Comments / New

Florida Panthers at the 30-game mark

Ahhhh. The more things change… the more they stay the same. Just when you think the glory days are here, you are forced to remember that you are indeed… a Panther fan. The horror, the horror. Being a Panther fan means you are going to ride the roller coaster. Every great moment will be matched by a poor one, and every patch of stable ground is actually built on a fault line. A crushing defeat to Pittsburgh. A win against Vancouver! That makes us all (including me) happy. But… the fault line… the Panthers also allowed one of the worst teams in the league (the aforementioned Canucks) to take 10 more shots against them than they have averaged all season. As Yahoo Sports wrote:

Florida Panthers: Meanwhile, the old Panthers snapped a four-game losing streak because they got to play Vancouver.

A bruising defeat in St. Paul, Minnesota, where, once again, the Panthers won the Corsi battle and got destroyed in the game. There has to be an easier way. Lets get this going.

Lets start with… LUCK:

For starters, our former colleague Shane wrote an excellent analysis of the Panthers recently that really demands you check it out. It can be found here. He is looking at the same issues we have been noting:

The Panthers have been an incredibly strange team to follow this season, as they typically control play for the majority of the game, but don’t exactly dominate their opponents. Despite having control of the puck, and despite winning the shot attempt battle, it seems like the team is typically out-chanced; their opponents get more Grade A chances, and end up scoring more goals as a result.

Its amusing that we were both working on the same thing around the same time period. We come at things from somewhat different angles (I have always valued zone starts and strength of competition probably too much, and he uses a lot of advanced math that is correct yet hard for me to comprehend), but we both feel that this team looks unlucky, but isn’t actually as unlucky as it seems. Shane is finding something similar to what I am looking at:

Despite their high CF%, the Panthers have actually struggled from a scoring chance standpoint, controlling just 45.7% of the scoring chances registered in their games. That total is only good for 24th in the league, a stark contrast from the strong Corsi numbers that the team has posted in their first 29 games.

For all the talk that has surrounded the Panthers since 2015, continuing to this very day, about “luck,” I am formally petitioning to have a 4-leaf clover sewn into the staring cat logo. “Yes,” my friends, there has been a lot of talk about luck. “The Panthers were luckier than good in 2015-16,” the “Panthers were incredibly unlucky in this game,” the “Panthers had the worst luck ever seen by the analytics community in the Flyers game,” the “Panthers ran into a lucky goalie.” Yada, yada, yada. Let’s put it down people: Heck yes, luck is part of hockey, even a big part of hockey, and “yes,” some teams get luckier than others over stretches of the season. But “luck,” to me, is the Conor Sheary goal in the Penguins game that bounced off the end-boards and went in off Luongo’s leg/foot. That is luck, and it happens sometimes.

When the term “luck” is being thrown around with the Panthers, its used most times in a different manner. The Panthers were considered unlucky because they didn’t score despite taking umpteen shots, or controlling play for long stretches of a game. Well, park me in the front row for those who say that beyond mere happenstance, luck (in that sense) is very much something that can be produced and influenced by play. Odds of being lucky go up when a player plays to high probability- you can’t have a puck bounce off the butt and in if you aren’t in front of the net- ya follow? And talent, it has an impact on luck also. Even some of the analytics folks most devoted to pointing fingers and calling “lucky!” are sliding down that slope, such as Yahoo Sports (in an article about the Blue Jackets):

You also have to say that at least part of their high PDO is talent-driven. To be specific, Sergei Bobrovsky, when healthy, is one of the eight or so best goalies alive, and for once he’s been healthy all year. That’s going to help you win a lot of games even when you’re not playing well

Remember last March, when the Panthers beat Toronto on 2 lucky goals by Jussi Jokinen? Allow me to remind you that Sportsnet wrote an article about the luck. In that article, though, was this gem, displacing the pure luck argument:

Yes, the Leafs outshot the Panthers Thursday, but ask anyone at the game who dominated play, and it wasn’t even close. Puck battles, cycling, passing… Florida was far superior to the eye.Interesting. So, while the Panthers scored 2 lucky goals and than sat on the lead, they also actually played better as well, in a way the metrics had difficulty measuring. Perhaps, Florida created an opportunity for lucky chances by simply playing better– whether they took 35 shots or not? Hmmm. Lets do an experiment.

Florida outshot its opponent by 11 shots. The Panthers out-chanced their opponent by at least 3 in the low slot. Florida got 2-3 more shots on the power play in the low slot than their opponent, but lost the game 3-1, in part because despite all that, they scored no power play goals and gave up 2 power play goals against. Bad luck? If so, the chance of being repeatable with the same opponent should be random.

Fast forward about 3-weeks. Florida outshoots its opponent by 19, out chances them in the low slot by about 7 shots, but only gets 1 power play shot off in the low slot, while its opponent gets 2-3. The Panthers score no power play goals, and its opponent scores one against them. This time Florida loses in overtime. Same result- a loss. Same scenario, Florida out-chancing and out-shooting its opponent. Same loss of the special teams battle. Repeatable bad luck?

