Was Corey Perry really the MVP last season? We'll get to that argument later.

The WHL and score effects

I went out to see the Kamloops Blazers Saturday night with a couple of friends. The 5-2 result in favour of the home team is rather elementary compared to what happened in the course of the game. I frequently butt heads with a guy I know out here. He’s a straight-up traditionalist when it comes to hockey and has no time for the more progressive theories that surround hockey.

So I’ve started making bets with him.

After the second period on Saturday night, with the Blazers leading the Edmonton Oil Kings 3-0 and outshooting them 37-7, (the shot count was probably generous to Edmonton. I don’t think they touched the puck in the second) I confidently stated that Edmonton would out-shoot Kamloops in the third period. How come?

Keaton Ellerby is the only Kamloops Blazer to make the NHL since I moved here.

My friend is vaguely aware of all of “my” theories, including the idea that a team coming from behind will generate more shots on goal. It’s not so much theory as it is fact that teams will play to the score and the leading team is more likely to give up more shots. Is this driven by strategy or psychology? I’m not sure and therefore can’t answer, but I can confidently state that, especially at higher levels of hockey, there is a noticeable effect on which team is taking the shots as another pulls ahead.

So, we watched as Edmonton scored two goals to pull it to within 3-2. With each goal, Kamloops played with a touch more urgency to re-gain their lead. With each goal, Edmonton’s next one became harder to achieve. This effect isn’t noticed by the casual crowd, mostly made up of old-timers who aren’t used to seeing their home team lose like they have in the past decade, so they hurled insults at the team, convinced that the Blazers were just dogging it after 40 minutes of hard work.

Kamloops’ JC Lipon scored on a rebound on a 2-on-1 to restore a cushion for the Blazers, and Chase Souto sealed it late with about five minutes to go. Still, the damage was done as I looked over to my confused friend. How did the Kamloops Blazers, leading 3-0 and outshooting their road-stricken opponent by more than 5:1 suddenly get out-shot in the third period 16-9?

Until we know for sure and coaches learn how to fix this trend, I will continue to make easy money by making the same bet every single time a team is dominating on the scoreboard after two periods. I don’t think I’ll keep making very many friends if I keep at this, however.

Oh, now we re-open the MVP debate.

Goals Versus Threshold

Justin asked me this weekend to explain to readers of the blog something called GVT, a statistic developed by Tom Awad and published by Hockey Prospectus. GVT is more a tangible beast compared to stats like Corsi or PDO, which require a bit of theoretical thinking and long explanations for why these things work. GVT is measured by easily-obtainable stats like goals, assists, plus/minus and ice time. Simply put, it is “goals versus threshold”: the total number of goals a player outperformed the next available replacement at his position.

I’ve written before that Corsi isn’t the be-all and end-all of hockey stats, and that the best players won’t necessarily have the highest Corsi. It’s just a stat, as is GVT, with its own benefits and drawbacks. GVT is a measure of production over performance, so designed to measure a player’s individual contribution toward’s the overall team success. If a player has a 10 GVT, for example, that means that he is 10 goals more valuable than the next available replacement player, so, for a first liner, the replacement is a second-liner. For a fourth-liner, it’s an AHLer. Without that player with 10 GVT, the team’s overall goal differential would theoretically be 10 goals lower.

Last year’s Hart Trophy, awarded by the hockey writers, went to Corey Perry. His GVT was 12.1, which means that he is worth 12.1 more goals over the course of his season than what a second line player would average in a first line role with identical minutes as Perry. While 12.1 GVT seems pretty low, keep in mind that:

  • Hockey is a team game, and lots of players are expected to make contributions. 12 goals is worth approximately 2 wins in the standings, which was the difference between playoffs and not for nine teams last season, so it’s still a lot.
  • Perry was the 25th most valuable forward according to GVT. With all his extra ice-time, he had more chances to beef up his goal and point totals.

GVT is calculated by estimating a certain value of events in four categories: offense, defense, shootout, and goaltending. Goaltending is goaltending, of course, and the calculation mostly involves weighting save percentage for the shots that a goalie faced. Goalies tend to dominate recent GVT lists given their importance: The top six players in GVT last season were all goalies: Tim Thomas, Pekka Rinne, Carey Price, Henrik Lundqvist, Jonas Hiller and Roberto Luongo.

Offensive is calculated by looking at goals and assists per minute, with goals weighted 1.5 times as valuable as an assist. Defensive values are assumed by looking at +/- and the amount of shots against, as well as penalty killing time. Shootout values are more straightforward, determined by which player sees success in the shootout.

GVT isn’t really a measure of how good a player is and doesn’t necessarily attempt to prove that. It’s to showcase contribution. Much like baseball’s WAR—or “wins above replacement”— the calculation remains unadjusted for luck and therefore relies on overall player production and not his performance. Since the statistics that are used have been tracked by the NHL since 1967, it doesn’t rely on modern techniques for player evaluation. Quality of competition or difficulty of minutes and even adjusting for good or bad luck isn’t factored in to the calculation. It’s just an objective view of production, and we can take it into consideration with many things.

Tim Thomas led all NHLers in GVT last season, worth 34.1 goals over a replacement goaltender.

Given the difficulty of Corey Perry’s minutes last season, I’d be comfortable arguing that he was more deserving of the MVP than GVT leader among skaters Daniel Sedin. What GVT instead argues is that Sedin’s production was worth more to the Canucks than Perry’s was to the Ducks, or, Sedin’s GVT, Awad put it in Hockey Prospectus “sums up all his pluses and minuses”.

The elements that make up a hockey game can be contested and debated until Chris Pronger decides to move the puck forward, but, at its very core, hockey is all about scoring goals and preventing the same. GVT, at its heart, shows us just how many goals a player contributed towards a team’s successes or failures. A difference in six goals is estimated to be worth one full win in the standings.

For a more in-depth look at GVT and far more than I could ever explain, I’d recommend a copy of the latest Hockey Prospectus annual. Meanwhile, the GVT standings for last season can be found at BehindTheNet.ca here. Fiddle around with them a little and try to find the players that excelled given their situation, or those whose overall contribution may be a little inflated given some extra ice-time in key situations, or who failed to back up their offensive game with defensive performance.

**CORRECTION**—As has been pointed out, the GVT numbers from BehindTheNet were last updated on March 7th. The full numbers, available with Hockey Prospectus, have Daniel Sedin leading among forwards with 26.2 GVT and Perry in second with 23.3 GVT.