Thursday, October 23, 2008

The Lie of Averages

There’s an interesting post over on JT's blog concerning the perceived underachievement of the Canandiens’ ‘first line’ early in the season. Her take is an admirably rational response to something that shouldn’t even be an issue. The Canadiens are currently 5-0-1 in six games with a positive goal differential- fabricating points isn’t a problem. It is only the myth of the ‘first line’ that causes any such premature worry over the Plekanec-Kostitsyn-Kovalev ensemble’s as-yet-unimpressive numbers. Let’s get this out of the way: There is no first line in Montreal, and there likely won’t be this season. There are simply guys who play various forward positions in various configurations, and score goals. Sometimes those configurations will be stable. Sometimes they won’t be. Sometimes one trio will be hotter than another. If it makes you feel more secure, you can call whichever trio is currently hot ‘the first line’, but all it’s going to do is cause you sleepless nights when they get cold again. Why, oh why, isn’t the first line scoring???

The real question, I think, is not about why that line doesn’t seem to be starting off this season as hot as they ended the last, but whether there’s any reason to expect that they will ever be that hot again. True, it’s the same three guys on pretty much the same team and only one year older; the environment is as close to stable as it will ever be in modern hockey. The difficulty is that hockey players are not stable. If human identity is largely malleable over the course of a lifetime, a hockey player’s on-ice persona- and especially his statistical profile- is even more so. Consistency is a precious commodity precisely because it’s an incredibly rare one.

Take Kovalev (no, not literally, he’s not for sale until January). Last year was, by virtually all standards, an improbably successful year for him. That success drove the success of his line, which in turn was part of what drove the success of the team. Consider: 2006-2007 was one of only three years in his career where he averaged better than a point per game, and also one of only two years where he played all 82 games. On average, throughout his NHL career, Kovalev averages .8 points per game, and plays 72 games a season.

So first, think about how happy one might be with Kovalev as a .8 points-per-game, 72 game player. 58 points in a season. That’s good. That’s useful. But it’s not necessarily spectacular. Armchair GMs could reasonably differ as to whether that’s worth $4.5 million per season, and the ultimate decision would probably come down to the weighting of the intangibles- whether one places more importance on qualities like his size, strength, and ability to make the tough shots, or on his reputation for being ‘enigmatic’ and difficult to coach.

But the fact is that Alex Kovalev has never been a player judged by his averages, because he’s never been a man who played to any average, including his own. His salary and his status are not those of a 58-points-per-season guy, they’re those of a sometime 95-points-per-season guy. Kovalev, like a lot of star players, is a site of eternal optimism for hockey fans and pundits alike. What he actually produces, day in day out, over the long grind of a fifteen-year career, pales to nothing in comparison to the flashes of greatness he’s shown in some particular seasons. His pay and the expectations of him are based on aberrations. Remarkable, glorious, aberrations, the aberrations that dreams are made of, but aberrations still.

Over the past five seasons, discounting the lockout, Kovalev’s point-per-game averages have been: 0.99, 0.58, 0.94, 0.60, and 1.02. To average those numbers yet again would be to deny the truth of them. It’s a yo-yo career, and (even leaving aside the dismal thought that this pattern portends a poor season in 2008-2009) it makes one wonder why anyone- any fan, any commentator, any coach, any GM, any teammate- feels comfortable predicting or expecting any particular result from the man. Buy Kovalev, and you’ve bought a huge maybe, you’re buying a certain percentage chance of greatness and assuming the risk of massive disappointment.

Remember, Habistan, in 2006-2007 Kovalev was considered the bane of the Canadiens. The media minds were competing to see who could demand his expulsion or demotion most loudly and creatively. That was a 0.6 PPG season. In 2007-2008, the competition- among the same minds- was who could demand his re-signing with the most effusive hyperbole. That was a 1.02 PPG season. Whichever way the present season goes, neither position was correct, for both were based on the anticipation that future results would replicate the most recent history, and if there’s anything we should know by now about Kovalev, it’s that history is not destiny.

[As a sideline, I’d like to pose the following question to those of my readers interested in hypothetical hockey management: given the statistical profile I’ve outlined- both the average and the inconsistency- under what circumstances does one consider Kovalev a good signing? What kind of contract is this kind of a player worth? Who would be willing to take the risk, and how much money would they pay for the privilege?]


Simonus said...

You are quite right to note Kovalev's yo-yo pattern, but there will come a day when the yo-yo stops and all that is left is decline. Hopefully that will not be for a few more years, but it will be relatively soon. As such there is a high degree of risk in re-signing Kovalev, especially if he is looking for a 3+ year deal.