I am, of course, referring to two games against Philadelphia, on November 22, 2016 (Panthers lose 3-1), and on December 11, 2016 (Panthers lose in OT 3-2). Several analytics writers have called the December 11th loss the unluckiest thing they had seen, as Florida dominated the metrics (and “no” I am not disagreeing with them because they are “analytics” writers). But luck, by its nature as a random effect, should not be similarly repeatable within a limited time frame, against the same opponent, because….well, its “luck.” The Panthers lost the two games in similar fashion- outshooting and out chancing the Flyers on paper, and losing special teams campaigns. But what if we dive into some of those numbers deeper?

The Philadelphia Games:

When I say the Panthers had 9 shots in game 1 against the Flyers from the low slot, it sounds great. But, 1 of those shots was by Derek MacKenzie with a season shooting % (so far) of 8.1%. Another 2 of those shots were by Kyle Rau, with a SH% (career) of 6.7%, and another was by Aaron Ekblad. Ekblad is a bona fide goal-scorer, but remember, that shot was in the low slot….by a defenseman. Maybe we can see why the team is giving up odd-man rushes so much this season? But I digress.

Four of the 9 low slot chances were by non-gifted goal scoring forwards or a defenseman. Its great to get production from bottom 3 forwards, but you’d like to see more from the top. Fast forward to game 2, where the Panthers got about 12 shots against the Flyers in the low slot. 3 of those shots were by MacKenzie. 1 of the shots was by AHL call-up Paul Thompson, and 2 more were by defensemen, for a total of 6 shots (or half) by non-gifted goal scorers who don’t take lots of quality shots.

Don’t get me wrong, I am happy to see those players getting their shots off, especially as someone who demands the Panthers attack the net more. My point is, a low team shooting percentage makes this team’s PDO appear below average. It makes them look unlucky. Yet, thats not all bad luck- its shooters you don’t expect to score lots of goals, and defensemen dropping in, taking half of the high quality scoring chances. Its not luck, because itsexpected they are not going to score on most of those chances. If they were expected to score, they’d play higher in the line-up. Contrast that with what we knew was actually going on last season.

We thought the system last year was concentrating on generating higher probability chances, over random shot attempts, and it turns out…. we were right. From that same Sportsnet article:

“I definitely wouldn’t call it luck. I know our first line is a real puck possession team, and the looks they’re getting are real quality looks,” says Vincent Trocheck, one of five other Panthers scoring on more than 13 per cent of his shots. “It’s just capitalizing on chances. When you get a chance, capitalize on it, and shooting percentages will go up.”

To increase “luck,” the Panthers need far more of their low slot chances to be by shooters with great goal-scoring talent, such as Barkov, Trocheck, and Jagr. In that first game against the Flyers, Jagr, Smith, and Bjugstad each got one shot from the low slot, and Barkov got one from the edge of the low slot. That was it for top-6 players on the team (and Bjugstad is used as top-9). In the second Flyers game, Trocheck had 1 shot in the low slot, Barkov 2, and Jokinen 2. Lots of names missing on that list. As an aside, that pattern was repeated against the Penguins, on December 8th, as the Cats only took 3 shots on Matt Murray from the slot. Those shots were by MacKenzie (1) andColton Sceviour (2). The power play had no shots from the slot in that game.

To increase luck, the Panthers maybe should have taken more than 1 power play shot from the low slot in game 2, and not allowed the Flyers 2-3 low slot shots on their power plays. In both games, power play goals by the Flyers crushed the Panthers.

Here is another one to chew on before you throw out the excuse of pure luck: in game 1 against the Flyers the Panthers took 39 shots, 15 of which were by defensemen. We know that only one of those 15 (by Ekblad) was in the low slot, and Ekblad took 9 shots that game. Only 22 shots were by Panthers forwards. By comparison, Philadelphia took 28 shots that game, 22 of which were by their forwards. Lets look at game 2. In that game, Florida defensemen took 18 shots, and again, we know that only 2 of those were in the low slot. The forwards only took 26 shots. Again by contrast, the Flyers defense took 5 shots, while their forwards took 21. If you want to talk about luck, the Panthers would have to be extremely lucky to score on upwards of 18 shots from the point (which has a very low scoring probability).

Lets stick with shot-analysis and something our own Rat-a-Holic researched:

I was looking at some stats and the Panthers lead the league in Slap Shots (248) but are last in the league in Wrist shots (335). Not surprisingly the Slappers are only scoring at a 6.45% while the Wristers are scoring at 8.35%. We also have the second most missed shots (wide) behind the Flyers with 320. Also we only have 4 combined goals by deflections and/or tips. Last in the league tied with Ottawa for those two combined.

This data points at exactly what we knew was happening (defensemen taking most shots, and the team playing from the perimeter), but shows how that fact impacts our shooting %. The conclusion is that the low shooting % a lot of folks are saying is bad luck and will improve….well, its not going to improve until the team starts taking higher percentage shots, using quality wrist shots, and starts hitting the net. It also decimates the power play plan of the deflection in front. Thank you Rat!

