2016-17 Flyers season review: Shayne Gostisbehere’s ‘sophomore slump’ clouds his future

This time last year, Shayne Gostisbehere was on top of the Philadelphia sports world. He had just concluded a rookie season in which he had exceeded all reasonable expectations, became an immediate fan favorite, and nearly became the Flyers’ first Calder Trophy winner in franchise history. Gostisbehere had provided a shot in the arm for a team in desperate need of dynamic defensemen, and established himself as the first successful piece of the team’s coming youth movement.

In evaluating Gostisbehere’s rookie season, there were a few minor red flags among all the positives, even if most fans were content to ignore them in the moment. While his goal-based outcomes at 5v5 were fantastic, Gostisbehere didn’t drive shot differentials especially well, even if that issue was largely attributable to extended time playing alongside Andrew MacDonald. In addition, it was always highly unlikely that Ghost was going to repeat an 11.2% shooting percentage, considering the vast majority of shots taken by defensemen come from distance, just by the nature of the position. However, Gostisbehere graded out very well by microstatistics, specifically those that measured defensive zone exits and neutral zone play. Even if the scoring rates regressed a bit, the thought was, Ghost’s underlying numbers should improve and make up for a possible points dip.

But even the most pessimistic observers couldn’t have predicted what actually happened to Gostisbehere in 2016-17. It wasn’t a just sophomore slump — it was a true Murphy’s Law season, where the young defenseman dealt with awful shooting luck, occasional defensive zone breakdowns, and a coaching staff that believed in repeated benchings as a way to fix what they felt ailed Gostisbehere’s game.

Raw scoring totals dipped dramatically

An offensive defenseman like Shayne Gostisbehere will always be judged, first and foremost, by his goal and assist totals. That’s not totally fair, as driving positive on-ice outcomes for the team is in many ways more important than individual totals, but it’s a simple fact. In those areas, Gostisbehere’s 2016-17 season was undoubtedly a disappointment.

After scoring 17 goals and 46 total points in 64 games during his rookie season, Gostisbehere regressed to a 7-32-39 line in 76 games this time around. But raw totals only give us a snapshot into a player’s production. In order to fully understand the nature of Gostisbehere’s points decline, it’s best to break up his metrics by situation and adjust for time on ice. Then, it should be obvious where the issues truly lie.

Interestingly enough, Gostisbehere’s power play point production actually improved in his second season, even if his goal totals dropped significantly. The bigger issue was in his 5-on-5 scoring, where Ghost fell from 12th in the NHL in 5v5 Points per 60 (among defensemen with at least 300 minutes) last season, all the way down to 187th in 2016-17.

This mostly lines up with the narratives that dogged Gostisbehere all season long. On the power play, he couldn’t find his mark as often with his point shot, and he was supposedly invisible at even strength for extended stretches. In terms of raw scoring, the tales are absolutely right. A player with Shayne Gostisbehere’s skating ability and puck skills shouldn’t be down in the region of Nick Schultz and Dion Phaneuf when it comes to 5v5 scoring.

However, Ghost’s scoring issues weren’t totally his fault.

When it came to puck luck, it was the season from hell for Gostisbehere

The concept of “luck” in hockey gets thrown around a lot in analytic circles, and it can be often criticized by traditionalists as a way for “the nerds” to discount qualities like leadership and intangibles. But anyone who has watched a goal scored due to a weird bounce off a player’s leg or a strange carom off the end boards would agree that puck luck absolutely exists in this sport.

Sometimes, the puck simply won’t find its way to back of the net when a certain player is shooting it. By the same token, sometimes every single mistake that a player makes seemingly ends up in the back of his own net. It may not mean that the shots he took were weak, or that the mistakes were especially egregious. The outcomes were just negative. That’s essentially what happened to Shayne Gostisbehere this past season.

