Tyson Barrie: reconciling the eye-test, the boxcars and statistical models

Tyson Barrie is a hard man to pin down. I tried three years ago and wasn’t absolutely sure then who he really was, once you dug him out of the pile of snow he’d been buried under on the Avalanche over his career.

Related

Tyson Barrie has some of the worst Corsi numbers in the NHL

It’s easy to say that the problem on the Avalanche then was just bad coaching and to shrug all that off. But the coaching change to Jared Bednar has proven that false, as they had one of their worst years ever under his reign.

The Avs in Brief

Colorado Avalanche

The Avs are the team that got a little better just in time to experience parity in all its glory. They are the go-to example of how hard it is to move up in the standings on modest improvements these days. They’re also a really good example of how luck can drive you to the top of the standings, but sometimes you’re in last place because you deserve to be.

Tyson Barrie and his Box Cars

Tyson Barrie

This all makes total sense. He gets better as he grows in experience, and except for that dip in the team’s worst year, 16-17, he’s fine, great even. Let’s just stop now.

But we learned by looking at his team that points and quality of play aren’t always related, and a second look shows one of his best years is 14-15, while the luck-fuelled run to third in the league was not that great a year for him.

Consider that most of Barrie’s points are assists — he is a defenceman after all — so he’s at the mercy of his on-ice shooting percentage. The first model of hockey you need in your head and your heart is “shit happens”. No one gets this game like Auston Matthews gets this game. The place you go first when you see that shit happened, like Barrie’s drop in production last year, is shooting percentage. We should expect that the shooting percentages of Barrie and his teammates were really good in 17-18 and not as good last year. That is true, but it’s almost entirely limited to the power play. Barrie’s own shooting percentage at five-on-five was in the normal defenceman range of 4.24% in 17-18, not the borderline forward level of 6.49% it was last year. On the power play, in 17-18, he shot 14.89%.

One thing we do when we look at results that vary widely is try to tell a story that fits those facts. We seek an explanation. And often we forget “shit happens”. But when we see consistency like Barrie’s points through his career we don’t dig any deeper, and the one anomaly is easy to explain. But in truth, Barrie’s offensive results are total chaos. The appearance of a simple progression is a fluke, and just happened to offset all the other wildness going on in his game, because his game is wild.

Barrie is Not the Model of a Modern Statistical Marvel

Remember that year where the Avs points percentage was .293 and their Corsi was almost as bad?

This isn’t it. This is the next season where the team made the playoffs, the year with Barrie’s incredible and totally unbelievable power play shooting percentage, and nothing else that was very good.

These Evolving Hockey RAPM charts, and the HockeyViz “Threat” calculation on heatmaps we’ll look at later, are statistical models that are adjusted for effects of teammates and other usage like quality of competition. What you see is Barrie, not his team — as much as that is possible. Part of Barrie is his shooting percentage, and normal fluctuation from year to year. Over his career, he shoots about like a league average forward, which is above average for a defender.

Just looking at a defenders impact on shots for and against, the Corsi For measures on these charts tells you useful things, and are much more meaningful that just looking at an individual’s raw Corsi numbers. The Expected Goals bars that sit alongside, start from Corsi weighted for quality of shot, so they’re where I tend to look most.

Look at Barrie’s defensive CF and xG in 17-18, and understand that it’s pretty much impossible to be in the NHL and have a worse looking season defensively than that, even if you’re a forward with a shot like Phil Kessel.

Kessel had never been that bad before, and he got better later — players can have bad seasons, but they generally come alongside a team’s historic failure, not the year after.

Let’s ground ourselves with a look at Barrie’s whole career in all situations, offence and defence:

All GAR (Goals Above Replacement) models have a lot of shooting percentage luck baked into them, but they include everything together about a player. In the two really bad seasons for the team, 14-15 and 16-17, Barrie has his high and low point during the mature part of his career, And even though we know last season was likely his best ever at even-strength, his power play performance watered it all down to barely better than his prior three years.

