The Numbers Game: Merlot's Gone Sour

What can you expect out of a fourth line? You want them to be capable enough players to eat a few minutes at even strength and give the scorers a rest. You'd like some specialists to call upon in special teams situations and maybe somebody to police the ice once in a while. You want them to play with energy, perhaps a banging game, and to hem the defense in until reinforcements can arrive if they lack the finish to do the job themselves. And maybe get more shots than the other guys in the process, why not? Sounds like we've pretty much got just that, doesn't it? Unfortunately, this appears to be a fallacy.

A Quality Vintage?

We'll come to this in a minute, but lets take a timeout and huddle around the bench for some explanation. We've been looking at Corsi and PDO pretty heavily, and I always yammer about the influence of zone starts, but we haven't really addressed much to do with Quality of Competition (QoC), or more specifically, Corsi QoC. Corsi QoC aims to give a snapshot of a player's opponents and how difficult they are to defend and get offense against by looking at their individual possession performances. Unlike QoC, which is measured in terms of regular ol' +/-, Corsi QoC is measured by taking the Corsi ratings of a player's opponents - basically how good each is at hanging onto the puck and generating offense - then multiplying these by the fraction of ice time they spend together, thus weighting the stat by the frequency that Corsi number is faced. If Kessel has Chara glued to him, his QoC is going to be heavily dragged into positive by the big man's big number. Next, average the results. Add a twist of lemon and a dash of bitters. Serve over ice. If this sounds like a hassle to calculate, you'd be right. But fortunately others are doing it for you pretty much on the fly with sophistimacated machines and you can just stroll in and check out the results. Still, it's helpful to have some grasp of just what's behind the number.

Since this stat is comprised entirely of Corsi, it shares some of that measurement's deficiencies. One can't really infer a great deal about the defensive capabilities of the opponent, save for their possession ability. It won't really reflect shut-down guys who get the puck out of trouble then change for the scorers, ala Kelly. What you will know is that the opponents can generate shots more often than they're facing, and while this often correlates with the best offensive players, it's not the case across the board. Since just possession and not scoring metrics are taken into account, guys like Justin Williams who aren't exactly lighting it up will sneak to the top of the QualComp list of tough competition. Also, because of the discrepancies between team Corsi across the league, there will be some extreme poles that make it difficult to nigh-impossible to use Corsi QoC as a metric for player-to-player comparison across teams. Bruins are going to have higher Corsi On's by nature of their game than, say, a player in Buffalo, whose rating suffers from the fact that they play much better possession teams over and over in this unbalanced schedule. That doesn't mean a negative player on their team is outright bad. Under a different team and thus against different teams, he may improve drastically. So try not to think in absolutes, as with PDO last week. Even so, noting leaguewide outliers can be informative. I'm going to skip over another variant of this statistic which aims to address this weakness, CorsiRel QoC, since it doesn't much play into our intra-team discussion.

Harvesting the Grapes

Now that we know one of the component stats involved ahead I think I'll spare you yet another list of numbers, so instead lets turn our peepers to some purty pictures. Below lets scan one of Greg Sinclair's (@theninjagreg) awesome player usage charts and see how our forwards are faring:

(click for larger image or hit the link above)

To quickly address how to read the chart, Corsi QoC is plotted vertically on the left, so players facing toughs on the other team are up high, guys facing crappy stiffs who are getting shelled all the time down low. Offensive zone starts are plotted horizontally on the bottom; dudes on the right aren't seeing much time in defensive situations. The dot size represents how positive or negative their Corsi Rel is, with blue representing positive and red negative. Big and Blue is good, Big and Red is bad. Tim Thomas disagrees with these value designations.

What we can initially read into this is that Claude is pretty confident in his top three lines as a whole. They are all pretty much seeing parity in their deployment as far as the Corsi QoC is concerned. They're being tossed out there against whoever's decent-to-good on the other team, with certain guys entrusted when trapped on the wrong end of the ice - mostly Bergeron and Kelly. The fourth line, on the other hand, is only being rolled out when the competition is bad at puck possession and tends to get outshot heavily. Paille's a touch higher, but only because of recent rides on the third line.