Let’s start with his offense. When evaluating the luck that he faced on the goal creation side, it helps to look at two things in particular — personal shooting percentage, and on-ice shooting percentage. The former should shed light on the causes of Ghost’s personal goal scoring problems, while the latter speaks more towards his teammates’ issues (and therefore his assist totals).

After posting a 11.2% shooting percentage in his rookie year that was probably inflated due to good fortune, Gostisbehere dropped all the way down to 3.2% in his sophomore season. Now, shooting percentages themselves don’t prove good luck vs. bad luck. Maybe Ghost’s shots in 2015-16 were simply better than the ones that he took this season, and that’s the reason for the discrepancy. Luckily, we can test that theory, using the Expected Goals statistic and its components. If Gostisbehere’s shot quality was truly worse this season, it should show in his Expected Fenwick Shooting Percentage, which attempts to determine what his goal scoring rate on all unblocked shot attempts “should have” been considering the locations and types of shots that he released.

What this chart shows is that Ghost probably scored a few goals too many in his rookie year, but then was hit with corresponding awful luck in Year 2. Yes, the year-to-year shot quality appears to have dipped a bit, but nowhere enough to explain Gostisbehere’s dramatic dropoff in goals scored. In addition, Ghost actually took even more shots at 5v5 this season versus last year, averaging 6.30 Shots/60 in 2016-17 versus 4.60 in 2015-16. Shot volume was up, shot quality appears to have been basically stable, yet Gostisbehere’s goal numbers fell off a cliff. His process sure seems sounds, but for some reason, the results just didn’t come easy for Ghost this year.

On-ice metrics even more unlucky

The same can be said for his on-ice goal metrics. Gostisbehere’s horrific plus/minus rating was noted on multiple occasions this year by detractors, and a -21 certainly jumps off the page. Advanced goal-based metrics told a similar sad story in Ghost’s 37.3% 5v5 Goals For percentage (when he was on the ice, the Flyers scored 28 goals and allowed 46). For some, that’s an open-and-shut case. Ghost was a defensive liability, had a terrible plus/minus, and therefore had an awful year.

But let’s take a step back for a second and analyze the possible causes for such a poor performance in that statistic. There are three main ways you can end up with a terrible on-ice goal differential at 5v5 over a full season.

1. Be on the ice when your team is constantly buried in the defensive zone (bad shot differential)
2. Be on the ice when your goalies can’t stop a puck
3. Be on the ice when your shooters can’t buy a goal

Category #1 would be especially concerning, because it implies that the Flyers are not driving play with Gostisbehere on the ice. Categories #2 and #3 are subject to more variance, but could potentially be impacted by the player’s on-ice actions, though it should show up in his shot quality-based metrics if the impact is substantial. So where does Ghost’s 2016-17 fall?

It’s clearly not the territorial play, as his Corsi shows. The Flyers generated almost 53% of the overall shot attempts with Gostisbehere on the ice, so raw volume isn’t the problem. At first glance, on-ice save percentage doesn’t seem to be the issue either (he ranked fourth on the defense), until you remember that 91.67% is still not very good. For reference, the league average save percentage at 5v5 this season was 92.32%, so Ghost was in the middle of the pack on a team that received bad goaltending.

And then we get to the on-ice shooting percentage, and the biggest issue shows itself. The Flyers simply could not score goals with Shayne Gostisbehere on the ice at 5v5 last season. He’s a full percentage point behind seventh-place Mark Streit, and around 4.5 points behind offensive dynamo Nick Schultz.

Of course, one could argue that Ghost contributed to that low number. After all, he struggled to get shots through traffic, right? (Not really.) And he looked like a shell of himself in the first half, correct? (As we’ll see later, not quite.) Maybe that’s why the Flyers couldn’t score with him on the ice.

