Hockey Prospectus, www.puckprospectus.com, is the hockey sibling in the famous Prospectus sports advanced stats analytics family. Some of their writers not only appear on the home site, they also do guest features on ESPN and other sports sites. They do a great job of analyzing the sport through the prism of advanced stats.
Tom Awad, one of the talented scribes and stat analysts on the site, developed a metric whose baseball comparable would likely be Runs Above Replacement-level (RAR). Awad called his player evaluation stat, “Goals Versus Threshold” (GVT) and with it determines a player’s on-ice contributions relative to a “threshold” or “replacement” player. (Before I go further I want to thank Tom for the help he has given me as I try to understand the GVT metric. Sometimes I just can’t grasp the nuts and bolts of these advanced stats. Tom has gone above and beyond in helping me to better understand.)
If you are a fan of baseball you may have heard of the RAR stat. This stat is a close comparable to GVT in that it represents a player’s value above a “replacement” or “threshold” player in a term (runs for baseball or goals for hockey) fans are familiar with. A “threshold” or “replacement level” player would be defined as the top AHL/minor league player in the organization or the highest ranking free agent on the market in-season (not a prospect) and carries a GVT or RAR value of zero. Basically the guy a club would go get in the event one of their regular players went down to injury or something to that effect. For Rangers fans think Chad Kolarik, Andre Deveaux or Kris Newbury.
The GVT metric takes into account the offensive, defensive and shootout contributions of skaters and the goaltending and shootout performances of netminders. The calculations are done separately based on position; forward, defense and goalie. That’s because the responsibilities associated with each position and the stats that accompany them are inherently different. Forwards tend to be expected to produce goals; the blueliners are tasked primarily with goal-prevention; and of course goalies are supposed to stop the puck from crossing their goal line.
The complex formulation also weighs ice time and the quality of that ice time. Obviously players who receive the most ice time have a better chance of accruing the type of counting stats that would help boost his GVT. Factoring the quantity of a player’s ice time serves to normalize his GVT. Just as important as considering the quantity of ice time is factoring for the quality of that ice time.
Clearly ice time is not created equally. For example; it is much easier to score goals while on the man-advantage than it is to score a goal while killing a penalty or at even strength. Thus the formula weighs the different types of ice time differently. It doesn’t punish a PK guy for allowing a goal as much as it would that player at even-strength or the PP. Conversely a player isn’t rewarded quite as well when scoring a marker on the PP as he would if scoring at even-strength or on the PK.
As part of a project I was working on at the time, I attempted to quasi-reverse-engineer the GVT stat so I could track players’ real time performances. Unfortunately, Hockey Prospectus has not been able to make that available to its readers… yet. Mr. Awad told me in an email that they hope to have the GVT stat live on the site by October.
Regardless, I quickly learned just how complex and comprehensive the stat is. I can’t imagine any potentially useful data that may have been omitted from this calculation. At the end of the day, I learned two things: One being that I had a love/hate relationship with Microsoft Excel; the other being that Tom Awad put a ton of work into the GVT metric. He accounts for offense, defense and shootout performance for skaters while factoring for a goaltender’s performance for regulation and in the shootout.
Here’s how some of our favorite Rangers ranked in GVT this past year and for comparative purposes I’ll include the finalists for this year’s Hart, Norris and Vezina trophies.
Ryan Callahan – 7.8
Brandon Dubinsky – 9.6
Marian Gaborik – 10.2
Artem Anisimov – 7.5
Brad Richards – 17.5
Mike Rupp – 1.4
Brian Boyle – 8.0
Brandon Prust – 6.3[jcol/]
Hart Trophy Finalists
Corey Perry – 23.3
Daniel Sedin – 26.2
Martin St. Louis – 20.4[/jcolumns]
Marc Staal – 8.7
Dan Girardi – 9.9
Ryan McDonagh – 4.6 (in roughly half a year)
Michael Sauer – 7.1[jcol/]
Norris Trophy Finalists
Nicklas Lidstrom – 16.1
Zdeno Chara – 12.8
Shea Weber – 12.7[/jcolumns]
Mats Zuccarello – 5.5 (in roughly half a season)
Derek Stepan – 7.0[jcol/]
Jeff Skinner – 13.9
Logan Couture – 16.1
Michael Grabner – 16.8[/jcolumns]
Henrik Lundqvist – 29.6
Tim Thomas – 40.0
Pekka Rinne – 36.5
Roberto Luongo – 27.3
What I find interesting is that the Rangers had seven players with a GVT between 7.5 and 10.2. Gaborik was the highest rated skater at 10.2 but he wasn’t close to any of the Hart candidates. This suggests that the Rangers were a balanced club but lacked players who performed at a truly elite level.
Given that Hank rated slightly higher than Roberto Luongo, it stands to reason perhaps he deserved a Vezina nomination rather than the Canucks backstop. Regardless, even GVT backs up the voters awarding of the trophy to the Stanley Cup winning Tim Thomas.
I was surprised that Zuccarello rated so highly considering he played just half a season. Since GVT is more-or-less a counting stat, it is feasible that if the “Norwegian Hobbit” had played a full schedule he would have finished with a GVT around 11; or twice what he earned in half the campaign. That figure would have led Ranger skaters.
GVT is the one player evaluation metric that is close to being perfect. Unlike PC, GVT doesn’t appear to overvalue goaltenders quite like Ryder’s PC metric and it’s hard to argue either with the sheer volume of data included in the formula or with the actual mathematics behind it.
[jbutton color="red" link="http://www.snyrangersblog.com/category/advanced-stats/" size="xsmall"]CLICK HERE[/jbutton] to see the Advanced Stats section.