To wrap up another summer of Expected Learning (A much shorter summer after a busy June and July), we will have one more history class. After looking at Matt Barlowe’s public work earlier this month, we will now see what we can learn from Sabres Data Scientist Dom Galamini’s public work.

John Vogl wrote a pretty good article about Dom’s hire in an interview with Sam Ventura back in October which highlights Dom’s background working with the Hamilton Bulldogs while he was an undergrad at McMaster University. He also worked with Meghan Chayka at Stathletes. The Hamilton Spectator profiled Dom back in 2015, which also highlights a lot of the themes that Ventura discussed in the praise of his new hire, centered around strengths in the realm of translating data in a way that communicates trends in a digestible way to coaches and front office members.

If you were familiar with hockey analytics in the mid-to-late 2010s, the name MimicoHero may ring a bell, as that was his Twitter handle for most of those years and also the alias he used on Hockeygraphs.

Unfortunately, of as is the case with a lot of sports hires, most of his public work has been scrapped. However, his work, namely through HERO Charts, remains available across different outlets.

HERO stands for Horizontal Evaluative Ranking Optic. Devin Slawson wrote an explainer for HERO charts in 2015 that highlights the important stats included in the charts, including TOI at Even Strength, Points per 60, and production and puck possession metrics of the player’s linemates. One of these chart comparisons from 2016 still exists on Dom’s Twitter, shown below:

These can also still be seen on sites such as Blue Seat, SB Nation (Hey look it’s Colin Miller), and CaliSportsNews (Hey look it’s Christian Ehrhoff). The only piece of his former blog, OwnThePuck, features an embedded Tableau report with stats from the 2015-2018 seasons to populate player evaluation graphs. For example:

The website DNVR also created its own Player Evaluation Tool (PET) that was influenced by HERO charts: which they introduced here.

Lastly, Dom has two HockeyGraphs articles that are still available. First from 2016, Quantifying the Importance of Handedness for D-Pairings, which highlights how pairs historically have performed better when they are each playing their strong side with a few examples of defensemen that can switch sides and still perform well (think Dahlin). Second, from 2018, Comparing Scoring Talent with Empirical Bayes. This is a method of comparing two players’ offensive performances. The Bayesian statistical method can be tricky to grasp (trust me on that), so I won’t do more explaining, but Dom explains the three main components of the concepts of the Likelihood Function, Prior Distribution, and Posterior Distribution very well. Essentially, think of this as going into an experiment with some level of idea of what a result or function would look like (prior), and combining it with a function that encompasses the probability of events given observed data (Likelihood – usually the trials of an experiment), and multiplying these two functions together to create a new understanding of the concept or function (posterior). In a perfect world, Bayesian statistics would be the cleanest way to make predictions, but obtaining good priors and good likelihood functions are usually challenging due to a lack of clean and unbiased data, so estimating with known functions is the typical way to create estimates.

As I said, it stinks that there isn’t more out there for these history lessons, but from what we can find on the ol’ internet, it’s reassuring that NHL teams are hiring really strong candidates to be part of their analytics departments, with the Sabres being no exception headed into 2022-23.