But one player on this list throws off that whole theory. Nick Schultz, the player who has never scored more than six goals in a season in his 15-year career, and is poised to retire because he’s slowed down so much, somehow finished with the best on-ice shooting percentage among Philadelphia defensemen. What’s more likely — that Nick Schultz found some way to magically help his teammates score goals at an elevated rate this season, or that on-ice shooting percentage is a stat subject to a ton of season-to-season random variance?

In fact, both on-ice save and shooting percentage are both notoriously erratic statistics. So if a player’s “awful” season is almost entirely due to that combination, it’s right to be skeptical of whether his year really was that bad, or if every break just went against him while he was on the ice. And let’s be very clear here — Shayne Gostisbehere’s 2016-17 was especially bad in terms of these percentages.

PDO, which is a essentially a shorthand way to look at both 5v5 on-ice shooting and save percentage by adding the two, is far from a perfect stat. But it does give us a rough measurement of “luck” over the course of a season, as the vast majority of players usually finish right around 100 in terms of PDO. Shayne Gostisbehere, on the other hand, posted a 96.51 PDO this season. Not only was that the worst on the Flyers, it ranks 1107th among 1,133 total defenseman seasons (with more than 500 minutes at 5v5) since 2010-11. 97.6% of all the seasons by qualifying defensemen came with higher PDOs than the one Gostisbehere posted in 2016-17.

The good news for the future is that defensemen who post horrific PDOs in one season tend to bounce back dramatically the following season, so long as they are given regular ice time (as Ghost certainly will receive). Out of the 50 defensemen who posted PDOs below 97 in full seasons since 2010-11, 30 played in at least 500 minutes the following year, and those blueliners averaged a perfectly-respectable 99.45 PDO after their disastrous one.

The lesson here is twofold. First, if a player has a bad PDO year, it doesn’t mean he’s likely to repeat it, even if that poor season is historically awful. Second, the nature of Shayne Gostisbehere’s particular bad year shouldn’t have anyone too worried. The places where he graded out the worst don’t have much in the way of year-to-year repeatability, and while his Expected Goals For percentage at 5v5 (48.38% score-adjusted) was worse than his Corsi, it’s nowhere close to the awful 37.3% Goals For percentage that he posted.

It was a rough year. But the numbers tell us that Gostisbehere isn’t likely to repeat the awful on-ice percentages that torpedoed his sophomore season.

Scratchings and the impact on Ghost

The Flyers apparently didn’t view Gostisbehere’s poor on-ice results as being caused by primarily by poor fortune. After starting out the season essentially being used by head coach Dave Hakstol as a first-pair defenseman, Ghost’s ice time gradually dwindled until he was receiving third pair minutes on a nightly basis. The change in role was the more subtle adjustment by Hakstol, however. Much more obvious was the coach’s decision to scratch Gostisbehere five times during the 2016-17 season — two one-game benchings on November 17th and January 14th, and then a three-gamer in early February.

Predictably, fans were outraged. Part of that was due to the continued presence of players like Andrew MacDonald and Brandon Manning in the lineup while Gostisbehere sat, but the bigger concern was that the Flyers’ coaching staff might be trying to change Ghost’s style of play and make him more risk-averse through the use of disciplinary measures like scratchings.

The concern was driven by Gostisbehere’s perfectly-fine advanced metrics. Prior to the first scratch, Ghost ranked fourth on the defense in score-adjusted CF% and second in xG%; prior to scratch #2, he led the defense in both categories. With Gostisbehere on the ice, the shot-based outcomes were all positive, implying that his overall process was sound. Yes, the bounces weren’t going his way, but an overreaction to an awful Goals For percentage (36.15% prior to the second scratching) risked de-incentivizing a process that was resulting in territorial dominance.

That’s not to say that Gostisbehere was a perfect defenseman over the season’s first four months. He was making some glaring mistakes in coverage while skating in the defensive zone, and his on-ice Expected Goals Against per 60 rate was correspondingly high (2.67 — third-worst on the defense on January 13th). But the Flyers were creating lots of offense with Gostisbehere on the ice to make up for the defensive issues (a defense-high 2.70 xG For per 60 and 63.95 shot attempts per 60 at 5v5), even if the puck simply wouldn’t go into the net for him. Had Gostisbehere simply been allowed to play through his awful percentages, things likely would have improved on their own.