It seems to me that Kovalev's future value to the team will depend highly on the development of Andrei Kostitsyn. If he can be a consistent game changing winger (a tall order), then the need to take a risk on an aging Kovalev is obviated.

Additionally, if the Gaborik rumours bear out (of which I am completely unconvinced), then I really don't see a place for Kovalev on the team.

This all assumes Kovalev as the franchise player... If, as his abilities break down, he could transition into a secondary scorer/leader with a commensurately lower salary (~$4/yr?) the whole calculus changes. I think he has already transitioned into a leadership role, but can he be the no.2 winger/powerplay specialist?

Success makes one generous and that could explain why Kovalev seemed so much more helpful and affable since the last season, but I hope and think that he might have come around to seeing himself as more of a leader than a loner. I'd be interested (but in no way anxious) to see how he'd react this season to being outproduced by other wingers (Tanguay? Kostitsyn?) on his own team.

Habsfan1993 said...

OK, so I'll take the bait. Kovalev will have a tremendous season. He has been known to produce marvelously when he is in a contract year. Does that mean the Habs will or should re-sign him? That remains to be seen.

As a make-believe GM, I would consider not only his average scoring, but WHEN that scoring happens: big-game situations, playoffs, end-of-season. Also, a GM likes players who fill seats and sell jerseys. This, Kovalev does like few others. One might even argue that the exposure created by Kovalev's less-than-stellar moments creates buzz that benefits the bottom line (As in, no publicity is bad publicity).

I would say that a five million dollar contract would not be excessive, but if Kovalev does shoot the lights out this year, he could fetch Briere-like money. After all, despite your protestations, past performance makes players more attractive to GM's. Kovalev is a marketable asset, and that counts for a lot. And if a GM is really savvy, he times the UFA year of his star to coincide with when he anticipates his team to be seriously competing for the Cup (ahem, Bob Gainey, ahem).

Jeff J said...

"...under what circumstances does one consider Kovalev a good signing?"

As far as the inconsistency goes, Stephane Richer had that yo-yo thing going for awhile too. It's all just random.

There is no real world circumstance that will make signing Kovalev in 2009 a smart move. There will be GMs willing to pay him based on historical counting numbers rather than what we can reasonably expect him to contribute in 2009.

First commenter:

"This all assumes Kovalev as the franchise player..."

He wasn't last year and won't be this year. If anyone is on this club, it's Markov.

"If, as his abilities break down, he could transition into a secondary scorer/leader with a commensurately lower salary ... can he be the no.2 winger/powerplay specialist?"

Before Tanguay came to town, A. Kost was probably the team's best EV forward. Kovalev was already a #2 calibre wing playing on the 1st line and a PP specialist last year.

Faceoffs this year: Carbo starts Kovy shifts in the offensive end of the rink more than any other player (26 def zone, 31 neutral, 34 off. zone). On the other end of the spectrum, Koivu/Tanguay are given the tough assignments (32 def, 27 neut, 18 off).

On top of that, Kovalev has lost more territory than any other forward as evidenced by the number of shifts ending in each zone (27 def, 24 neut, 27 off). He's given up a net 8 rink lengths. Only Brisebois is worse, at -14. Koivu/Tanguay are even-steven, and doing so against roughly equal competition. Not only are they outplaying the Kovy line, but they're doing so with the handicap of more often starting shifts in their own zone after the Kovy line has lost territory.

The upside of Kovalev's game is the PP. To start with, his PP performance was doused with magical pixie dust last season. He was the runaway league leader with 8.1 PPpts/60. That won't continue. He will regress to the mean, which for Kovalev will still be stellar, just not 8+ pts/60.

For MTL in 07/08 the PP/PK/EV icetime breakdown was 12%/11%/77%. By goals, it was 25%/18%/56%. To realistically pin a value on Kovalev, you would have to figure out how many goals for/against he contributes compared to replacement-level player in each situation and weight it by the 25%/18%/56% breakdown. He's average at EV, does not contribute on the PK, and is very good on the PP.

For the sake of simplicity assume forwards get paid by points alone. This should flatter Kovalev because he gives up more chances against than most wingers in his pay range. Figure replacement level for MTL is ~4.0 PPpts/60 (Ryder) and 1.2 EVpts/60 (Begin).

Assume Kovalev gets the same minutes: 4 PP, 14 EV. Assume he scores on the PP at 6.0 PPpts/60, which is still enough to lead the league last year and 1.8 EV pts/60. I doubt he'll do that well this year but I'll give him the benefit of the doubt. Plug in the numbers and he gets 22 points over replacement value. If you grant him a more realistic 5.75 PP pts/60 and 1.5 EV pts/60, he's 15 points over replacement.