One more thing stands out as non-luck related impact, because of its repeatability- Wayne Simmonds. The Panthers weren’t just unlucky, they did a very poor job of containing one of Philadelphia’s top offensive weapons. In game 1, Simmonds had a goal and an assist. In game 2, Simmonds had 2 of the Flyers 3 goals. The Flyers averaged 3 goals per-game against the Panthers….because they scored…..3 goals per game. The Panthers averaged 1.5 goals-per-game against the Flyers in the two games, which sounds at least balanced. But when you separate the games, the Panthers only scored 1 goal in the first game and 2 in the second. In game 1, while the Flyers star power forward- Simmonds- was unstoppable, the Panthers top line was held off the board, as only Reilly Smith scored. In game 2, Aleksander Barkov and Jussi Jokinen scored, but were still outplayed by Simmonds.

Before you write the Panthers results off as “unlucky,” understand hat the team is not playing to increase their percentages. High value goal scorers are not the ones getting the vast number of chances from high probability scoring areas, and special teams remain a problem. Return to 2015-16, when the Panthers “looked” lucky, but in actuality worked to deny high probability chances, and took more high percentage chances, instead of low percentage chances. Of course they were lucky- they increased their odds. This season, in a game that metrics folks called incredibly unlucky- luck played a part, but the Panthers also did not increase their odds by play.

And, we probably now realize why Derek MacKenzie kept getting all that playing time– he of the team leading 4 shots in the low slot against the Flyers this season. Amplify these facts against 30-games, and you see a few things that I am going to discuss further below:

1) The Panthers are still “jobbing” Corsi;

2) special teams are still costing the team;

3) the defense has cut back on their rushes into the low slot in general, which is also cutting back on scoring;

4) the non-defensive defense is still giving up too many high probability chances.

This is not to claim that the team should shoot less. What I am saying is that its hard to claim you were killed by bad luck when the majority of your shots (that are effecting positive Corsi) come from areas of the ice that you typically need some luck to score from, or by players who don’t light the lamp much. Its hard to say you were unlucky when you let the opponent’s top power forward run all over you, while your own top forwards fail to match (and Ivan Provorov completely shut Jaromir Jagr down in game 2).

All this available information can really send a writer down a lot of “rabbit holes.” Analyzing an NHL game is an unbelievable “stew” that metrics attempt to reduce to useful numbers. But the “noise” is difficult to account for in its entirety. “Noise” is a term that analytics folks use to describe random, or outside, or temporary contextualized impacts on otherwise stable statistics (amongst other things). A trend that goes 10-games starts to show a little less noise, and a trend that runs 20-games is going to start to give you very useful information. I have set about to break the season into four 20-game blocks (hence my 20-game trends article). Due to the coaching change, I will throw a 30-game analysis in here as well.

Allow me to point out that there is some inherent bias in a 30-game examination on the coaching front, not because its aimed against former coach Gerard Gallant, or present coach Tom Rowe, but because Rowe will have only coached 8 of those 30-games. I have not chosen to do a 30-game analysis to point a finger at Rowe. I put it together at this point because of some of the comments that were made after the coaching change by the new coach and team management.

The Panthers organization, in the week following the firing of Gallant began giving more interviews about the coaching change. In those interviews it became clear that Gallant was actually fired because the organization felt the team they had assembled was under-performing (apparently due to coaching) and that they could not continue on that path. Ownership believed (after the Carolina game) that a change had to be made immediately to save the season. Rowe indicates he sought an extra day, or the remainder of the road trip for Gallant, but ownership demanded it be that evening. Immediacy was apparently critical. The firing (according to the interviews) had nothing to do with analytics. The more that was stated about the coaching change being about under-performance based on coaching, the more I scratched my proverbial head.

As my 20-game article showed, there were areas the team was slipping in, from the 10-game mark, to the 20-game mark, but most of those slides were minuscule, and I pointed out that it was not necessarily fair to base a lot of conclusions on a 10-game start to a season; conclusions would come more into focus after 40-games. 40-games is half a season, so (perhaps) its unwise to take a look after 30-games. After 30 there is still time to change course, while that is more difficult after half of a season has passed. So lets take a look back, and include where this same group has gone after the change in direction:

Stat: After 10: After 20: After 30: Trend:
Goals-For 26 52 68 Down
Goals-for-per-game 2.60 (19th) 2.60 (14th) 2.30 (25th) Down
Shooting Percentage 8.3% 8.5% 7.5% Down
Goals-Against 24 53 75 Improved
Goals-Against-per-game 2.40 (10th) 2.65 (18th) 2.67 (11th) Steady
Save Percentage .910 (14th) .911 (19th) .910 (16th) Steady
Shots-For-per-game 31.3 (7th) 30.5 (11th) 31.1 (7th) Steady
Shots-Against-per-game 26.8 (4th) 29.7 (12th) 29.5 (11th) Steady
Power-Play % 12.5% (24th) 14.5% (23rd) 12.6% (30th) Down
Penalty-Kill % 78.6% (22nd) 81.4% (19th) 84.8% (5h) Improved
PDO 99.8 99.7 99.1 Steady

The Panthers are steady across many categories, but now worst in the league in power play %. Someone explain to me again how it was Gallant who got fired, but not Barr? Last season the Panthers were 24th in shots-for per game, 13th in shots-against-per game, 5th in opponent’s shooting % per game, and 5th in shooting % per game. The team has massively improved its shots-for per game since last year, and mildly improved its shots-against. BUT….the team is now 17th in the league in opponents shooting %, meaning worse than average, and 27th in the league in shooting % differential. Let me translate that for you- The Cats are taking way more shots, but their shooting % per game is one of the worst in the league when compared to what they are allowing their opponents per game. Yes, be concerned.