This must be common right? Most teams must roll their least skilled guys out against comparable opponents. Not really, I'm afraid. Take a look at Anaheim. They're giving easy minutes to Teemu Selanne and Bobby Ryan in hopes of them feasting on a goal buffet. Montreal gives the same to Max Pacioretty - no wonder his Corsi is so good. Inexplicably, Nashville gave up a 1st rounder to have Gaustad take draws against garbage competition. Vancouver famously gave Malhotra ALL the defensive starts so the Sedins could have ALL the offensive starts. To lesser extents, many teams prefer to use the bottom line as a checking line to let the skilled guys focus on filling the net, and so giving the scrubs tougher competition. Our coach don't roll like that. He likes to roll three lines - not four as the oft-repeated narrative goes - and reserve his fourth against the opposition fourth, letting everyone above Merlot handle equal measures of scoring and scoring prevention. Few other teams use this model. The Leafs are one, and they're having better results with that bottom unit this year.

Now that we've established Merlot aren't totally the norm and are getting somewhat uncommonly easy deployment, allow me to scrutinize. Note their massive red circles. As clear as day, these guys aren't doing well against bad puck-possession players and are getting outshot heavily themselves. Since the dot color and size is based on Corsi Rel, measuring them in comparison to much better teammates, I should note their Corsi On, the raw +/- of shots when on the ice. Paille is a not-great -4.21 and Campbell's is a dreadful -15.27. Conspicuous in his effort to hide from his poor performance off the edge of the chart, Shawn Thronton isn't just getting murdered by John Scott, he's also getting killed in Corsi, breaching -20. Given the softest competition and the cushiest starts, he's still in the bottom 7 league wide in Corsi. Ho-Lee-Crap, there's that outlier I was talking about.

The Dregs

Shawn Thornton, as we've been saying over the past few years, is an uncommon goon. Sure, he can punchisize a face, but he's not lost with the puck and can capably keep the rubber in the zone and chip in some timely secondary scoring. While this may have been true a few years ago, it would appear we've bought into the myth of this pugilistic folk hero. This town loves a fighter, and it is clearly clouding its judgement of a guy in sharp decline. To dramatize said descent, I've taken the liberty of gif-ing his full-time seasons with the Bruins:

You see him start with a wee blue circle, indicating a positive Relative Corsi. In other words, good in comparison to some of the team. Then you see it turn red-but-small for a couple years, then blow up like a neutron bomb over the course of the past three. For a frame of reference, the first larger red dot is the cup year, the beginning of a personal decline.

Obviously, since I zeroed in on the dot you can't see the grid showing him receiving fewer and fewer defensive zone starts, but what you can see by the last two frames is that his dot is bleeding over the bottom line of QoC, showing that he faced worse players as time went on, and he performed worse even so. He's now facing the worst players available to him, coming in at 6th worst competition in the entire league. And he's getting shelled more than he's ever been shelled before.

To play devil's advocate, I'll provide an alternate reading: the team stunk when he got here, hence the positive Corsi Rel, and got progressively better around him, thus exploding his dot in comparison to his teammates. This could certainly be a valid interpretation based on the information available in the chart. Unfortunately, there's more out there. If we take into account his Corsi On, he hasn't been positive since 09/10. Subsequent years have seen him drop gradually from -4.29, -6.92 and this year's -22.97. There's still plenty of time for the last number to improve, but it's indicative of a continued trend.

Clearly sagging in possession terms, he's also entering an apparent drop off in production from his career high shot count in 10/11 and concurrent height in shooting percentage and scoring. Right now he's actually for the time being beating his top percentage, but that's easy to do with one goal in ten shots over 14 games. At his height, even as his possession started to tail off, he was at least directing around two shots on net per game. Should he be unable to up the firing rate above one, his percentage will dip and he's going to have little left to recommend him beyond his fists.

