds a tool before this package does.

Tools

Method names match the API tool names. Arguments are keyword arguments; required ones can also be passed positionally. Each method's docstring lists its arguments (help(client.list_teams)).

Full API reference: https://puckapi.com/docs

License

MIT