Hockey Alumni Analysis: The Results

It always hurts when your team is scored on – it’s that much worse when the opposing player used to be on your favorite team!


This is Part 1 of a two-part post, which will answer the question of why it seems that often while watching NHL hockey, your team is playing a former teammate.  This part will focus on the results of the analysis.  Part 2 will focus on the technical details of how data science was used to create this analysis.  For now, I’ll mention that the data used is a snapshot, based on Wikipedia, and is a point-in-time analysis.  If teams move their players up from their farm teams, or lose players to injury, it won’t be reflected here.


My hometown team is the Buffalo Sabres.  After last being a playoff contender in 2011, the Sabres have struggled competitively, and had changes in ownership, coaching and most of the players.  It seemed that lately the Sabres were continually matched up against an alumni player (someone who was formerly a Sabre).  I wondered how often this occurs (or is it just noticeable when someone traded away previously gets a goal against the Sabres)?  What team had the most former players active on other teams?  Was there a team that didn’t have any former players active?


It turns out, there’s quite a wide range, and the Sabres are not the top of the stack in active alumni.  It ranges from a low of 8 players for the Detroit Red Wings to 33 for the Chicago Blackhawks.  The Buffalo Sabres come in at 27 in the top 10.  Almost half the teams have enough players to field an opening day roster, currently playing elsewhere (although some players are counted for multiple teams as they move around a lot).


Some teams have more than one alumni playing for a particular team, such as the Winnipeg Jets, who have 4 former Sabres on their roster at the time of this writing.  This means that the number of actual teams with an alumni play that your team might play against is lower than the total number of players outstanding.  The Pittsburgh Penguins edge to the lead here with 21 teams staffing a former Penguin.  Detroit is, not surprisingly, at the bottom with only 6 teams playing a former Red Wing.  Buffalo is a little further down the ranking, possibly due to the heavy placement at Winnipeg, with 16.


During the 82 game regular season, a team will play certain teams more often that others as the schedule is designed around a hierarchy where a team plays within-division and within-conference teams more often than without- teams.  By combining the 2016-2017 regular season schedule with the presence of at least one alumni, one can quantify how often this will actually occur.  Here the Penguins are the leader, with well over half of their games playing against a former Penguin.  In fact, over half of the NHL teams are scheduled to play a former play half the time, with the Sabres, the Penguins and 16 other teams scheduled with 41 or more regular season games.


With the anticipated activity around the spring trade deadline, the numbers will shift around a little.  The addition of the Las Vegas Golden Knights as an expansion team will be a point of inflection as well, due to the expansion draft process.

So in summary, it turns out that our Buffalo Sabres are not the most prolific feeder team to the league.  But, if you put the Sabres together with the Pittsburgh Penguins, then every other team in the NHL but the Dallas Stars and Tampa Bay Lightning have one of our rust belt alumni playing for them.  And the Stars have Lindy Ruff as Head Coach, former Sabre both as a player and as a coach, which we’ll count for this analysis.  So, on behalf of the Sabres and the Pens to the rest of the NHL, you’re welcome.  And to Tampa Bay, maybe after the next round of trades you’ll get on board and simplify the Venn diagram below.


Total counts are higher in the Venn than show above, as this includes the fact that the Buffalo Sabres is of of course full of current Buffalo Sabres, and same with the Penguins.


The Python code used to create this analysis is available at my Github.  I’ll have a post up shortly explaining the web scraping and data analysis in detail.  Go Sabres!




Free Machine Learning eBook!

Andrew Ng’s early access copy of his new book Machine Learning Yearning is now available!  You can sign up for it at the below link.  Add your email to the list and you’ll get a link to download the book – I just downloaded my copy.  Andrew is an influential driver of machine learning and AI, and also has a talent and passion for teaching other, so I’m looking forward to digging into this early access copy.

Who is Andrew Ng?  He’s Chief Scientist at Baidu, co-founder at Coursera, author of a Machine Learning MOOC I mentioned in an earlier post, and professor at Stanford.