The 14-15 season is the first hint of Barrie dancing to a different beat than his team. He had 63 points while the team around him was astoundingly bad. I watched every single game the Avs played in that season, so I know Barrie was paired almost exclusively with Nate Guenin, a man who should have been a good pro in the AHL, but there he was on the ice waving his arms and shouting, telling the players where to stand in the defensive zone. Coach Patrick Roy experimented with switching between zone and man-on-man defence, and succeeded in confusing his team made up of really old guys (Alex Tanguay, Jarome Iginla, Danny Briere, Max Talbot, Jan Hejda, and so many more) and a cast of younger forwards who were hopeless at executing whatever the system was Roy was expecting. Matt Duchene couldn’t cope as a centre in that setting. Nate MacKinnon looked like a draft bust. Ryan O’Reilly and Gabriel Landeskog ventured out across the 50% CF line when they played with Barrie.

I have always been confused about Barrie. What’s him and what’s his astonishing mix of teammates? I remember a lot of shifts that season that he played with a fourth line centred by Marc-Andre Cliche, (Frederik Gauthier without the charm or the offensive skill), and with a mix of Briere, Talbot and Cody McLeod on wing. Barrie played as a winger for real on those occasions. He carried the puck over the blue line, charged the net and played successful offensive shifts with Briere and Talbot, two guys on their way out of the NHL, who nonetheless knew how to play hockey.

We know the Avs were hopelessly outshot that year, but hidden in their terrible CF% is the fact that their offensive pace was terrible. They weren’t high event failure, it was just failure to ever have the puck. They were worse than the Leafs that season in all categories.

Barrie, even with his fourth line time, was carving offensive opportunities out of a very hard granite that year. He wasn’t even at league average, but he was above average on his team. Neither Landeskog nor O’Reilly were any better.

The next season was worse in many ways. O’Reilly was gone, MacKinnon was not quite the dominating centre he is now, and Landeskog had a really down year. Barrie looked about as he had the year before. It wasn’t until 17-18 that the top forwards on the Avs, who now included Mikko Rantanen, looked like above average performers, but here’s where it gets confusing. We already know Barrie had an epic defensive disaster in 17-18, but his offensive results didn’t change. He clocks in at around a -10% offensive threat. He didn’t flip over into positive offensive threat until 18-19 when suddenly he’s a +10%, and he played with the same forwards both years.

Prior to last season, if you shut your eyes to the points, he was uniformly not much better than team average in offensive impact by any model. And, therefore, if you look at any multi-year set of results, he looks bad, but that’s the bulk of his career! Compare last year to his prior three, and it’s hard to see the same player.

Barrie is not just out of sync with his team, he’s out of sync with himself. Are the models doing less to remove teammate effects than they should? Did Barrie, at 27, suddenly figure out how to effectively channel his offensive energies while not being the worst player ever at defence as he used to seem? Is this just a coaching system that killed a power play made up of the hottest line in hockey plus Barrie, while the same four guys were killing it at even-strength?

I’ve largely ignored his defensive contributions, because he’s just not all that hot at it, and that’s clear from watching him, from his usage and from just looking at simple Corsi stats. But his 17-18 is the worst performance by any Avs defender in the last four years barring Francois Beauchemin in his last season on the team.

That horrible season for Barrie, the one where the rest of his team was rebounding from a year of hell the season before, came with a couple of interesting events. First, he was injured, then Erik Johnson was, and when Barrie came back, he played out the season as the top pairing defender paired up with Nikita Zadorov. A fellow who I think of as Roman Polak with a better press agent.

Zadorov is only just an average defender who adds nothing but exciting-looking hits to a game. Top pair minutes with him put Barrie on the ice with the best forwards almost entirely in the offensive zone because Bednar isn’t stupid, he knew they needed a cushy job. And that’s the best part of Barrie’s season. Prior to that he played with Patrik Nemeth, who was a bit sub-replacement level. That’s where Barrie’s season seems to have been torpedoed. Last season he spent with Ian Cole, and it is both true and a little amazing that a solid and uninspiring player like Cole is one of the best defence partners Barrie has ever had. He and Barrie clicked, and they were both excellent.

