First off I want to thank you for joining me in my endeavor to explore the potential of the science of statistical analysis in hockey. It is important for me to convey to the reader that I do NOT advocate for statistical analysis to ever REPLACE the traditional, hands-on method of player analysis currently in use in the NHL.
Instead, these statistical analysis will prove to be useful in augmenting scouting reports when NHL Managers finally understand the usefulness of “advanced statistics” and start to apply both methods when evaluating players in terms of trades or free agency. I imagine most of you are at least familiar with statistical analysis from either my first blog or from its prevalent use in the sport of baseball.
Therefore I would like to take this opportunity to address the “old school” hockey fans that are likely to dismiss the ideas I intend to put forth in this column by falling back on the argument of “math can’t quantify certain characteristics of hockey such as chemistry” and other similar positions. With that in mind, here we go.
As fans, we think of hockey as a game or a sport and that is why it is easy for us as fans to vehemently oppose something like statistical analysis. Statistical analysis is math-based science and math-based science is cold and unemotional. Fans are emotional and don’t like to look at our sports through cold and unemotional lenses.
To the owners and managers in the NHL, this is a business and they should run it as such. Players should not be considered as just employees, they should be viewed as assets. In my company, I am an employee and an asset (so to speak) even though I can’t be traded.
Technically I am a free agent every day but my company doesn’t have a salary cap, though it has a salary structure. Yes there are some differences between the company I work for and an NHL franchise but in many ways they are the same. For the company that I work for it’s all about controlling costs, increasing sales and measuring performances and in order for NHL franchises to be successful they must do the same.
Statistical analysis attempts to attach metrics to performance and studies trends just like most ordinary businesses.
Managers in the NHL should be conscious of managing their assets (players) just like any other business would. Neither a typical business nor an NHL franchise should make a habit out of investing in declining assets. It’s all about buying low and selling high.
Former Dodgers executive Branch Rickey may have put it best when he said, and allow me to paraphrase, that it is better to let a player go a year too early than it is to let a player go a year too late. The hard part is determining when a player reaches that declining stage.
Scouting is obviously one way to help determine when a player is in a decline. Most scouts can recognize when a player’s skills begin to diminish. Unfortunately, you can’t always get scouts to agree about a player. How many great NHL’ers have almost gone completely unnoticed by scouts, were not drafted highly or were not highly sought after as undrafted free agents?
Another area in which scouting can only help some is in determining the real worth of a player in dollars and cents. The salary cap has turned the world of the NHL upside down. With a salary cap in place it is imperative that NHL managers correctly assess the true value each player has before awarding them rich contracts. One or two overpaid players can kill a team and put them in a position where they have no flexibility to improve their roster.
Paying a player in decline based on their performance level of past years can handicap a franchise as well. Teams must establish and adhere to their unique salary structures or pay the price. This is where statistical analysis can help.
Ask any executive of any business whether they can ever have too much information when making an important decision. Chances are they’d want any and all pertinent information no matter how much. Statistical analysis is just additional information to be used with scouting reports to help NHL Managers make better decisions.
Tom Lynn is the former assistant GM of the Minnesota Wild. Currently he is doing some work on The Hockey News as a blogger. One of his more recent blogs talks about the Wild’s decision to move Goaltender Dwayne Roloson at the 2006 trade deadline. Once the decision to move Roloson was made, the Wild front office gathered as much information as possible about which teams were in need of goaltending help, the depth charts of those teams and also the depth charts of other teams that were possibly looking to deal a goaltender as well.
All of this info was reviewed to help determine which teams were the best possible matches with the Wild. The front office settled on Edmonton and Tampa as the teams that needed goaltending help and that both had assets the Wild were interested in and might be willing to part with those assets.
At that time, Edmonton was in a fight for its playoff lives and Minnesota knew how critical it was for them to qualify for the playoffs. Tampa on the other hand was coming off of their cup win and had seen Stanley Cup hero Nikolai Khabibulin depart as an UFA. Both teams needed goaltending but which one would give up the most?
After conducting all of this research and having already been offered just a fourth round pick for Roloson from Tampa, the Wild management could have simply called Edmonton and asked for a 3rd round pick. That makes sense; it’s more than a fourth round choice so they would have done well to make that move, right?
Actually, Wild GM Doug Risebrough and his staff were confident that Roloson was worth a first round pick to one or both of the interested teams and held out for that asking price. Ultimately, the Oilers ponied up with the No. 1 and rode Roloson all the way to that season’s Stanley Cup Finals. This is a fine example of management gathering as much information as possible to prepare them to make the best decision possible.
In baseball, where the science of statistical analysis is light years ahead of hockey in its development, GM’s utilize information gleaned through research to help them make important personnel decisions every day. An example of the usefulness of analysis in baseball was the contract extension that the Baltimore Orioles gave to second basemen Brian Roberts.
Roberts had been an all-star caliber player for several years but was entering the stage in his career where a decline could have been right around the corner. Before committing vast sums of money for a long term, Orioles GM Andy McPhail had his staff gather statistical information of players throughout baseball history who were “comparable” to Brian Roberts.
Some of the criteria used to determine a “comparable” is age, playing style, service time, injury history, production, etc. The purpose for doing this was to see historically how long comparable players were able to sustain their levels of production.
After extensive analysis, McPhail and the Orioles felt comfortable offering their second baseman a four year deal. How that decision pans out in the end is anyone’s guess but the Orioles are more likely to be satisfied with the results because of all of the in-depth analysis than not.
Comparables are a big factor with the VUKOTA projections that Puck Prospectus has developed. As I mentioned in my last post, Puck Prospectus recently introduced their VUKOTA projections which use historical comparables to gauge the potential performance of current NHL players. Presumably, these projections can be utilized for not only this season but for future seasons too. The VUKOTA projections would then be a potential tool to determine dollars and cents value of current players.
I guess I have just taken a long time to ask you, the reader, to give me and statistical analysis a chance before condemning me a fool and statistical analysis as foolish. With time I can open up a lot of eyes regarding the long term potential benefits of the science in the sport of hockey.