Obviously, changes were made after last season, with the intent to model the Pittsburgh Penguins system. The engineer behind those changes (now Coach Rowe) has now explained in recent post-game pressers that he team is still young, and now has called out the work ethic of his young players. Lets take a moment and look at last season’s Stanley Cup Champs.

The “Penguins Model,” in detail:

Much has been made of the “Penguins” model, that the Panthers have admitted they are admirers of. Where once the Panthers were heralded as a “Western Conference” team in the east, due to their size, talent, youth, and speed, they changed tacks and are looking for speed and skill throughout the line-up without regard to size. Gallant apparently felt the team needed more size in the line-up, and that was turning into a point of constant confrontation with the management team. After the Philadelphia game, Gallant called out the team for a lack of toughness. That was allegedly seen as a shot at management. Yet, when we looked at the number of shots from the slot and who was taking them, as well as what Wayne Simmonds did to the Cats, he likely had a point. Speed without regard to size sounds good- speed and skill are the modern game. But lets dig at the 2015-16 Penguins a bit here.

The Pens had the 2nd best regular season Corsi-for % in the league, and the 8th highest PDO (widely seen as a puck-luck stat). But here are some things that stand out from that team:

– The Penguins were 15th in the league in shots-against per game

– The Penguins had the 2nd best save % in the league

– The Penguins had the 2nd worst opponents shooting % in the league (they kept opponents from taking good shots)

– The Penguins were best in the league with shots-for-per-game with 33.2 per game

– The Penguins were 18th in shots-per-goal. The Pens needed 11.29 shots to average a goal.

– The Penguins were 18th in the league in shooting %.

What this tells us is that, at least in the regular season, the Penguins took a lot of shots. Let me emphasize- a LOT. Whether the majority of those shots were “good” shots is hard to say. The team’s shooting % was below the league average, as was their shots-needed per-goal. In effect, the team needed a lot of shots to score its goals. On the other side of the coin, the Penguins gave-up a league average number of shots-against. Half the league was better, half the league was worse (Florida was slightly better). Yet, both the Penguins and Panthers kept opponent’s shooting % to some of the lowest levels in the entire league. We know what caused that result for the Panthers, and we can guess that the Penguins engaged in something similar (based on those results).

When pundits say that goaltending got the Penguins to the promised land, they are 100% accurate. While San Jose made its goalies look good, giving up the 2nd fewest shots-against-per-game last regular season, the Penguins relied on their goalies, which is what led to a top-10 finish in PDO. The Penguins were not especially lucky on offense: they took a lot of shots to generate goals. But they did ride a wave of high shooting percentage (similar to the Panthers). That is worth discussing momentarily.

As discussed above, the Penguins had a below-average team shooting %, while the Panthers had one of the best Shooting % in the league last season. As we saw, the Penguins had the most shots-for-per-game last season in the league, while the Panthers were near the bottom at 24th. The optics are that the Panthers were lucky, and the Penguins far less so, and that is somewhat accurate. But, we also saw last season that the Panthers tended to eschew low percentage scoring shots, and took far more high percentage shots as a portion of their total. As “high-percentage” shots there is actually less luck involved in goals. I did not study the Penguins shot-location charts from last season, so I cannot answer that question for them other than to say- if they similarly took mostly high percentage shots than they were extremely unlucky last season (based on the high numbers of shots needed to score goals).

And the speed team? Yes, the Penguins are very fast, but they were not small. The only player who saw regular time shorter than 5’11 on the team last season was Conor Sheary. Sheary posted 10 points in 44 regular season games, averaging only 9:45 in time-on-ice a game. The small left wing posted a great Corsi-For % of 57.5% but (you guessed it) also started a whopping 61% of his shifts in the offensive zone. He exploded in the playoffs with 10 points while averaging 13:58 in TOI, but he also had a shooting percentage of 10.5%.

Yet, the remainder of the Pens line up was all 5’11 or bigger (and only 6 players in the entire line-up were under 6’0), and that included Patric Hornqvist, who, at 5’11, parked himself in front of opponent’s nets and took heaping spoonfuls of abuse to score 22 regular season goals. He was, and remains, a huge part of the Penguins game.

Writers have been critical of the Panthers for having an unsustainable shooting percentage in 2015-16. They were talking about this:

Name: Shooting % 2015-16:
Jagr 18.9%
Barkov 16.4%
Trocheck 14.4%
Smith 14.5%

You may recall that the team was criticized in some circles for re-signing 3 of those players to sizable contracts because they were running “unsustainable” shooting percentages (I was not one of the critics, nor am I now). Before we decide to emulate the winning game plan, you’d best realize that in the playoffs last season, the Penguins ran this:

Name: 15-16′ playoff Sh%
Bryan Rust 17.6%
Matt Cullen 16.0%
Tom Kuhnhackl 13.3%
Patric Hornqvist 11.3%
Conor Sheary 10.5%
Phil Kessel 10.2%
Carl Hagelin 10.2%

Combine that with Matt Murray in net running a .923 save percentage and 2.08 GAA for the playoffs, and your getting a strong indication that as good as they were, the Penguins also ran serious hot goaltending and shooting at the right time (as did San Jose). What can we take-away from this?