If you figure Tanguay can put up 5.0 PP pts/60 and 2.5 EV pts/60, given the same minutes he's worth 30 points over replacement value. At 8.1 PP pts/60 and 1.9 EV pts/60 last year, Kovy was 35 points over replacement value. Plekanec was 30, AKost was 21 (because of poor PP numbers). And remember this is based entirely on points (favouring Kovy) instead of goals contributed for/against.

Sorry for beating a dead horse on this subject, but it's clear that Kovalev's PP contributons continue to obscure the fact that at even strength he scored at a rate identical to that of Chad Kilger despite having much better linemates. Plekanec and A. Kost were top 40 scorers at EV. Tanguay was not far behind them in CGY with less skilled linemates.

Mistaking historical counting numbers for expected future contributions and mistaking PP contributions for overall contributions is what got Ryder $4M per year. The only way a Kovy re-signing makes sense to me is if the dollars are equivalent to another FA signing elsewhere who contributes 15-20 points over replacement value. That's just not going to happen.

E said...

three responses that cover most of the range of possible (reasonable) replies. i love it.

simonus- kovalev's age is, obviously, the elephant in the room that i didn't directly address. given that, it's unlikely that resigning him is a reasonable unless he's willing to go for extremely short term contracts. and it's true that a declining player can still be useful, particularly if he fades gracefully and takes on an 'elder statesman' role (and a lower, although possibly still inflated paycheck). but i don't know if kovalev- even the reinvented kovalev- is that guy. (he might be, although previous history doesn't suggest it.)

1993- leaving aside the question of whether kovalev really does better in 'big games' (i'm skeptical- i think you can find plenty of examples either way, but collective memory is kind and favors the dazzling moments), given the habs current crop of popular young talent, i don't think gainey is in need of kovalev to spur fan interest. and while i'm sure somebody in the canadiens organization considers these things, do you really think they matter urgently to silent bob, compared to the on-ice issues?

jeff- i can't possibly respond to all your points without writing up an entirely new post, but i wanted to mention a couple of things:

1) i'm not persuaded by the boa post about consistency. it's an interesting speculation, but i think he overgeneralizes wildly from his initial evidence. yes, sometimes people use the term 'consistency' to describe getting better, but unless you can show me that no player is more steady in day-to-day and/or season-to-season production than any other, i'll continue to believe it has value as a descriptive concept, if not a prescriptive one. i don't know that kovalev is inconsistent by choice or that he could do anything to become more even, but that doesn't change the fact that he is inconsistent. moreover, as one of the comments mentions, 'random' is also an overused concept in hockey, often deployed to describe things that fit an approximately random-looking pattern for which we cannot or will not pursue the causes.

2) i get what you're saying about even-strength, but a powerplay is not an inconsiderable part of a team's success. yes, if it's a choice between having a good even strength team and one that has to rely on the pp for it's goals, than obviously it's better to have the former, but if kovalev can be the difference between an elite pp and an average one, that's still worth a lot. on the other hand, the habs have in the past couple of years routinely given up the guy who was thought to be key to power play success, without suffering a dramatic collapse...

Topham said...

Interesting piece. And welcome back.

Something that I learned well from watching him intently over two seasons is that Kovalev is not as wildly inconsistent as his stats would have us believe.

This is key. In 2006-07, he was nowhere near as bad as his 47 points had the media thinking. He was personally responsible for at leat 4-5 wins that year and still played with flair and control throughout. 2007-08 was a truer reflection of his contribution in statistical form. The bump coming as he became the focal point of the PP.

If you sign players based on statistical analysis alone, you are a poor manager on a par with anyone who ran the Rangers, Flyers or Leafs in the 1990s. For that reason, the prospect of signing Kovalev has little to do with how his totals look at the end of the year, and more to do with how he continues in his role as leader and mentor. If he comes anywhere close to how he performed in this capacity last season, then he should be signed, without hesitation.

As a side note, I think your opener about there being no first line in Montreal is fine as far as opinions go. But it is a naive take given that Koivu was playing 6 minutes a game less than Plekanec over the first two weeks. Carbonneau clearly has a first choice, and it happens to reflect what everyone else thinks. Even though Koivu and Tanguay were hot, they were not the first line during that period. The situation has evolved to the point that it is now debatable based on ice time, but I beieve that has something to do with Kostitsyn's injury...

