I debated whether or not I wanted to write this post. I’ve heard of the stat GVT (Goals Versus Threshold) and when I finally found available data, I was overjoyed. However, my hopes of a predictive catch all statistic were quickly dashed. My research on the stat led me to believe that this is a deeply flawed metric that measures player output and not player input. It’s still a stat that provides value, so let’s discuss it.GVT is an attempt to replicate something similar to baseballs WAR (Wins Above Replacement) stat. Instead of wins, GVT looks at how many Goals a player contributed to his team compared to what a replacement level player would have contributed. It is calculated by adding a skaters offensive plus defensive GVT numbers.
Offensive GVT is a calculation that uses goals, assists, and time on ice. A player is then rewarded a higher GVT for having more goals, assists, or playing time. It’s a little more complicated than that, but I’m just trying to simplify it for you. Now since this stat only looks at production, it doesn’t illustrate a players overall offensive talent. David Clarkson putting up a 30-goal season would produce a pretty good GVT. David Clarkson produced for his team. He added many goals, played a lot of minutes, and had a good GVT. We know David Clarkson isn’t a good player. That’s why GVT may be misleading to the general public. It doesn’t tell you who is better, but who produced more for that season. It also doesn’t compensate for teammates. Teuvo Teravainen isn’t going to produce many points while playing three months with Andrew Desjardins and Phillip Danault, which hurts his GVT. Likewise, Artem Anisimov is going to get a nice boost in GVT by playing with Patrick Kane & Artemi Panarin. If a player gets more power play time, they will get a boost in icetime and likely a boost in points, and thus a boost I their GVT.
So we start to see why this stat is flawed and not a very predictive stat. One does not use this stat to determine future success or a players overall talent. It is best used simply as a way to illustrate who produced for the team. But hell, if that’s the case, just look at Chicago’s point leaders and you will see who produced. I don’t feel like there is much to gain from looking at GVT. It’s interesting just to see who basically produced the most goals. But we know there is so much variance in hockey. So, so ,so ,so ,so much luck is involved with goal based stats. That’s why most people utilize shot based stats when analyzing players. Most sample size, less variance, more repeatability, and thus a better overall description of talent.
Again, remember offensive GVT mainly utilizes goals, assists, & playing time to determine goals above replacement. This stat is an all situations stat. It looks at even strength as well as power play and penalty kill numbers.
Defensive GVT utilizes shots on goal stats, a form of plus/minus, time on ice, and goalie numbers to determine how many goals a player was better than a replacement level player. That really doesn’t sound like the best way to try and analyze defensive talent. No corsi numbers which are simply all shot attempts are involved in GVT.
So overall GVT is offensive GVT + defensive GVT. In this post, I will produce for you offensive GVT, defensive GVT, overall GVT, GVT per 60 minutes, and Goals Versus Salary.
Here is chart number 1: Offensive Goals Versus Threshold.
I just want to note here that GVT is not a rate stat. That’s why we will have GVT per 60 minutes as out 4th chart. With GVT, with more icetime, one will be able to increase their GVT. If a player only plays a couple games, their GVT will remain closer to 0. This stat is about overall production. Think about the goal stat. A player that plays 82 games will likely have more goals than a player who plays 30 games. So games played and ice time will really influence GVT.
Chart number 2: Defensive GVT.
As I noted before, this isn’t a rate stat. If we looked at average goals above replacement per 60 minutes, things would change. Players with less games like Erik Gustafsson and Richard Panik will increase while players like Artem Anisimov and Andrew Ladd will decrease.
Here is the 4th chart: GVT per 60 minutes.
Finally let’s take a look at GVS, Goals Versus Salary. Instead of comparing goals above or below a replacement level player, we will now see goals compared to salary. A $5 million cap hit player’s GVS will be his goals above or below what a $5 million cap hit player should produce. Let’s look at the chart and I will describe it in more detail.
Artemi Panarin was on a pretty cheap deal. His +10.7 GVS suggests that he was the best value player on the Blackhawks last season. He provided 10.7 more goals than a player with a similar cap hit should produce. Patrick Kane had an enormous cap hit of $10.5 million, but yet he was still able to produce enough to be the second best value player on the team.
In the cap world, value is extremely important. The Blackhawks don’t win Stanley Cups unless they have players like Brandon Saad putting up top-6 forward numbers while playing on a cheap entry level contract. The best way to build a team is to get good value players on good value contracts. A $4 million player that performs like a $6 million player. A $925,000 player that performs like a $5 million player. This is the only way to create a super team. NHL teams can’t just buy players like the Yankees or Dodgers of the MLB. NHL teams must find good value players to be able to fit more talent under the cap.
Chicago did have some negative value players. Jonathan Toews produced 7.7 goals worse than a $10.5 million player. Bryan Bickell produced 7.5 goals worse than a $4 million player. This chart DOES NOT compare Toews to Bickell straight up. This is Bickell compared to his salary and Toews combed to his salary. So while Toews is worse than Bickell on the chart, it simply suggests that Toews was a worse VALUE than Bickell. This does not mean that Toews is a worse player. It’s much harder for Toews as he is being compared to a $10.5 million cap hit player. If Toews only had a $4 million cap hit, he would have a positive GVS and be a good value.
Brandon Mashinter made the league minimum salary last season and was still a poor value player. Let’s just hope he stays in Rockford this season.
I want to remind everyone again not to take too much from any of these charts. Don’t look at player A being higher than player B and think that means player A is a better talent that will produce more this season than player B. When you look at GVS, think of it as if you’re looking at last seasons point totals. In 2015-16, Andrew Desjardins scored more points than Richard Panik, but that doesn’t mean Desjardins is a better player or that he will outscore Panik again. Brandon Mashinter had more points than Marcus Kruger last season. Again, this doesn’t mean Mashinter is a better player or that he will outscore Kruger this season. That’s the type of mindset one has to have when looking at GVT. This is a stat that does not show who is more talented or who will produce next season. Instead, it’s like last seasons point totals. The stat shows who produced more and that’s about it.
GVT has been around since 2009 and it’s not very mainstream in the hockey stat world. Most people were probably like me. They read about it, realized it’s not very repeatable, and thus tossed it aside. There are more WAR-like stats, but none are considered good enough to become prominent yet. Or at least I haven’t personally seen a prominent War-like stat. GVT is probably the most mainstream one I know of.
I imagine a prominent War-like stat will soon be upon us. As you read this, there are hockey statisticians working on creating one. So while GVT might not be the best, it’s a nice starter stat to learn as we start to enter the “one number for everything”, catch all stat era.