Why I’m not worried about Jake Gardiner and Nikita Zaitsev

I stumbled into an argument about who was the Leafs number one defender the other day. It was fairly amusing as these things go. A fan of some other team was trying to construct the killer put down: "Well with [insert name here] as their 1D, no wonder the Leafs stink."  They failed on the delivery, however, because they didn't know which name to toss in there. The joke died on the vine as their expert advisor kept changing his mind about which player was the villain of the set piece.

More amusing than merely watching the timing on a cliché joke fall apart was watching the expert never once consider the obvious as he frantically compared five-on-five and all-situations ice time to come up with one name after another: The Leafs don't have a number one defender.

The Leafs have four defenders who play virtually indistinguishable minutes and competition. Each pair of them has one heavy penalty kill minute player, and one power play specialist. Their five-on-five minutes are all close together with less than two minutes per game separating the lowest from the highest.

They are a collective set of 2.5Ds, or perhaps a pair of 2Ds and a pair of 3Ds.  There is a difference in their usage by score state, Jake Gardiner and Nikita Zaitsev (GaZa) play more when the score is close. That’s usually a measure of who the coach trusts defensively.

What is a 1D anyway?

By one definition, the 1D on a team is whoever is best, so someone takes the crown on the Leafs. By another definition, the same sort of thing we mean when we say 1C and mean a guy who leads the team, there aren't necessarily 31 of them in the NHL, and some teams have two.  Or maybe three.  The NHL is unfair like that.

I took the top 10 teams (at time of writing) in the NHL by the standings and looked up their top four defenders by all-situations TOI per game.  Then I added up their cap hits. I picked 10 because I'm way too lazy to do them all, and this early in the season it gives a nice cross-section of teams from the mediocre to the very good.

Top ten NHL top four defenders on November 12

So as you can see, Toronto's defence by committee, which isn't totally unique, is usually less costly than a two stars and two guys arrangement.  I guess we should ask which is better, and the answer is, as usual, that it depends on the team and the stars.

I think Tampa and Pittsburgh are overpaying for their top four, and Los Angeles knows they have one more year of that bargain price for theirs.  Some of those totals are depressed because they've got a pricey guy injured or the team have given up on a mistake of a contract, as well.

But now that I've belaboured that point, which is that GaZa are in some ways pretending to be the top pair and are at least not costing top-pair money, why doesn't that worry me?

I just don't worry about things like that is partly the answer. You can analyze all you like in advance, but the reason we watch hockey in the first place is because you don't know how something will work until you try it.  The Leafs have been trying this arrangement for 300 minutes of five-on-five.  All we know is how it’s gone so far.

Committee defending

GaZa and their two friends Morgan Rielly and Ron Hainsey (RiHa) form the committee defence on the Leafs. The third pair is used in the gentlest manner possible, so they are getting mostly ignored today.

The Score and Venue Adjusted Corsi For Percentage for the top four is: 48, 47, 53,  and 52.  That's in the order of GaZa and RiHa.

And just a note about Zaitsev here. He has more ice time than Gardiner, and it amounts to some shifts with Andreas Borgman for the most part, while Gardiner has occasionally played with Connor Carrick.

So, the difference between the "worst" defender, Zaitsev and the "best" Rielly is 5 percentage points, and in gross numbers of Corsi For - Corsi Against, the difference is 73 shots one way or the other.  And per game? that's less than four shots.

Obviously it's better to have a higher CF%, but seriously, sometimes, we get so close to the decimal places and percentages that we lose touch with how meaningful those differences are and are not.  The joke up there about Darnell Nurse isn’t just because someone on TV wants to trade Mitch Marner for him, it’s because he has a CF% of 58 right now, second highest in the league, so obviously he’s better than everyone on the Leafs.

If you won’t accept that argument, and you shouldn’t, than you need to work hard to convince me that GaZa are bad and RiHa are good based on their small spread in CF%. I'm extremely uncomfortable making broad sweeping judgments about players based on their results in a few games, and sometimes even in a lot of games. Three years of data comes closer to telling the tale on a player, so 300 minutes is descriptive of what has been in a given set of circumstances, not who players are.

I'm a lot less comfortable with using just, or even mostly, CF% with defenders.  Driving play, that is, getting the puck out of the defensive zone, through the neutral zone and keeping it in the offensive zone is important, but it’s a job that's done by every player on the ice in concert.

Defenders defend more in the defensive zone, and that's both an awkward sentence, and so obvious it shouldn't need to be said, but I weight defensive zone execution pretty highly when I look at defencemen.  The best defensive play is the one you never have to make, but every team spends nearly half their time without the puck, so that pithy truism won't help you when you are standing there in front of an annoyed Frederik Andersen, deciding what to do next.

To move beyond just CF% for analysis, the most obvious step is to look at the two component, unrelated parts.  There is no evidence skill at shots for correlates with skill in reducing shots against.  So if we look at our four defenders in Corsi Against per 60 minues, we get: 58, 58, 58 and 62.  The odd man out is Zaitsev.

That number 58 is exactly the Leafs team CA/60 this year and last year. While the Leafs rank better in that measure against other teams now than they did last year, they themselves haven't improved.

Raw pace of shots against is only part of measuring defence.  Before there were expected goals formulas there was this idea (not mine) where you looked at the proportion of the shots against that are high-danger shots.  The idea was that you could tell how tough or easy the goalie's job was with that defender in front of him. Now we can just look at Expected Goals Against which weights all shots and presents a clearer picture of the shot quality allowed.