There can be no doubt that speed and skill throughout the line up are THE game in 2016-17. But don’t be fooled by the NHL story line of the “small and fast” Penguins. The Penguins are fast, but they also feature world class talent across 3 forward lines with Crosby, Kessel, and Malkin, and they are joined by proven high-level producers in Kunitz and Hornqvist and Hagelin.

Sydney Crosby is on a goal scoring tear this season, and the vast, vast majority of those goals are being scored in the low slot. Patric Hornqvist does his work around the net. Chris Kunitz gets down and nasty in the low corners. Gino Malkin is a power forward. The team uses both size and speed and tenacity to control possession. As NBC Sports recently noted:

A lot of this is the result of having a team that rolls out four lines of forwards every night that possess the ability to score (including three of the most talented forwards in the league in Sidney Crosby, Evgeni Malkin and Phil Kessel), combined with a blue line that is made up almost entirely of puck-movers and offensive-minded defensemen.

Beware, however, modeling from a team that caught fire. Fast-forward to this season and the Penguins are still one of the league’s best teams, but there are concerns. The Penguins have one of the worst GAA in the league, are leading the league in shots-for, but are one of the absolute worst in shots-against. That is not a recipe for long-term success (as NBC Sports also pointed out(:

Because of that, is also not a style of play that has resulted in a lot of success in thisera.

Over the past 10 years only one team has won the Stanley Cup finishing worse than seventh in the league in goals against (the lowest ranking over that stretch: The 2008-09 Penguins were 17th. Six of the Cup winners were in the top-two, including three that were the best in the league).

The Penguins have not yet shown that they can stabilize their defensive end, and as the stat quoted above shows, it is still defense and goaltending that wins championships.

The Possession stats:

Lets return to our regularly scheduled program, evaluating the Panthers possession metrics at the 30-game mark:

Name: Face-Off Win % 10 games: 20 games: 30 Games: Trend:
Aleksander Barkov 72-66 (52.2%) 142-152 (48.3%) 215-233 (48%) Steady
Vincent Trocheck 94-105 (47.2%) 206-202 (50.5%) 301-297 (50.3%) Steady
Denis Malgin 17-36 (32.1%) 40-70 (36.4%) 49-83 (37.1%) Steady
Nick Bjugstad N/A N/A 24-26 (48%) N/A
Derek MacKenzie 63-59 (47.3%) 102-101 (50.2%) 141-140 (50.2%) Steady

Name: Corsi-For % after 10 20: 30:
Vincent Trocheck 58.4% 55.6% 55.5%
Reilly Smith 57.6% 56.3% 56.0%
Jaromir Jagr 57.5% 55.9% 58.7%
Jason Demers 56.6% 53.1% 53.1%
Jonathan Marchessault 56.4% 54.4% 55.0%
Colton Sceviour 55.5% 51.4% 51.3%
Aleksander Barkov 55.2% 54.6% 57.5%
Mark Pysyk 54.7% 51.1% 52.7%
Aaron Ekblad 53.2% 51.1% 52.1%
Michael Matheson 52.9% 51.4% 53.6%
Keith Yandle 52.8% 50.5% 52.4%
Denis Malgin 51.1% 50.8% 53.0%
Derek MacKenzie 50.0% 45.6% 43.3%
Kyle Rau 47.6% 46.5% 46.5%
Nick Bjugstad N/A N/A 52.3%

Lets adjust those numbers with a stat that I still think matters:

Name: Off-Zone Start Rate % 10-game: 20-game: 30-Game:
Denis Malgin 71.7% 65.5% 66.5%
Keith Yandle 62.6% 55.6% 57.3%
Aaron Ekblad 60.8% 56.1% 56.9%
Jaromir Jagr 59.8% 52.7% 53.6%
Aleksander Barkov 58.3% 53.5% 55.1%
Colton Sceviour 58.3% 48.5% 50.7%
Jonathan Marchessault 54.9% 53.1% 55.5%
Jason Demers 53.3% 49.7% 51.7%
Mark Pysyk 52.4% 48.3% 48.7%
Derek MacKenzie 52% 44.5% 42.9%
Reilly Smith 51.5% 50.7% 52.2%
Mike Matheson 50.6% 49.7% 50.5%
Kyle Rau 50% 48.8% 46.2%
Vincent Trocheck 49.1% 47.6% 52.1%
Nick Bjugstad N/A N/A 59.7%

Are the Panthers jobbing the Corsi system? Only two players on the team are running a Corsi % under 50%, and both of them are buried in the defensive zone. Looks like the remainder of the team is posting better than 50% Corsi rates. Yet, the team is scoring virtually no goals. “Yes,” it appears the Panthers are taking a lot of shots-for, but not many “quality” shots-for.

The Defense:

I (along with many others) have questioned the sum total of the off-season defensive changes the Panthers made. It is evident that the Panthers are giving up more high-probability scoring chances than they did in 2015-16. As I have posted in comments recently, we now know what we suspected was the case last season, which is: the Panthers coaching staff was, in fact, using metrics and video to identify Roberto Luongo’s weak-spots, and designed the entire defense to protect those shooting areas of the ice. Think of it as hockey’s equivalent of the “left-tackle” in football, but it became known to the public (opponents surely already knew) in last-season’s playoffs that Lu is weak on the blocker side, meaning the right defensemen is the key to high probability shut down defense. Last season that position was held by Aaron Ekblad, Erik Gudbranson, and Alex Petrovic.