Instead, the Flyers evaluated that a benching was the right course of action. And when Gostisbehere’s flashy plays came back late in the season, it seemed like the coaching move may have worked. Ghost’s strong finish was also attributed by some to a return to health, as it had been theorized that the defenseman was hampered by his offseason sports hernia surgery during the season’s first half. The combination turned into the prevailing narrative at the end of the year — that health in addition to renewed confidence had brought back the old Gostisbehere, and the scratchings either helped him get back to that point, or at least didn’t have a major negative impact.

However, there is a nagging issue with that narrative, as perfect as it seems. Gostisbehere’s advanced metrics, so strong in the early season when the defenseman was supposedly a total disaster, were declining just as Ghost was supposedly regaining his old form.

Gostisbehere stayed above the 50% Corsi mark through the entirety of the first half, and hovered right around break-even in xG. The second half, however, saw Ghost spend the majority of his time below the 50% line in both categories. Things did pick up at the very end of the season, but it wasn’t like those last few weeks were markedly better than Ghost’s results prior to being scratched in the first place.

It’s easy to build a narrative around this, as well. Maybe Gostisbehere shook off the first benching easily, made few changes to his game, and continued to drive positive shot-based outcomes as a result. But with his goal-based percentages staying low, the team benched him a second time, and that’s when Ghost’s play really fell off a cliff. Maybe it hurt his confidence. Or even more concerning, maybe Gostisbehere was trying to make changes to his game that were actually detrimental to his on-ice results.

Luckily, other evidence calls into question the theory. Gostisbehere’s manually-tracked microstats didn’t dramatically change over the final 38 games of the season versus the first 30, and such a shift would seem like an obvious indicator of an intentional change in style of play. Ghost still was driving controlled exits better than any Flyers defenseman not named Ivan Provorov, and he was still aggressively breaking up zone entries in the neutral zone.

His rate of forcing dump-ins when being attacked directly on the rush did decline, but it did the same for every Philadelphia defenseman as the season wound down. Ghost remained second on the club in both datasets, implying that even as team-wide tactics changed, he continued to attack opposing forwards more than his peers did. All of these are good signs. In addition, he spent a large portion of the second half playing with Nick Schultz, who is probably an even worse play-driver than MacDonald.

Still, these charts are a bit worrying. Ghost took a major step forward in the first half of 2016-17 in terms of play-driving ability, a legitimate weakness of his rookie season. Then, he gave back most of those gains in a second half that supposedly saw the defenseman “improve.” At least by the numbers, Gostisbehere’s first half looks like an example of “good process, bad results,” while the second half is more “mediocre process, mediocre results.” The concern is obvious — will Gostisbehere be trying to replicate the right process next season, and does the team even want him to do so?

Next season will be pivotal for Ghost’s career trajectory

At age-24, Shayne Gostisbehere has reached what can be viewed as a career crossroads. His rookie season was undeniably fantastic, as he rode the hot hand for 60+ games to a surprise Calder Trophy nomination. But he did need unsustainable shooting percentages to put up those eye-popping numbers, and his underlying metrics were just decent, not outstanding. Season No. 2 was also mesmerizing, at least if you consider spectacular accidents the primary reason to watch a NASCAR race. The unsustainable percentages didn’t just come back to earth; they burrowed all the way through to the planet’s core. On the other hand, those pesky underlying numbers did get better, even if they trended downwards in the second half.

