November 06 2012 07:55AM
In his book The Signal and the Noise, author and statistician Nate Silver, who you've surely heard about the last couple of weeks, advises humans to become "less and less wrong" when making predictions. We need to gather as much information as possible to make consistently good forecasts, but, as importantly, Silver suggests that "you can benefit from applying multiple perspectives toward a problem".
A lot of the book deals with consensus and aggregation, and the way that market necessities aid humans in working together to solve problems. I thought it was interesting when, earlier this summer, a writer at Puck Prospectus named Rob Vollman, sent out a list of all the summer unrestricted free agent signings and simply asked a handful of statistical analysts to rate the signings.
As it turns out, there was a lot of disagreement. Usually, the disagreement is expected to be between one of two camps: "stats" and "scouts", the groups using divergent methods to forecast a future player's role or success on a particular team. However, there was disagreement in the "stats" crowd on the list. For instance, 8 of "us" liked the Ryan Suter signing for Minnesota, 10 who disliked it, and 5 who were neutral. 13 liked the Jason Garrison signing for Vancouver, but 7 disliked it and 3 were neutral. [Full list here]
Remember, this doesn't mean that the consensus thought Garrison would be better than Suter. Every dollar spent on a player is supposed to bring in an expected win value. It is more likely that Garrison lives up to 6 years of $4.6M a season than Suter does to provide a positive return on a 13-year deal worth $7.5M per.
Where are we going with this? Well, I'd like to see if we can form an educated guess, as a Canucks Army community, for how many points Henrik and Daniel Sedin will earn in their next full season.
Thom wrote a post yesterday that suggested that the Sedin twins were not on the decline, that Henrik Sedin's lowest point total since 2008 was not a reflection of his career trend as a 31-year old hockey player, but that overall circumstances worked in the twins' favour in the 2010 and 2011 seasons, resulting in a massive increase in points to the twins, and that this past season was simply a return to the mean.
That's fine enough, but I'd like to tie up some loose ends. For one, despite being known as the "stats guys" in this corner of the Canuck-ternet, I think a lot of readers may be surprised to learn just how much we disagree about certain players and teams and quantitative methods. Not to discourage Thom, because he does a lot of good work here, but seeing as he's the guy with the day job and I'm not, I'm sort of left to myself some days to copy and paste filler information into spreadsheets and see if anything clicks.
What does this have to do with the Sedins? I think other than the shot numbers, point numbers, and other things, there's enough information online to show that the Sedins overall production in the NHL has flat-lined, and it ain't going up anytime soon. For simplicity's sake, I'm going to use Henrik as the proxy for both Sedins here, since not only was he the only forward to play every regular season game between Lockout '05 and Lockout '12, but having followed the twins for more than a decade, my eyes have me convinced Henrik is the superior of the two.
Anyway, here's Henrik Sedin's career, looking at points per minute played (relatively easy to suss out from Hockey Reference) from season to season:
You see a peak in 2010, and two years of decline. But what's special about this? I think to the un-trained eye, it looks like the trend line is moving upwards.
That said, remember that the twins never broke out until late, the 2005-06 season playing alongside Anson Carter. By the fall of 2006, Todd Bertuzzi had been shipped out of town for Roberto Luongo and the twins were first liners. But something else happened: An overwhelming number of penalties were being called. While the median powerplay percentage increased from just 16.1% to 18.0% after the lockout, the median number of powerplay goals scored in the NHL went from 56 to 88. It took for a couple of years for referees to put their whistles away, but part of that did help elevate the Sedins to "elite" players.
This is the first year that Alain Vigneault appears to have seriously massaged the Sedins' zone-starts in order to maximize their offensive-zone opportunities. As a direct result, their puck possession game improved significantly and they were able to generate shots at a higher rate than in the two seasons previous. In addition to their increasingly favorable deployment and their higher shots rates, the Sedins saw their On-Ice Sh% spike to an outrageous 13+%. Basically the Canucks were scoring a goal every 7.7 shots they took with the twins on the ice.
As a result of this correlation of factors, the twins' even-strength production spiked. Henrik Sedin managed 83 points at even-strength that season (and won the Hart Trophy and the Art Ross), while Daniel managed 64 points despite missing nearly twenty games. Sounds sustainable.
