Sometimes serious TV game shows are intentionally anti-competitive: the most egregious example I can think of is the PBS children’s game show, “Where in the World is Carmen Sandiego?” The game was played as follows: Round One – Geography Trivia; Round Three – Geography Trivia; Round Two: Concentration. (If you’ve forgotten your children’s games, Concentration involves two players turning over cards – first randomly, then by memory – looking for pairs.) In other words, a game of chance was sandwiched between two games of skill. The point was obvious: the show’s creators didn’t want the best player to run away with the game - presumably viewers enjoyed seeing overconfident kids lose because the cards fell the wrong way for them in the second round.
What does this have to do with the NHL? There is no relationship between team winning percentage during the first 65 minutes of the game and winning percentage in the shootout. Tie games might as well be decided with a coin toss. Essentially, when the league introduced the shootout after the 2004 lockout, it turned professional hockey into a children’s game show.
How can this be? Clearly there are some NHL players with great shootout skills and some with exceedingly poor skills. But most shootout shots are taken by a small number of players – last season, roughly four players per team had four or more shootout shots. The skill spread between forward #1 and forward #4 is very small. And a player who performs very poorly in the shootout will likely drop out of the top three before he can run up a lot of misses – a rare exception is Marian Gaborik, who’s 1-for-15 overall. And, as Gaborik can attest, there is no relationship between being a great offensive player and being a great, or even good, shootout taker.
It might seem surprising that performance is so unpredictable and that there isn’t a big difference in talent among shooters. Among players with 20 or more career shootouts, Vyacheslav Kozlov leads with a 56% shooting percentage, while Stephen Weiss trails at 17%. But if we look at the top half of shooters in 2005-06 and 2006-07 (minimum 10 shots) and compare their performance to 2007-08 and 2008-09, we find that their shooting percentage drops from 50% to 33%. At the same time, the bottom half of shooters increase from 28% to 35%. In other words, if I told you that a player had a shooting percentage near the league average of 33%, he is equally likely to have scored on half his shots the season before as he is to have missed three out of four.
What about goaltending? Surely the vast gap between the league’s best and worst goalies must give certain teams an advantage in the shootout. If we look at the 15 best shootout goalies in 2005-06 and 2006-07, we find that their save percentage is approximately the same in the next two seasons – 73%. But the league’s 15 worst shootout goalies aren’t that much worse: they stop 61% of shots in 2005-06 and 2006-07, but improve to 67% the following two seasons. Again, the gap between the “best” and “worst” isn’t as large you might expect: truly bad goaltenders will end up on the bench or in the minors and simply won’t face enough shots to end up in our study. Our group of the 15 “worst” goalies is actually a group of 15 league average goaltenders, and their shootout performance as a group in 2007-08 and 2008-09 is right at the league average, making their teams less than half a win worse per season than the teams with the best goalies.
Normally, if the talent spread in one particular area of the game is too small to differentiate between good players and average players, we wouldn’t expect to see it have much of an impact in the standings, with teams clustered very closely around .500. But the shooting percentage in the shootout is high enough that we get results based on very small sample sizes, which lends itself to very extreme results that distort the league standings.
For example, in 2006-07, Carolina finished with a 34-34-14 record in regulation, while Tampa was 29-33-20. They each went 5-3 in OT, but Tampa was 10-2 in the shootout while Carolina was 1-5. Tampa made the playoffs; Carolina was out. If we had no OT, Carolina would have made the playoffs; if we just had OT, they make it too. But their playoff picture came down to a few dozen penalty shots – and as a result, they ended up 11th in their conference, nowhere near the playoffs.
But it’s not just luck that the shootout rewards: the more teams drag the game into the shootout, the better they do. In 2005-06, Vancouver went 34-32-16 in regulation; Edmonton was 28-28-26. They each went .500 in the extra frames. Which means Edmonton got more extra points – OTL and SOL – than Vancouver, giving them the 8th seed, while Vancouver finished 3 points back.
This second effect is by far the most damaging to the NHL. It was always an accepted truism in hockey that if you didn’t score, you couldn’t win. In the post-lockout NHL, if you don’t score – and you keep your opponent off the board – you’ll end up in 82 completely random shootouts and finish the season with 123 points, good enough to lead the league every single year. Trying to eliminate ties is a noble cause, and fans certainly like the shootout, but the NHL needs to get rid of any incentives for slowing the game down. Ending the game with a roll of the dice might decrease the gap between the best teams and the worst, but no professional league should have intentionally anti-competitive measures in its games.
Gabriel Desjardins is a contributor to Puck Prospectus and runs the statistical hockey sites Behind The Net Hockey and Behind The Net. You can contact him at: info at behindthenet.ca.