We might as well re-address this issue at this point: the Gudbranson trade. It would be overly simplistic to say that the trade is a net positive or negative at this point (and quite frankly, it may be that the loss of Kulikov will- long term- be a more negative impact). Its not really either, yet. Gudbranson has continued to struggle with Vancouver, especially with inconsistency. He shut down Edmonton’s Connor McDavid line admirably in one game, but is still posting terrible Corsi-for numbers. He remains a much maligned player by the analytics community. Yet, some interesting things start to emerge when you look at Gudbranson’s numbers from last year with some context. Lucky for us, the Vancouver Courier did just that, starting its look with Corsi (where Gudbranson was near the bottom) :

One interesting thing to note is the one name consistently at or near the bottom of the chart, below Gudbranson’s: Willie Mitchell. At 39, the former Canuck doesn’t appear to be the defensive stalwart he once was. He was also one of Gudbranson’s most common defence partners.

There’s an argument to be made, then, that Mitchell dragged Gudbranson down, with the caveat that they faced difficult minutes together.

Gudbranson’s other defence partner last season was Brian Campbell. When they skated together, they posted a 52.5% corsi and had a goals for percentage of 62.5%.

You could argue that Campbell carried that pairing, but it seems clear that Gudbranson didn’t drag Campbell down, at the very least. When Campbell was paired with Mitchell, on the other hand, they had a corsi of 43.5%

Now, recall that I said the right defensemen were the key to the entire defensive scheme? Gudbranson was 17th in the league in defensive zone starts last year, and got the most playing time of any Panther in the playoff series with the Islanders. Only 16 defensemen in the league last season started more shifts in the defensive zone. It is clear which of the right D-men the coaching staff was using to help Roberto Luongo. That becomes even more clear when you realize that Aaron Ekblad started 55.6% of his shifts in the offensive zone last season. Gudbranson and Petrovic were doing the heavy lifting in the defensive scheme, and Petro only averaged 16:57 in time-on-ice last season. It is quickly becoming apparent what a seismic shift the Panthers undertook this summer. But wait, there’s more.

Who was tasked in 2016-17 with replacing the heavy minutes of defensive zone use, against opponents top goal-scoring lines the Panthers lost in Gudbranson and Dmitry Kulikov? We just saw Aaron Ekblad’s zone usage last season, but Keith Yandle was even more protected by the Rangers last season. Only Dan Boyle faced less competitive opposition by use for the Blue-Shirts. Lets pause here and take a look at last season for the Panthers d-men:

Name: 15-16′ OZ start rate: 15-16′ Corsi-For %: 15-16′ Avg. TOI:
Yandle 66.7% 50.3% 19:58
Pysyk 55% 51.9% 15:54
Ekblad 55.6% 51.4% 21:11
Demers 49% 54.1% 20:52
Petrovic 45.7% 48.7% 16:57

Perhaps you believe that I am over-hyping zone start information, but we will get to that in a moment. Reports are starting to trickle out that maybe the Panthers analytics team is using “first-pass” rate-data to evaluate defensemen. Let me be clear- I do not have proof of this, but that stat has been mentioned by some Panthers management in recent interviews, so it seems to be something that is being used. To be further clear- I don’t know what modifiers the analytics team is using, but based on what we see above, there are some warning signs.

Start with this: a defenseman’s first-pass success rate is probably going to be better when they are being used against low quality of competition, and when they are used less in the defensive zone, or playing fewer minutes per game. Mark Pysyk, who is now being tasked by the Panthers with heavy defensive zone minutes (by comparison to his fellow Panther defensemen this season), had a 55% offensive zone start rate for Buffalo last season. A review of the Sabres blog “Die by the Blade,” reveals that many Sabres fans felt Pysyk was an analytics darling whose lack of physicality let opponents walk all over the Buffalo defensive zone. Yet, against lower quality of competition, Pysyk performed well analytically. Thats easy to explain, isn’t it? He played relatively few minutes, against lower quality opponents, giving him more time for puck collection after dump-ins (the preferred zone entry for 3rd and 4th line forwards, as well as by defensemen who are retrieving those dump-ins), and a better ability to separate opponents from pucks at or between the blue lines.

Usage is a big deal, and coaches know it. Returning for a moment to our look at the Penguins, Pittsburgh Coach Mike Sullivan has been lauded for what he has gotten from defenseman Justin Schultz. Yet, consider that Sullivan is using the exact methodology we are discussing here (from NBC Sports):

Sullivan too deserves credit, for playing Schultz to his strengths and bringing out the best in an imperfect player. An offensive defenseman, Schultz is not being asked to do much of the heavy lifting defensively. He starts a lot of shifts in the attacking zone, and he doesn’t play very often against the opposition’s best.