For xGA/60, our guys turn in 3, 3, 2.6 and 2.5.  Zaitsev is best here, with Rielly and Hainsey being equal at the bad end.  And that difference? It could be nothing but random variance in a small set of shots against, or it could back up Babcock’s faith in GaZa defensively.  That faith is relative. None of the these four players show up on the good end of a league-wide list of defencemen by this measure.  They start around 30th place. That’s 30th worst for big minute defenders.

While it is true that CF%, particularly the adjusted variety, is more predictive of future success than expected goals, I'm not predicting, I'm describing, and as far as I'm concerned, defensively, these four guys are as indistinguishable from each other as four grey cats in the dark, despite what a lot of eye-tests are telling people.

But you see where I'm going, don't you?

Offence is for defencemen too

If our four defenders all have very similar rates of shots against, but different, CF%, then we know where we should be looking for the distinction between them: the CF/60.  Right now, it is 53, 55, 62 and 64.  The order there is Gardiner, Zaitsev, Hainsey and Rielly.  The spread between best and worst is much larger here, and in real numbers is 30 shots more for Rielly than Gardiner, while Gardiner has been on-ice for more minutes.  That's not even two shots per game, however.  This isn't the difference between winning and losing the game every night, but to see Gardiner so low on offensive pace is very odd.

We know who else has been sluggish offensively, and you might expect the explanation to be that GaZa is on the ice with the Bozak line or the fourth line a lot.  If you’ve read other analysis of GaZa, you already know that's not it.  The actual answer is mostly their failure to execute well with the Kadri line.

Here is a somewhat complicated chart that show this:

What you should notice is that down in the bottom left, 51 with 43, 47 and 12 is really bad.  This combination is accounting not just for the terrible Corsi For Gardiner has, but also most of his not stellar Corsi Against.  With the Matthews line, the results are okay, not great, and the Bozak line's troubles  have had a small but still real impact on Gardiner's offensive average. (The obscured combination in the bad area is 22, 51 and 28.)

The interpretation here could simply be that playing checking line support with Kadri just isn't GaZa's thing, and all of this could be rebalanced with some tilting of the deployment for that line toward RiHa.

A quick look at the same chart from Kadri's point of view shows that it's Gardiner and the Kadri line that's truly bad, Zaitsev has done at least tolerably with them.  And Kadri's deployment is already tilted towards the RiHa pair.  GaZa have 71 minutes each with Kadri, the other pair is over 130.  That difference between Gardiner’s and Zaitsev’s results with Kadri is exactly the kind of exaggerated effects from one or two shifts apart that small samples of ice time produce.

I could claim here that the problem with Zaitsev is obviously Gardiner, but that isn't supported by these small slices of game results.

Jake is dull now?

There are three games where Jake Gardiner's offensive pace was virtually nothing.

One: The St. Louis game, where the Leafs got rolled and GaZa played mostly with Kadri and against Pietrangelo and Gunnarsson.

Two: The Detroit game, where the Leafs got rolled in Corsi, won anyway, but GaZa played mostly with an uncharacteristically bad Matthews line and against Babcock's other son, Justin Abdelkader, who just owned them.

Three: The Vegas game, where the Leafs smacked up against a defensive system that has chewed up more teams than seems plausible, even if you've watched it happen.  GaZa played mostly with the Matthews line who stank in terms of Corsi.  They stank because they played against Vegas's one good line fronted by Jonathan Marchessault and Matthews likely shouldn’t have even been in the lineup.

There are a few other examples, moments where GaZa are hemmed in with the fourth line or nights where either the Matthews line or the Kadri line can't handle the competition.  But through it all runs the strain of poor results with Kadri in particular.

That does suggest they are being overplayed a little, and I emphasize the word they here. There’s also no evidence to suggest that GaZa’s offensive dullness is Zaitsev’s fault. But, so what? That's what the committee approach is designed to do, play everyone a little over their heads some of the time.  The goal here is to maximize their individual strengths and mitigate individual weaknesses in combination with the forwards and get a good aggregate result from all four.

That sounds dull, though. I do realize that. No one in this group is going to top any lists or stop the trash talkers from trying to make jokes. There is no star here.

Fix it!

This arrangement is not permanent. Ron Hainsey is not one of the defenders with term, but you can’t just order up a star 1D from IKEA, even if seems like they’re all manufactured in Sweden these days.  There are ways to improve the picture, however.

Try this: Take away Hainsey and add a much better right-shooting defender in his spot, now skew the usage, make RiXx your top pair, and what happens to GaZa's results?

Or this: Find a centre who is a real matchup guy, not a good fake like Kadri.  Rearrange your lines, sort out your defenders, and what happens to the results?

Or this: Play a few more games and see if everyone settles into their role a little better and the individual results start to converge towards team average as they often do.

I won’t be upset if all four grey cats get better in their own end, but don’t hold your breath waiting for really big 50+ CF% numbers. Parity in the league affects individual results too, and hovering around 50% is all Victor Hedman can even manage so far playing a role similar to the Leafs top four.

At the end of the day, the biggest failure of GaZa is in offensive pace, and there's a list of players on the Leafs contributing to that, a long enough list that it starts to look like a problem when anyone but the Matthews line is on the ice, although we’ve already seen some encouraging signs from Bozak’s line.  The Leafs, by the way, are second in the league in Expected Goals For, which makes complaining about anything on the offensive side of the ice seem churlish.  Their team pace of play has declined over the season, but that might be normal ups and downs.

So, I'm not worried about GaZa. Like everyone on the team, they can improve, and they've both had periods of shaky play, but their results are hardly a horror show or all of their own making.  When you're building a team up from scratch, and it was a star 1C that descended from heaven for you, not a star 1D, you will end up playing some defenders over their heads on some nights.

No one looks like they're drowning.

Acknowledgements: All numbers are from Natural Stat Trick except Expected Goals, where I used Corsica’s version of that stat.