If you compare the career defence partners of Tyson Barrie and Cody Ceci, Barrie’s are likely worse.

But hold on here. These models do a good job of removing teammate effects and while Barrie’s defensive partners have often been terrible, his defensive shortcomings should be seen as his not theirs. You can see some effects in these charts of players changing teams, but don’t go mentally adjusting for a historically bad team more than a little. I could decide Tyson Barrie is a guy who is overly influenced by those around him beyond what the model can sort out. That explains why his Goals For Percentage is nearly always above 50% when his  Expected Goals For Percentage is nearly always well under. But is that just another too-easy answer?

Barrie is a lot of fun to watch, and the temptation to bend the stats to make someone that exciting into someone really good has to be resisted.

The Eye Test

The scouts eye-tested Barrie first. He was drafted in the third round in 2009 after he’d scored 52 points in 68 games in the WHL. Victor Hedman, who was ranked as the number two on Bob McKenzie’s list that year behind a certain Maple Leafs centre, had 21 points in 43 games for MODO.

Scrolling down the list of players and reading the scouting reports from a decade ago is an education. There’s nothing about defensive play for the defenders. The closest you get is a mention of the transition ability of some of those Swedes. But Ryan O’Reilly sounds exactly like the man he is today, down to the “crazy ice time” and faceoff prowess. So it’s not like scouts couldn’t see a forward for who he is, but after the top two defenders, Hedman and Oliver Ekman-Larsson, the hilarious mistake of Jared Cowen  at seventh isn’t even all that remarkable. Everyone a decade ago was all wrong about the men on D.

At the bottom of TSN’s list, outside the top 60 and in with the guys who are grinder forwards today in the NHL mixed in with names I’ve never heard, not even on a European team, I find Barrie as an honourable mention. Here’s his scouting report:

After a successful regular-season in 2008-09, Barrie helped lead his team to a WHL Championship and the Memorial Cup tournament, recording 18 points (4-14-18) and a plus-18 rating in 22 playoff games.

Is the son of Tampa Bay co-owner and President Len Barrie -- who was drafted in the sixth round (124th overall) by the Edmonton Oilers in the 1988 NHL Entry Draft and played 184 NHL games with four teams (Philadelphia, Florida, Pittsburgh and Los Angeles).

Points and his famous dad. You’re left thinking that without the dad, the points wouldn’t have mattered much. It’s fair to say no one knew what he was back then beyond not very tall. Nothing’s changed.

His hottest offensive season was last year, so let’s run our own eye test:

That’s a lot of appearances by a particular trio of forwards. But remember that horrible year in 14-15, when he was good by the models when no one else was? He helped Tanguay and Duchene to a lot of goals, but he really did succeed with the fourth line too. He teamed up with Briere for three of his eight goals that year.

Conclusion

If you look at his box cars, he’s almost boringly consistent. if you look at the models, he’s not even good at offence. If you watch him, he’s neither a defender nor a winger, he’s just doing his own thing, wherever on the ice he finds a spot he likes.

In his best seasons he shoots the puck from everywhere, and in his worst, he’s tethered to the right dot. He ranges around and shoots more than Morgan Rielly does, more than Brent Burns does, more than Erik Karlsson. (Go play with the HockeyViz Spray Charts and see this for yourself.)

If you put him on the ice with a finisher (even an old one) and a forechecker (maybe not even as good as Landeskog) he finds the white space to be the guy who takes the puck and hands it off to exactly the right guy in the right place to score. Or he takes a shot that goes in more often than most defenders manage. He’s riffing off of those around him like a jazz musician who is only good when the band is cooking. Not necessarily the most skilled band, but one with chemistry.

Or is that just a poetic way to call him a passenger on the hottest line in hockey for his one good year? I’m still not sure.