The League of Extraordinary Statisticians: The Great Debate

The League of Extraordinary Statisticians (LOES) is a weekly forum bringing together the top analytical minds in the hockey world to answer a variety of questions that straddle the line between stats analysis and something you might hear floating around section 304.  They have agreed to answer these questions in a few paragraphs or less, and with minimal formulae.  Because this is a forum, we'd encourage you to use the comments section to answer the questions yourselves, or to discuss or debate the answers given.

The LOES is not meant to represent the entire of the hockey stats community.  There are a number of people that either were too busy or too difficult to contact for the purposes of the forum.

Talk about a potentially taxing topic...if you were to think about the greatest teams in NHL history, you would likely think about the amazing offense of the Oilers dynasty, or the astonishing combinations of offense and defense in the Canadiens dynasties of the 1950s and 1970s, as well as the Islanders of the early 1980s and the Red Wings of the late 1990s and early 2000s.  This kind of recognition identifies "offense" as goals, "defense" as preventing goals, ultimately expressed in winning in the regular and playoff seasons.  Posing a question to the LOES like I'm about to do, concerning whether the offensive element is more or less (or equally) important than the defensive element, will not necessarily result in those definitions being taken for granted, nor should it.  Instead, it requires that we look a little deeper at what makes both offense and defense, and whether those things might even be separable.  Maybe they are, maybe they aren't...

This week's question: What is more essential for a successful NHL team, a "good offense" or a "good defense"?  Why?

Fair warning: in the hockey metrics world, this is a weightier question than it appears in the responses.  The answers are the result of a lot of studies on value, talent, etc.

Definitely good defense.  Generally speaking, good defensive teams consistently make the playoffs relative to good offensive teams, and there's a very good reason for that.

 
Assuming a good defensive team and good offensive team have an equal chance of winning a game, why would there be an advantage of being a good defensive team?  Because there's a better chance of a low-scoring game going to an OT/shoot-out than a high scoring game.  In OT/shoot-out you get points for losing.
 
The ultimate defensive team would go to an OT/shoot-out every single game, and get 123 points on average, since OT/shoot-out is essentially a coin toss.
 
Also, teams aren't very good at rating defense, making it far easier to get good defensive players than good offensive ones.

- Rob Vollman, Hockey Prospectus

A goal saved is a goal earned.  What is essential is talent, and it doesn't matter how it's split up between offense and defense.

- Tom Tango, tangotiger.net and author of The Book: Playing the Percentages in Baseball

From a team-building point of view, a good offense. A lot has been written in sabermetric circles about the variability of goaltending, but good defense in general is very transitory, in hockey as in many other sports. Case in point: last season, New Jersey and Buffalo were first and fourth in goals against per game. This season, they're 24th and 21st. This isn't to say that you might not be able to buy improvements to defense for less, though.

- Timo Seppa, Hockey Prospectus

Generally speaking, unless you're special, offense is everything in hockey.  No team will be successful trying to defend in their own zone.  The only safe hockey is played in the other team's zone.  For the most part, the teams that won the Stanley Cup did so because they were good at scoring goals.  However, team defense/structure can differentiate an average team from a good team (you still require a competent goaltender as well).

- Chris Boersma, Hockey Numbers

I don't think I can write much for this question except to say that offence and defence are equally important.  Hockey is a game of outscoring your opponent.  This comes from preventing your opponent from scoring as much as it does from scoring your own goals.  There is no reason to break symmetry between the two.

 
Some people make statements that offence wins games and defence wins championships, but they are not backed up by facts.  The Gretzky Oilers won championships and the Lemieux Penguins won championships.  The Langway Washington Capitals and the modern days Boston Bruins haven't.  
 
There just is no answer to this question.  Both are equal and that is the nature of the game.

- Greg Ballentine, The Puck Stops Here at Kukla's Korner

The numbers, of course, say that it doesn't matter either way.  But if you were building a team, you want a "good defense" that has shot suppression abilities.  It's cheaper to buy defensive skills because
of the over-emphasis on counting stats in the free agent market.  And in the playoffs, unless I had a dominant team, I'd prefer to play low-event games rather than high-event ones because it increases the
role of the luck in winning.  None of this considers team revenue and attendance, of course.  What may be optimal for a GM or coach may not be optimal for the organization.

- Gabe Desjardins, behindthenet.ca and, of course, Behind the Net

In other words, the answers are all over the board.  You will notice that offense and defense, in some cases, are still held as extricable concepts, as Corsi numbers and goal-differentials which are currently used by many hockey statisticians are presented as shots-for/shots-against and goals-for/goals-against, with either side representing offense and defense.  The format of the hockey rink lends to this a bit too, as in one zone you are simply far more likely to be trying to create offense and another far more likely to be trying to prevent offense than you are the other way around, and the extreme zones are bound by rules and strategies to concentrate play in either end.

Honestly, I think each answer kind of stands on its own as an interesting statement on the question.  Rob points out the value in the current NHL point system of playing to the guaranteed points, a strategy that can get you into the playoffs.  Gabe has noted a similar phenomenon in the value of playing to the tie.   Tom's noting that, really, any combination of the two can succeed if the overall talent is sufficient to win games; a Sabres team with zero offense came within two wins of a Stanley Cup in 1999, the Oilers teams of the 1980s had marginal defense but world-beating offense, and I mentioned some of the dynasties above that had a pretty decent balance of both.  Timo takes the team-building perspective, and what this approach notes (as Rob had also noted) is that defense can be assembled for less; in Timo's case, this means that emphasis can and should be placed on the offense, as it will be the talent level of the offense which will set a team apart beyond the "transitory" environment of NHL defenses.  Chris alludes to the advantage of a strong offense, that it reduces time spent in one's own zone, with the contention that a focus on defending well does not produce offense itself, which is necessary to win the game.  Greg says that there's no point in pulling the two apart, that they are both essential and you can look across the breadth of Stanley Cup winners and see that.  

One of the more interesting metrics in recent memory, PDO (shooting % plus save %) by Vic Ferrari, brings together two of those key elements of offense and defense; interestingly, it pulls strongly towards 1000 league-wide, and among your truly sensational teams you have very strong defenses just as frequently (if not more frequently) as you do strong offenses.

For discussion, I'd like to build a little on Gabe and Rob's points, and ask a question on the distribution of talent: assuming that the distribution of defensive talent holds over a number of years, would an emphasis on defense truly set a team apart so as to make the sacrifices on offense worthwhile?  In today's league, can hockey games be consistently and deliberately made into battles of attrition as they were in the Dead Puck Era (Gabe's noted playing to the tie, but what about an entire game)?  On the flip side, how much is too much of a sacrifice on defense?  As we come away from a summer where the goaltending market depressed, the defensive forward market depressed slightly, and the defensive defenseman market at least held (if not went up), are we moving closer to an idea of the minimum defense needed for a winner?

Have at it in the comments.

P.S.  And they certainly have.  Be sure to check out the comments below; there are a lot of good ones worth thinking about.