E said...

topham- my point was a little broader than the last two seasons- his inconsistency in terms of results goes way back in his career, and it may be unrealistic to consider the three seasons where he got more than one point per game to be more representative of the player than the several where he's gotten less than .7 per game.

as to him being worth re-signing for his contributions as a leader and a mentor, i agree that it might be a good idea, but that's part of why i asked under what circumstances he'd be worth retaining. fair or not, there are different pay scales in the league for leaders who are expected to put up big numbers and leaders who are expected to put up moderate numbers while being a good influence 'in the room'. so i wish you'd given me more information on what kind of contract you'd be signing him to- how many years for how much money?

finally, i wasn't saying that there's no first line (assuming you call the first line the one that gets the most ice time) at any given moment, but merely that that line will not be consistent. if recent history teaches us anything, it's that carbonneau scrambles his lines readily in pursuit of better results. just because plekanec-kovalev-kostitsyn came out on top by the end of last year doesn't mean they will this year, or even that they'll still be together in a few months. all i'm pointing out is that as long as the team is doing well and somebody is getting goals, there's no point worrying about 'the first line'. if they start heating up again, they'll get their ice time and live up to all the expectations. if they don't, then other things will be tried.

Topham said...


Well since we weren't all Pittsburgh and New York fans in the day (I take some liberty in making this assertion, it's possible some of us were), I don't think it is valid to compare his time in Montreal with his time in those other two places.

Consider he had one good season in Montreal, one bad one and one great one - a veritable microscosm of his career. In doing so, I wouldn't say that his play on a personal level changed dramatically.

It is easy to be convinced that it did in retrospect with statistics and stories about him being a floater. But I have personally witnessed now how reporters labelled him a floater back in 2006 and I can tell you it only happened following games with no points and was often totally irrespective of how he played on all those other shifts where goals aren't scored.

I suppose that means I do everything in my power to sign him, keeping in mind of course that other important pieces must be secured first. Namely Plekanec (and the money to pay Andrei Kostitsyn, Andrei Markov and Carey Price in their future contracts). At Gainey's current standards, I see an offer for $5 million a season for Alex.

And, despite your original assertion and reply, I still disagree about the first line. For the first 6 games of the season there most certainly was a favourite offensive line. With good reason it was the line with the two most gifted goalscorers and the centre they like to play with best. Carbonneau played them and played them.

E said...


we are going to have to agree to disagree here. it disturbs me that you're unwilling to do me the credit of acknowledging that i, too, watched games in 2006-2007. quite closely in fact, and moreover i wrote a good deal about kovalev's performance during that season. i don't need you to tell me how he played, but i do disagree with you. there were plenty of games where i thought he was as invisible on the ice as he was on the scoresheet- hanging back at the fringes of the zone, hesitant playmaking, lack of cooperation (i.e. sharing the puck) with his linemates (especially plekanec). you can find a few interviews where he himself talks about changing the way he played in the 2007 off-season. personally, if i was going to give an explanation for the difference, i'd say that it seems like his attitude exacerbates both hot and cold streaks, that he takes a lot of pride in his performance but reacts in self-and-team-punishing ways to adversity. perhaps that was only one year and he won't ever do that again, or perhaps it really is luck both ways and attitude is irrelevant, but either way i wouldn't bet on him producing at last year's rate consistently for another contract equivalent to the one he has now. i'd bet cash money and quite possibly the souls of my children on that, if i had children.

but it seems clear that nothing is going to resolve this disagreement short of rewatching the entire past two years game-by-game and debating them in detail. otherwise, all we have is your assertion that you watched and saw certain things, and mine that i watched and saw different things. we'll both trust our own eyes and disregard the other person's assertions about the evidence of theirs. i'm not a huge stats person, this post notwithstanding, but that is the virtue of them- when discussing numbers, we can share a common frame of reference. with personal perception, we clearly can't.

Topham said...

Quite right, E. I was wrong not to acknowledge your experiences in 2006-07 here.

I can only plead getting carried away. I think at a certain point I stopped responding to your specific points and started railing against people (other than yourself) who would be happy to let stats do the remembering for them.

And to be honest, I think this article just hit a nerve. I basically started blogging out of frustration re the Kovalev file. My first entry was on Kovalev. We watched him with hawk eyes all last season to document some of the things he did other than score points.

So, I apologise for my disrespect here. I hope we can agree to disagree and still get on.

I still sign Kovalev, because I still believe what I said when wrote this:

"What Alexei Kovalev gives the Canadiens is a unique proposition. He gives them alternate routes to scoring when traditional avenues are blocked. And, to my mind, he's the only one who does on this team. And that's why he is valuable."