The Panther’s new issue is, who doesn’t have to be started in the offensive zone? We have seen that Alex Petrovic played limited minutes last season. But in those minutes he did see a 45.7% offensive zone start rate. He was used heavily on defense, allowing Ekblad to get the offensive use. Petro was clearly expected (with Pysyk) to play a shut down role this year. His injury has complicated defensive problems for the team. In a limited data pool (prior to Petrovic’s injury), Pysyk and Petro were the Panthers best Corsi d-pairing, albeit in limited minutes. What happens with more minutes and better opponents can only be seen on Pysyk now.

Perhaps the Panthers were counting on Petrovic, Pysyk, and Demers to bring defensive ability to go along with offensive prowess. But the zone starts this season show that the Panthers are still heavily protecting Yandle, and to protect him further, are using Demers as his partner (resulting in mostly offensive use for a player that has shown he can handle defensive use). I will call this the “McIlrath effect.”

Much has been made of Dylan McIlrath’s success last season in New York….. when paired with Yandle. Yet, by receiving that assignment, McIlrath benefitted from the same offensive zone use, and low quality of competition that Yandle was getting. In fact, McIlrath saw an even lower quality of competition than Yandle, the lowest of the Rangers 7-defensemen last season. Thus, he got the attendant boost, and his beneficial Corsi numbers got him traded for. Demers is now getting those favorables, rather than being used more defensively, a role in which he has shown himself capable.

The problem now is that there are fewer options behind the Yandle pairing, because Ekblad is also being heavily sheltered, with his new second pairing partner, Kindl. The results along the Panther blue line have shown what could have been predicted from history.

Let me be very clear- I do not know what metrics the Panthers analytics department relies upon for defensive evaluation. Yet, I do know that there is a great deal of controversy in the analytics community about evaluating defensemen. Take a look at this article for some of the controversial questions that still exist. One of the conclusions from that article:

Goal stats are notoriously tricky to deal with because of small sample size and variance, and because of that we’re likely missing a large part of possible teammate and competition effects in our analysis.

There’s so many factors that go into a player’s results that I don’t think we fully understand just yet. We’ve got a good idea, but it’s an extremely elaborate web that’s still being untangled. How the other nine players on the ice affect a single player is perhaps one of the biggest hurdles left to tackle.

So many variables come into play that pundits are still arguing about, such as time-on-ice, usage, quality of competition, quality of teammates, and quality of line mates. That article sums up a likely, but still unfounded conclusion:

Some minutes- for d-men at least- are in fact easier than others, with a range much larger than whats previously been measured.

Lets translate that via the Panthers: Aaron Ekblad posted a far better Corsi-for, and Corsi-for-Relative than Alex Petrovic last season….while starting 9.9% more of his shifts on offense. Petrovic (we noted, had less than 50% of his starts on offense) had a Corsi-For % of 48.7% last season. That difference in zone usage is 6.9% this season, before Petro was injured, which was alarming because Ekblad’s Corsi-for was still 2.7% worse than Petrovic’s after 20-games. Some of that likely is explained by the difference in time-on-ice. There is a very muddy collection of conclusions you could draw from this, but the easiest (after reviewing last season’s data as well) is that Ekblad was not yet ready for un-sheltered defensive zone use.

Whatever it was the team had planned for the defense based on numbers suggesting that 3 of its 6 d-men could not necessarily play shut-down defense, another is a barely tested rookie, and the other struggled in defensive use last season, the results have been clear. The Panthers are getting scored on from high percentage areas of the ice, and are no longer protecting Lu’s blocker side. Nonetheless, Lu (thankfully) has posted admirable numbers.

The Goaltending:

Here are the goaltending numbers after 30-games:

Name: GAA 10-game: GAA 20-game: GAA 3-Game SV% 10-game: SV% 20-game: SV% 30-game:
Roberto Luongo 2.30 2.31 2.24 .912 .921 .924
James Reimer 2.61 3.05 3.02 .907 .897 .894

Roberto Luongo has been carrying the team for the first 30-games. His Save percentage is above the league average, despite facing 29.75 shots per game. Put Reimer into the equation and it knocks the team save percentage down to league average.

Point-production:

Name: Goals 10 game: G 20-game: G 30-game: Assists 10-game: A 20-game: A 30-game:
Marchessault 6 9 10 5 7 10
Sceviour 5 5 5 2 4 6
Trocheck 4 6 7 0 7 8
Matheson 2 2 2 2 5 7
Barkov 2 2 6 3 9 14
Ekblad 1 5 7 0 0 2
Smith 1 4 4 2 3 7
Jagr 1 2 6 3 7 9
Pysyk 1 1 1 2 2 5
Malgin 1 4 4 2 4 4
Yandle 0 1 1 4 10 12
Demers 0 2 4 3 6 6
MacKenzie 0 2 3 2 3 4
Jokinen 0 1 2 2 3 3
Bjugstad N/A N/A 0 N/A N/A 0

Over the last 10-games, Barkov, Jagr, and Smith have had the only somewhat noteworthy jump in goals or assists. But that is not saying much. The team is not producing much in the way of goals. The Panthers only scored more than 2-goals in 2 of their last 10 games. In Tom Rowe’s 8-games coaching, the Panthers have been out scored 24-14. With what I discussed above, its hard to say this is surprising.

The biggest news is that the Panthers are suddenly getting zero production from Vincent Trocheck. Vincent scored 25-goals in 2015-16, and added 28 assists, and basically, single-handedly dropped Nick Bjugstad into the bottom-6 forward pool.

