I was looking at the goal difference numbers the other day and happened across something interesting. Edmonton is almost certainly going to finish 28th or 29th in the NHL. The Oilers, however, have a goal difference, net of empty netters and the asinine SO goals that the NHL includes in the official totals to pretend that offence is up, of -15. We’d usually expect that to put a team at somewhere like 77 points over a full season, plus whatever they can add in OT or the SO. Call it 85 points. Someone on Twitter pointed out that Montreal’s number is even better – they’re tied for 16th in the NHL in net goal difference with -7. They could plausibly get that to zero and pick 28th. Insane.
This got me wondering about how the best and worst goal differences in the NHL have changed since the new CBA. I’ve had a theory, which I’ve never gotten around to testing, that the range between best and worst would begin to narrow as time went on. Very simply, the idea is that under the old CBA, you could do an actual rebuild in which you didn’t bother spending money on guys who could play hockey but just flat out sucked. Teams did this. Under the new CBA, you can’t really do that – you’re forced to have a cap hit within $16MM of the ceiling. This should, in theory, sforce the hockey talent to spread out.
And it looks like theory is backed up by reality. What I’ve done is split the league into six teams by goal difference – 1-5, 6-10, etc. I’ve then started in 2003-04 and gone to present, charting the goal difference for those groups. You can see that it’s converged significantly since 2003-04. (I assume this goes without saying but the dark blue line is 1-5 in goal differential, etc.) It turns out that there has been something of a convergence in strength of teams since the lockout. In 2003-04, the best five goal differential teams in the NHL were +265, the worst five teams in terms of goal difference -333. This year, pro-rated to 82 games, it’s +254 and -225.
What’s interesting is how much more expensive it’s become to have a lousy hockey team. The change that’s taken place in goal difference is undeniable and large but…those teams still by and large don’t come close to the playoffs. In 2003-04, the average payroll on a bottom five goal differential team was $35.5MM. This year, it’s $58.1MM. A top five goal differential team in 2003-04 cost $52.8MM, on average. This year, the cost of such a team is $61.6MM. Call it a 6.4% increase, annually, for the bottom teams and a 1.9% increase, annually for the top teams. It’s a bizarre result of this CBA – owning a terrible hockey team has become much more expensive on a year by year basis.
You don’t spend that money without some sort of result and, as you can infer, lousy teams are indeed now less lousy than they once were. I’m not sure that spending an extra $23MM annually for a 21 goal differential improvement is really worth it, from an economic perspective though. It’s generally accepted, I think, that you make more money by winning more hockey games. Cutting the margins between great and terrible doesn’t really do it. I sort of wonder if the cap floor has been responsible for some of the problems that we’ve seen teams have under this CBA. Phoenix, Atlanta, New Jersey and Dallas have all had serious financial problems. Columbus asked the city to bail them out. There are rumours about other teams. Under the old CBA, you could respond to a bad team by stripping it down, running it cheaply and collecting talent. That’s no longer an option here – you just need to keep pouring money into it. Revenue sharing will cover that to a degree but I don’t know that it’s enough.
I wonder – and again, this is idle wondering, something I haven’t really checked – whether randomness starts to come into things a lot more as the margins between good and bad get smaller. It seems reasonable to expect that it would. As the margins get smaller because talent is more equally spread between teams because of financial constraints, there’s going to be less to separate teams outside of randomness. When this CBA came in, people were thrilled by the advent of parity; at the time, I kind of understood that to mean that all teams would have an equal chance to be good. What I didn’t realize was that it seems to have meant that virtually anyone could make the playoffs because talent is so equally spread between most teams that randomness will have a much more significant role in determining who makes the playoffs versus who has a horrible year.
It’s an odd sort of a thing for a league to aspire to but maybe it makes perfect sense if you’re operating a cartel like the NHL. Any system in which any team can develop any sort of a long term competitive advantage will be opposed by the majority. What will be acceptable is a system in which randomness governs pretty much everything. It’s the logical endpoint.