Got part way through “Stat Shot“, the new hockey analytics book by Rob Vollman, recently. It took longer than I expected to get to writing a review as the book is damn dense and I could not get through it in one sitting instead reading a chapter at a time and then doing some of my own thinking and research into the various concepts discussed.
I had been anticipating the release of this book for quite a while as it seems there is an appetite for a good hockey analytics book that could appeal to a wider audience, i.e. the general public. Plus, I wanted to review it here on my hockey blog as it is relevant to the types of things I like to write about.
As far as I can tell from following some of the discussions from writers & bloggers, a lot of the hockey analytics experts are fairly hard core in that they are fully converted to the value of analytics in hockey and have traveled far down the road of studying it (the book author started getting interested in it back in the year 2000 and has written a number of hockey analytics books before writing “Stat Shot”). Then there is a much bigger group, the general public, that loves watching the game and is curious and interested in learning more about analytics in hockey.
Side note: When the Erik Gudbranson for Jared McCann trade was made I really noticed the work of Travis Yost on TSN. He is one of their hockey analytics writers but the way he analyzed the trade really sparked my interest. A lot of charts and comparisons against the mean – what other players in similar positions produce analytics-wise. Canucks GM Jim Benning does not seem to be on board the analytics train yet, seemingly still old school in his approach to assembling his teams.
Back to the book, I don’t think this book really appeals to a wider audience, as I’d hoped it could. Many of the formulas that were demonstrated were somewhat proprietary and, for me, not very intuitive or interesting. Also, from what I can tell, a lot of the better quality hockey analytics information is online in blogs and at sites similar to the author’s (or google “hockey analytics”) and not in books and, therefore, reading a 300 page book is not really required to get a general understanding.
The book consists of about ten chapters discussing applying statistical analysis to everything from how to build a team in the salary cap era, if looking at a players junior numbers can tell us if he’ll be a good NHL player, who are the top face-off specialists, hitters and goalies. Also, detailed analytics around shots taken and the last chapter looks at the most one-sided trades in NHL history. It’s a bit of an assortment of hockey topics where analytics can be applied.
Of all the stats in this analytics area, I think that the Corsi /Fenwick number will become mainstream the soonest. These are complementary statistics and are calculated as the number of shots taken by your team divided the sum of your teams shots for plus shots against. Corsi, unlike Fenwick, includes in the shots for the number shots for that are blocked. For example, to calculate Corsi of a player that is on the ice when 100 shots are taken by his team and 80 against, the Corsi number is 55% (100 shots for divided by 180 total shots). A Corsi number above 50% reflects a player who is on the ice when more shots for are taken by his team than against. There are a number of websites that show Corsi % such as this one (see CF%; CF60 would be the Corsi over a 60 minute game). These types of calculations and numbers are fairly easy to understand.
The idea of directing more shots towards the net (not necessarily hitting the net) is a key indicator of an important metric and that is puck possession during a game. This is something former Flames coach Bob Hartley was criticized for – poor team possession numbers as it was said Calgary was chasing the play too much and not possessing or controlling the puck/play.
In this “Stat Shot” book, the author starts off interestingly enough talking about how the whole hockey analytics movement got started. It’s borrowed from baseball. There is a baseball writer named Bill James who back in the 1970’s and 80’s wrote a book called “Baseball Abstract“. James subsequently wrote a few more similar books on applying analytics to baseball and Oakland A’s GM Billy Beane got interested and applied the strategies to building his A’s baseball teams in the early 90’s. The movie “Moneyball” was created from Billy Beane’s analytics work with the A’s.
Vollman wrote his first book, “Hockey Abstract“, in about the year 2000 by following Bill James’ ideas. Some of the formulas are named the same but apply hockey stats instead of baseball’s. This idea of “Value Over Replacement (Player)” is a Bill James creation that, for hockey, Vollman named “Goals Vs Threshold“. Take a read of the explanation of those key stats and see if they are interesting to you, they were not particularly interesting to me. GVT is basically what a player is calculated to contribute to goal production vs what a calculated replacement player would. It’s the same as Wins Above Replacement (WAR) in baseball but in hockey wins are replaced with goals.
As far as the NHL scene, it was around the year 2014 when teams started to hire analytics people and the author made a good point recently that teams are using these stats as much for cost efficiency reasons as anything. Hiring a data person to calculate and pull semi management information reports is fairly inexpensive within the overall budget of the average NHL team and a few good analytics-based decisions can save an organization valuable dollars.
The other part I was interested in was how technology and software is permeating into sports and the NHL specifically (not related to the book). At the World Cup (that I’ve not been watching much of), they had puck tracking technology. Not sure if that was reported on TV or what was done with the info but I believe this is done by embedding a chip in the puck. This puck tracking technology has not moved forward to mainstream NHL games because the pucks with the chip cost about $200 per puck and that’s deemed too expensive to have in every NHL game. At the league-level, though, they are capturing more and more data about the game and that data will be used in analysis and decision making moving forward.
The John Chayka, as GM in Arizona, hiring showed their organization’s appreciation of the importance of hockey analytics at the management /decision making level. Chayka is only 27 years old and is a bit of an analytics whiz kid; he started the hockey analytics company “Stathletes“. The Florida Panthers are another team who are very on board with hockey analytics. As well, the Toronto Maple Leafs have a whole department of people under AGM Kyle Dubas doing analytics work. Their software partner is the company SAS.
Teams will have data analysts and database administrators (if they don’t already, not sure) for getting data to management on financial / stat-related info.
The Calgary Flames have Chris Snow, Director of Video and Statistical Analysis, who is a well-respected hockey analytics person. The Flames use a system called “Pucks” that he says about half the other NHL teams are using. Here he talks about “Pucks” but it’s just data capture software that includes video links to demonstrate different scenarios a player is in. Getting the statistical data involves a daily download.
The NHL has invested in SAP and their HANA database technology for their website’s advanced stats and this area will only grow more and more into the future once their data collection grows. Businesses or start-up’s that can offer software/technology cost savings at the league or team level may have an opportunity to sell to teams in the NHL and other junior / related leagues. SAP seems to be the company that is closest to this NHL software wave but I expect there will be room for other software analytics companies to play a part (SAP is a bit known for the high cost of their software as well). I heard today SAP and the NHL held a “Hockey Innovation Summit” in Toronto about the use of technology /software in the NHL.
This trend of software and technology in hockey is just getting started. There will be a lot more innovation in this area in the next 5 years for sure.