Hockey Modelling Resources

Natural Stat Trick – https://www.naturalstattrick.com/

Provides downloadable/scrapable tables of player and team data from 2007-08 to the present
A TON of filtering options – date ranges, strength states, home/away splits, leading/trailing, etc.
They also have numerous “advanced” tools, like player comparisons, WOWY’s and D pair breakdowns

MoneyPuck – https://moneypuck.com/data.htm

Provides downloadable data for skaters, goalies, lines/defensive pairings, and teams from 2008-09 to the present
Also provides game level data for each team and unblocked shot attempt data (a very nice resource for people taking their first crack at building an expected goals model)

The NHL API – https://api-web.nhle.com/

The ‘rawest’ data, but also the best choice for integrating into your scripts. Natural Stat Trick and MoneyPuck data is just cleaned up/aggregated versions of data from the API
Has everything available, but can be tricky to use sometimes because there is no official documentation
Best documentation: https://github.com/Zmalski/NHL-API-Reference

NHL.com – https://www.nhl.com/stats

Provides downloadable/scrapable tables of player and team data sourced from the API
‘Basic’ data (no expected goals) but there are some interesting filter options
Downloads can be frustrating – you are only able to download 100 rows at a time
I use this page to find useful API endpoints – if you filter the table the URL changes and the same syntax as this URL is used in the API

This is a collection of interesting and useful articles that I have come across over the years while researching and developing my own models.

MoneyPuck.com – About and How it Works – Interesting and easily digestible information about how some of MoneyPuck’s models work

A New Expected Goals Model for Predicting Goals in the NHL – a detailed description of Evolving Hockey’s Expected Goals Model

Reviving Regularized Adjusted Plus-Minus for Hockey – a detailed description of Evolving Hockey’s RAPM model

Building an xG model – v. 1.0 – Hockey-Statistics – a very ‘different’ approach to an xG model that I really appreciated (this guy has a ton of amazing blogs and video tutorials on his site that are worth checking out)