Finishing the Bottle

Thornton may also be having a strong negative impact on his linemates. Taking a look at his

WOWY chart, he's clearly projecting the Bourque Effect on Paille, who soars when relieved of the fourth line and paired with Peverley and Kelly. It bears noting that we appear to have a prime candidate for a move to the third line, as Paille conversely has a positive impact on teammates on both lines. Campbell, while also pulled down by Thornton, doesn't seem to have as positive impact on his pugilistic partner. Their whole is less than the sum of their parts thus far this year, combining for a drop in Corsi for both and for Campbell a drop in production when together.

In previous years, there had been no marked advantage for any of these players between playing together or split up with other groups. Campbell and Thornton did have a similar muting effect on each other last year in scoring terms, but no major impact on each other's possession. Paille's stayed fairly flat during their time together, so this team-elevating quality of his game is a new development. Don't question it, just enjoy.

While struggling mightily at even strength, Campbell still has considerable value as a PK specialist. He's Claude's favorite in penalty killing situations, earning the most PK TOI, and from the look of his offensive zone finishes at 4v5 he's pretty damn good at clearing the puck. You really shouldn't bother with Corsi when it's not at evens, unless you like negative numbers thanks to the inherent imbalance of the situation. So this is about all we have to go on by individual stat for the PK. Paille is serving similar duty, though isn't as good at clearing. Hard to gauge his impact at slowing down the play through the neutral zone though, which the eyes would say he's rather good at.

Distilling It Down (into grappa?)

The overall performance of the Merlot line is in sharp decline, and it appears to be largely due to a deterioration of Shawn Thornton's game, though abetted by iffy showings from Campbell. Right now, this much lauded unit is playing worse than the Begin and Yelle fourth lines, who maintained possession better against tougher competition and scored comparably. Were it not for the four rings among them, the current group would be held in similar regard.

Paille has developed serious value to the team. Even though my observational brain can't quite wrap itself around the concept after witnessing a few years of failed breakaways and meager production, he's proving himself an objectively decent third liner. Given the competition, Julien should seize what he has in Danny and kick Bourque to the curb, at least until trades are made. It will serve to kickstart a dormant third line even if it diminishes the fourth.

Campbell's value, lest he can rebound with new linemates, is becoming that of PK specialist alone. Which isn't inconsequential, but it does beg the question of why he's not being deployed in more defensively demanding situations during 5 on 5 play. Is he being saved up for the inevitable lapse in discipline by his teammates, knowing our propensity for over-the-edge play? Or do we just have a lot of defensive options with greater offensive tools? I'll go for B.

Thornton, on the other hand... are his fists still so valuable we can risk his negative influence on his own line and his gradually declining skill set? I know he's here for another year, but we're going to have to get used to life without him already since he's a shadow of his former self. I'm just gonna leave this here for you:

Tasting Notes

- Remember how Bergeron was being used in an exploitation capacity? Yeah, that's over. Back to Selke duty for Patrice. He's now number two for defensive zone deployment.

- The depiction of the Bergeron line seeing low quality of competition earlier this year could easily have been a misinterpretation based on limited data. Provided just a couple couple opposing teams got outshot for a few games early in the season, it could be enough to make good competition look bad in the eyes of Corsi with such a small sample. Short of looking at shift charts and examining each individual opponent's Corsi at that point, it's hard to say, but that would my hunch.

- The defense offers zero surprises. The QoC is tiered by pairing, Z and Seidenberg are making up for Hamilton's light defensive zone usage, and Ference and McQuaid are rarely to be trusted. An upgrade on the bottom two would be welcome.

- With last night's action, we moved into number 2 on the Fenwick Close charts. We're almost number one in something other than the PK! However, the Kings look dominant by this measure, so we probably won't be overtaking them.