In the first half of 2016-17, Gostisbehere didn’t look quite as dynamic with the puck as he had the previous year, but was very aggressive without it, particularly in his offensive zone pinches on the forecheck, and in his neutral zone defense. The result was improved play-driving metrics, despite extended time alongside Corsi deflator Andrew MacDonald. Yet the goal-based metrics didn’t follow suit, as Gostisbehere’s defensive aggressiveness seemed to burn him repeatedly, while he and his linemates couldn’t seem to buy a goal at the other end. The result was five scratches by the coaching staff, heralded as “teaching experiences” for the young defenseman.

But what was being taught? It’s a question that is impossible to answer without inside information regarding the content of those meetings. If the advice was “keep pushing the play and being aggressive on the attack, but work on your effort in defensive zone puck battles and recognize coverage switches quicker,” then it had the potential to be very helpful to Ghost, as those issues did exist, and were contributing to the high goals against totals. But if the advice was more along the lines of “cut down on o-zone pinches, and keep onrushing forwards to the outside rather than directly challenging them,” then we’re looking at an overreaction driven by unsustainable percentages.

Elliotte Friedman implied that in early February, the Flyers and Gostisbehere were butting heads on just how much the defenseman needed to change his style. Now, I have no idea what was being said in internal meetings between Gostisbehere and the coaching staff. But if Ghost was making the point that his play in the first half was generally fine and that the bounces would come eventually, the hard statistical truth is that he was probably on the right side of the argument.

That’s why, in my mind, Gostisbehere’s second half was in many ways more concerning to me than his first. The underlying processes were strong in 2016; Ghost just needed to clean up a few issues and get anything remotely resembling decent puck luck, and he’d be fine.

Photo by Bruce Bennett/Getty Images

This is where I believe the Flyers went wrong in their treatment of Gostisbehere in 2016-17. The scratchings didn’t help his on-ice performance — if anything, they sent his underlying play into an utter tailspin, at least for a bit. Gostisbehere talked about a loss of confidence as a big reason for his struggles in his exit interview, and I tend to think that an easy way to lose that confidence is to believe your game is structurally fine and then hear from your bosses that actually, you’re performing so poorly that Nick Schultz and Andrew MacDonald deserve to be in the lineup over you.

By the end of the season, Gostisbehere did look mostly recovered from a confidence standpoint. But the underlying numbers never returned to those early season levels, which calls into question whether he did make some subtle changes to his game to pacify the coaches, serving to weaken his play-driving ability.

Now, it remains totally plausible that the Flyers organization is committed to letting “Ghost be Ghost” and that this season was just a bump in the road of an otherwise-idyllic development process for him. Maybe the poor second half advanced metrics were primarily due to a “Nick Schultz effect.” But the trajectory of his stats does leave open a possibility I had previously discounted — that the treatment of Gostisbehere this season may have legitimately hurt the player for the long-term.

Let’s be clear — the Gostisbehere from 2015-16 isn’t coming back. That’s not saying that he can’t be a dynamic, effective defenseman at the NHL level. It’s merely an admission that he can’t expected to rack up tons of points with just-decent underlying metrics due to a shooting percentage over 10 percent. For me, the question is whether Ghost is content to be the guy from the second half of this year, or would prefer to be more like first half Ghost with better on-ice percentages.

The latter is the path of least resistance. He’ll still make tons of eye-popping plays with the puck, be great on the power play, and easily provide second pair value to the Flyers when looking at his overall contributions. But replicating the early-season play-driving ability while cutting down on the glaring defensive zone mistakes is the highest upside move for Gostisbehere. Going heavy on the offensive zone pinches and neutral zone aggressiveness may make his coaches cringe, but it’s also what allows him to have first-pair potential at even strength in the first place.

Is that guy gone for good, eviscerated in a blaze of awful puck luck and ill-advised scratches? I certainly hope not.

---

All stats courtesy of Corsica.Hockey, Stats.HockeyAnalysis.com, or the manual tracking work of Corey Sznajder at The Energy Line. On-Ice data derived from Corey’s manually-tracked metrics courtesy of Muneeb Alam.