Allow me to work under the heretical assumption that Henrik Sedin's 2010 year was an outlier and can't be repeated. If you take away the time in the NHL the twins' spent developing, look at the data from the start of their prime and eliminate the outlier season, you get this:
It's a bit cleaner. There are more downs than ups, and the trend line appears to be flattening out, or even declining, if we just visualize it mentally.
I'm not going to toss the trend line in there, partly because I don't think that I'll ever perfectly predict the amount of points Henrik Sedin will get in an 82-game season. I'm quite confident in saying that without the outlier seasons, both pre- and post-lockout, the Sedins haven't shown an improvement in their game.
Keep in mind this is points at all strengths, however. With fewer power plays called, it's reasonable to suggest that the twins are getting better at even strength to make up the difference on the powerplay. On the other side of the equation, they're getting more chances in the offensive zone than any other players perhaps in the history of the league.
But what about the zone starts, Cam?
Gabriel Desjardins tweeted out this season that the zone start phenomenon helped the Sedins out 7-9 points over the course of 59 games (Vancouver had just played its 59th game, a 5-2 win over the Edmonton Oilers. Sedin had two points in that game, a goal and an assist). That could be anywhere between 10 and 13 points over the course of the season.
Also, why points? Assists are fairly arbitrary and I don't like them and they factor prominently on points. Instead of only looking at Sedin's point production, why not the Canucks' offence when Henrik Sedin was on the ice? Surely, plays that don't show up in the scoresheet ought to count, if the Canucks are still maintaining a very good offence at even strength with Henrik out there.
Rob Pettapiece once tried to figure out how many points an extra offensive zone start was worth, eventually figuring out that an offensive player separated by 202 zone starts against a defensive player, would wind up being on the ice for seven more goals for. Using that knowledge, we can use the number of extra offensive zone starts Henrik Sedin saw to create the "Real" Goals For Sedin was on the ice for per 60 minutes in the last five seasons, or our "Behind the Net" seasons:
|Goals For / 60||Extra O-zone Starts / 60||"Real" Goals For / 60|
(In case you were wondering, I'm fine with using "goals for" instead of the more accepted "shots for" when it comes to the Sedins. Eric T with NHL Numbers has called him "a shot quality king", and setting up teammates for better opportunities is not only a huge aspect of his game, but something most other hockey players don't have.)
Sedin has spent his last two seasons below his "combined" score. Eliminate the outlier season wholly, and he's below his combined score for just the single year, but is on the decline since 2009. It's slight, but you wouldn't be able to notice it if you weren't looking at offensive zone starts, as we are here.
|Goals For / 60||Extra O-zone Starts / 60||"Real" Goals For / 60|
Obviously, the outlier season counts in the overall scheme of the things. The problem is that it's such a large peak. There are no group of sherpas so mystified by the size of Mount Everest that they don't respect the treacherous valleys of the surrounding Himalayas. The problem with Henrik Sedin is that the eye is naturally drawn to the outlier.
So here we go: I don't think that Henrik Sedin is necessarily going to age at the same curve as a different forward. I do agree with Thom that since the twins' game isn't that of a natural athlete, they can last longer playing a systematic, cerebral type of game. That said, they will be declining, but ever so slightly based on the factors I listed above.
I think there will be a few commentators who look at Henrik's point totals and see nothing but "112 to 94 to 81 to just 75!!! Is this guy a real leader???" You should ignore those people, since the cataclysmic drop is more like something from 82 to 81 to something just slightly below 80. I think anywhere between 70-80 points over an 82-game season is a reasonable expectation for a 32-year old Sedin.
That's assuming the conditions don't change. There probably won't be a spike in powerplays like there was coming out of the last lockout, and I'm not anticipating too many other rule changes, or any trends that would involve goalies playing worse, and I'm also assuming that Alain Vigneault goes a little easier on the zone starts. I think those were a big factor in extending the prime of the Sedins, but I think it's clear that after strong conditional seasons in 2010 and 2011 that led to a Sedin winning a scoring title, Vigneault has run out of tricks up his sleeve.
The Sedins are very good players, but they're human. Like all humans, they age, and like all humans who age, their abilities begin to wane with age. The more information I look at on these guys' careers, the more sure I am that the Sedins are closer to trending downwards than upwards. Maybe another year of data, to confirm that 2010 was a clear outlier, and I'll be a little more sure of myself, and a little less wrong when I make next season's forecast.