Hockey Projects

I am currently leading a collaboration to archive the majority of analytics projects in women's hockey that are under threat of being lost to the chaos of the internet. In the near future, we hope to host a one-day event to organize and support a set volunteers to hunt down women's hockey projects, data, and resources and archive them on Meta Hockey. We are also helping to revamp Meta Hockey to make it more accessible and suitable to the hockey analytics community's needs.

LINHAC Slides for this project

This fall, I hope to join the Carnegie Mellon University Stats in Sports group to work on some of my projects that need more support from the academic sphere, such as potential projects applying statistical mechanics to football and using information theory to analyze passing in hockey. I also hope to bring my expertise in physics to this group's existing football research to provide an alternate perspective that can be beneficial in interdisciplinary fields.

I also will be helping Rebecca Nugent, Sam Ventura, and Ron Yurko to put on the 2021 CMU Sports Analytics conference.

The Women's Hockey Analytics Conference occurred on July 10th, 2021, receiving about 400 unique viewers to the conference stream and averaging 50 viewers over the 8 hours of the conference. In hosting this conference, we aimed to give an effective platform to the growing number of women's hockey analysts and their projects. We helped facilitate collaborations between media members and women's hockey analysts through this conference, resulting in WHKYe and other novel projects.

We also hosted analytics-focused panels to help connect analysts, players, and media members. These panels featuring viewpoints of players and ex-players such as Colleen Coyne, Karell Émard, and Emma woods, as well as media members such as Erica L. Ayala, Gabs Fundaro, Nicole Haase, Marisa Ingemi, and Spencer Fascetta.

WHKYHAC hopefully will be an annual event going forward.

Archived presentations and data can be found on whkyhac.com and on the WHKYHAC https://www.youtube.com/channel/UCd_owNMk_0mkqSLnRdXVLiQ

With this novel project, we used several established metrics in conjunction with machine learning to identify data-driven player archetypes in the Big Data Cup women's hockey dataset. We found four forward archetypes, dependent, balanced, playmaker, and shooter, as well as three defensive archetypes, disruptor, two-way, and defensive. Particularly, we found in comparing men’s and women’s hockey, forward archetypes are very similar across the two types of hockey, but critical differences in women's hockey rules cause defender roles to be different.

We submitted this project to the Big Data Cup (paper), where we were named as Honorable Mentions for this project and presented our project at OTTHAC 2021. We also presented this project at the Ohio State University Sports and Society Institute Undergraduate Research Fair, where we were honored for our Outstanding Application of Analytics in this project.

An Nguyen made a comparison radar plot based on this project.

When this project was conceived, it is unknown at what age a typical women’s hockey player produces their “peak” performance at. The careers of female hockey players can be somewhat sporadic, with breaks for the Olympics, pursuit of other professions, and various life events. This further complicates the task of discerning the ages of top production. In this project, I was able to identify peaks in offensive and defensive ability at certain ages among women’s hockey players despite these career interruptions. Following previous studies on aging curves in baseball and men’s hockey, I used the most advanced CWHL measures available at the time with the “delta method” to construct an aging curve using twelve seasons of CWHL data. This aging curve is then analyzed to determine when, on average, the peak performances in player’s career occur should they exist. Ultimately, this project offered novel insights into the evolution of female hockey players and helped break ground for other aging curve projects, particularly Mike Murphy's NCAA aging curve project and Ben Howell's work on NWHL aging curves.

This project was presented at RITSAC 2020: (slides) https://docs.google.com/presentation/d/1Ej0b_gSgh5R93TE9l9gNBH62meV0AEifvqI2K5dg1FE/edit#slide=id.p, (video @ 4:41:00) https://hockey-graphs.com/2019/09/21/2019-ritsac-slides-and-video/

Though my first foray into women's hockey analytics is no longer publicly available, it was a part of an initial wave of projects in 2018 that put women's hockey analytics on the map. Being an exclusively visual project, it was also easy and succinct women's hockey content to engage with on a Twitter timeline, which I believe helped grow awareness of women's hockey. My event location map featured all available PHF (at the time NWHL) location data, which included shots, blocks, saves, turnovers and penalties, and could be filtered by a number of categories.

I presented this tool at SEAHAC in 2019: https://youtu.be/RlZs5Ue4Rpc?t=7899

This Tableau tool was written about on the Ice Garden: https://www.theicegarden.com/2018/9/21/17871558/nwhl-event-map-buffalo-beauts-metropolitan-riveters-connecticut-whale-boston-pride-analytics.