Suddenly, over the last dozen (and more) games, he can’t find the net (unless its empty). His shooting % has dropped from a season average in 15-16′ of 14.4% to 9.6% this season. This is odd, because, while 14.4% may not have been sustainable, Trocheck is the only center on the Panthers who returned this season with the same wings he had last season (Smith and Jokinen). At least over this short term, his Corsi % is also up over his season % last year. What is going on?

To start with, Trocheck’s PDO has dropped through the floor (so far). His 102.9 PDO last season was high. Yet, we have now discussed ad nauseum how the team system last season resulted in a high PDO possibly based on something more than pure luck, due to high percentage scoring area shots-for. So what else has been happening?

Aleksander Barkov. Barkov’s cold, as in sub-zero, start to the season, along with Trocheck’s hot start caused opponents to focus on…..(you guessed it), Trocheck. Further, after 20-games, Trocheck’s usage had adjusted negatively for offensive production. After 20-games, the 2nd line center was getting even fewer offensive zone starts than he did last year (when he was under 50%). Apparently in an attempt to get the young star going again, that % has massively flipped over the last 10-games, as Trocheck now sports a 52.1% offensive zone start rate. But that has still not worked. So, what else?

Time-on-ice. Trocheck averaged 17:46 last season. So far, this season, he is averaging 21:01 ATOI. He is also hitting more. Last season he averaged 1.64 hits per game. That is now up to 2.43 per game. He is taking about the same number of shots per game. Last season, he averaged 2.28 shots per game, which is up slightly this season to 2.43 per game. But with the extra minutes he is playing, and more offensive zone time, that is concerning.

None of this is to say that Trocheck is not playing hard. Frankly, he may be the forward playing the hardest through the first 30-games. But its likely that some combination of factors is shutting him down: 1) yes- poor luck; 2) better opposition; 3) more playing time with more hits and possession is wearing him down; 4) systems are not geared to produce.

That last one is of interest. Note that Trocheck was one of the team leaders in goals over the first 10, cooled off in the second 10-games goal-wise, but exploded as a set-up player with 7 assists over those 10-games, and has now collapsed across the offensive production scheme over the last 10-games…..while suddenly being usedmore offensively. And we noted from the Philadelphia games above, that he is not getting much in the way of high % scoring chances.

I confess that I am not entirely sure why this is happening to Trocheck, but can definitely see a trend that goes like this: first 10- goal scorer. Second 10- set-up player (as defenses worked to shut him down). Third 10- more offensive zone minutes and zero production. Could this be what is now allowing Barkov to produce more in the last 10? Barkov is playing fewer minutes than Trocheck and getting even more offensive zone starts. Have opponents decided to use their shut down defense on Trocheck instead of Barkov? That could be the case. Nonetheless, it appears the team is not playing to Trocheck’s strengths.

Lets look at another Panther missing-in-action so far. Since his return, Nick Bjugstad has also been a negative player. Bjugstad never really jumped back to his old scoring totals from 2014-15 after returning from an injury and being placed on the 3rd line. I try to work from evidence though, and if ever there was evidence of the player he still is, it was in the playoff series against the Islanders.

With Trocheck injured, Bjugstad lined up between Jussi Jokinen and Reilly Smith, and the results were amazing. It was the Panthers best line in the series, and Nick was a key to its success. He averaged 19:08 in ATOI against the Islanders, and in 6 games had 2 goals and 2 assists, while going +6. He passed the eye-test as well, as a physical beast along the boards and around the net. In game 6, he was moved back to the 3rd line and production ceased.

Why than, with none of those four players producing much this season, has a return to that line not been attempted? Do the analysts (who are reportedly feeding the Panthers coaches line ideas) remain unaware of the magic that line combined for against a team in the high-pressure playoffs that shut-down all the Cats other forwards?

In 2014-15, when Bjugstad put up 24 goals in 72 games, he averaged 16:35 in ATOI and a 52.8% offensive-zone start rate. After returning last season, he averaged only 15:31 in ice time-per-game with terrible wings. His production dipped to 15 goals in 67 games (during the regular season). As a 3rd line center, Bjugstad only started 46% of his shifts in the offensive zone last season.

So far this season, in 10-games played, Bjugstad is averaging only 15:31 in ice time, once again with rotating wings. Put simply, before you toss Bjugstad to the curb, try giving him 2nd line minutes and pairing him with wings he had proven success with. If he fails there, you have an answer.

Lets finish this up. The Panthers are still hanging around a playoff berth, so hope is not lost. Yet, play is trending in a bad direction. The team’s goal scoring has collapsed over the last 10-games, the power play has actually gotten worse, and the improvement in shots-against has been mild, and not substantial enough to off-set the loss of offense. Very simply, the Panthers remain a Corsi devotee’s darling, but the results have been terrible. Have the Panthers gone too small? The Cats have 7 forwards playing right now (although Malgin has been in the press box) under 6’0. The list of the NHL’s top 15 goal scorers includes only two players under 6’0: Sydney Crosby and Nikita Kucherov. It may still be a big man’s league. Thats an overly simplistic response to the data. But one thing is certain, and that is: the team that has been fielded does not look capable of doing whatever it is Rowe and company wanted them to do.

Talking Points