2016-17 Flyers season review: Andrew MacDonald saw a larger role, but results remained poor

The career of Andrew MacDonald in Philadelphia will always be boiled down four simple words in the minds of fans: six years, thirty million.

That of course, was the contract that MacDonald signed way back on April 15, 2014, while playing under another Flyers general manager and for another Flyers coach.

But long-term extensions don’t just disappear when those responsible no longer have personnel power. MacDonald played just one season under the massive contract before he and his five million dollar cap hit found themselves in a battle for a roster spot at training camp in 2015, a battle that would end in defeat. Yes, the player envisioned by Paul Holmgren as a key member of the Flyers’ blueline through the end of decade just a year and change earlier would ply his trade in Lehigh Valley while defensemen like Luke Schenn and Brandon Manning received regular NHL minutes.

To be sure, the demotion to begin the 2015-16 season wasn’t solely the result of a poor camp, or struggles the previous year. Unlike his competition for a bottom-pair role, MacDonald’s remaining four years and $20 million on his contract made a waiver claim by another club highly unlikely, allowing for general manager Ron Hextall to essentially stash him in the AHL in the event of injuries to the big club, and clear necessary cap space simultaneously. Still, a player signed to a contract befitting a high-end second pair defenseman shouldn’t even be an option for a demotion — he should be viewed as an essential part of the team. At that moment in time, Andrew MacDonald clearly was not seen in that light.

After spending the majority of the year with the Phantoms, MacDonald finally received a recall in February of 2016 after Michael Del Zotto suffered a season-ending wrist injury. Surprisingly, he immediately joined a pair with hotshot rookie Shayne Gostisbehere, and received second-pair minutes right off the bat. While it was obvious that MacDonald had earned a degree of trust from NHL head coach Dave Hakstol, the fact that he played 43 games in the minors in 2015-16 loomed large over his status in training camp come September 2016. After all, Ivan Provorov was pushing for a roster spot, and with Del Zotto back, there was no guarantee that MacDonald wouldn’t return to Lehigh Valley, again the victim of a roster crunch.

Instead, it was the conclusion of the 2015-16 season that best predicted Andrew MacDonald’s 2016-17 usage. Not only did he never once see the minors, the 30-year old defenseman ended up receiving top-pair minutes for the majority of the season, right alongside the rookie Provorov. It was a complete reversal of fortune for MacDonald and his career, as he had clearly earned his way back into the good graces of the organization and was being given minutes worthy of that six year, $30 million contract.

There was only one thing keeping MacDonald’s return to prominence from being a true feel-good story: his statistical results.

MacDonald’s usage ascension

Of course, the Flyers’ season did not begin with Andrew MacDonald jumping straight into a heavy-minutes, first-pairing role. That required a number of events in succession to allow for MacDonald to climb the depth chart and take on a prominent role in terms of minutes. However, once Dave Hakstol gave MacDonald that spot in the lineup, it became his permanent role.

To start the year, MacDonald was paired with Shayne Gostisbehere, a reunion of the duo that Hakstol seemed to like so much during the 2015-16 stretch run and into the playoffs. But MacDonald was clearly viewed as the lesser of the pair at even strength. While Ghost often received extra shifts with Brandon Manning and Mark Streit, MacDonald usually stayed on the bench when not skating with Gostisbehere. He was receiving lots of minutes on the penalty kill, but at 5v5, his 14.47 minutes per game ranked sixth among Philadelphia defensemen in October. Essentially, he was a PK specialist/third-pair defenseman, an understandable role for a guy who spent two-thirds of the previous season in the AHL.

To hammer home the point that he was viewed as strictly a role player at this stage of the season, Hakstol scratched MacDonald for eight games during an 11-game stretch, starting on October 29th versus the Pittsburgh Penguins and concluding on November 19th against the Lightning. At this point, it would have been impossible to predict the substantial increase in ice time that Andrew MacDonald would soon receive.

But three things happened around the same time that MacDonald rejoined the lineup in late November. First, Shayne Gostisbehere’s minutes began to decline. Ghost was a surprise scratch on November 17th, and even after he returned, the days of Gostisbehere receiving first-pair minutes (as he did to start the season) never really returned. Second, Radko Gudas’ minutes dropped as well, which is a bit more difficult to understand considering his strong play all season long. However, there were rumblings around the team in November and December that Gudas was battling through injury (he was eventually scratched twice in late December/January), and that could help to explain why Gudas was not chosen to pick up the even strength slack with Ghost’s minutes dropping.

The biggest adjustment to the defense, however, was the increase in role for Ivan Provorov. Around Game 20 (mid-November) Provorov took the lead in overall ice time among Flyers defensemen and never relinquished the honor, a shocking development for a 19-year old rookie. But Provorov’s steady play made the decision understandable. It also opened up the potential for a corresponding increase in role for whoever played alongside Provorov, which is where MacDonald stepped in.

Ghost wasn’t getting the job; Hakstol’s trust in him was waning. Gudas’ minutes were dropping too. Mark Streit was Provorov’s main partner to start the year, but that duo saw awful goal-based results and Streit was about to turn 39, so his ability to take heavy minutes was in question. That left two choices — MacDonald or Michael Del Zotto, and the latter had struggled after his return from a preseason injury. Obviously, Hakstol chose MacDonald, and the rest was history.

Aside from a brief period around the season’s midpoint when he dropped below Del Zotto, Andrew MacDonald was clearly the Flyers’ #2 defenseman for the remainder of the season. From December on, he averaged over 16 minutes per game at 5v5 and spent 55 games as the primary partner of Ivan Provorov. Combine the increased even strength role with the fact that only Provorov received more shorthanded minutes for the Flyers in 2016-17, and you end up with an undeniable truth — Philadelphia viewed Andrew MacDonald as one of their most important defensemen this past season.

One problem — his stats remained awful

Just because a coach is giving a player a prominent spot in the lineup and the minutes that accompany that role, it does not mean that the player should be in that role in the first place. That was the big issue with the Flyers’ usage of Andrew MacDonald — it’s difficult to find a statistic that looks remotely favorable upon his on-ice performance during the 2016-17 season.

This really shouldn’t come as a surprise, particularly to longtime readers of BSH. Before the Flyers traded for MacDonald, we noted that his play-driving metrics were flat out awful. Following the trade, we dug deeper into video, identifying his neutral zone defense as a primary reason for his poor underlying metrics. We disagreed with the contract. We argued last year that the belief that MacDonald’s play improved dramatically upon returning to the Flyers after his sojourn to the AHL was a myth and not actually backed up by statistics. I even made the case back in September that MacDonald has been one of the worst NHL defensemen in term of relative 5v5 play-driving metrics since 2013-14.

Basically, Andrew MacDonald has a long track record of failing to produce positive outcomes while on the ice, especially at even strength. A good season was theoretically possible, but seemed unlikely based on the preponderance of evidence regarding MacDonald’s true talent level. That’s why it was so odd to see MacDonald elevated to a top pair role by Hakstol. He had received second pair usage in 2014-15, and apparently looked poor enough in those minutes to put him on the roster bubble entering the following camp. If he had struggled in that role the last time that he received a full year in the NHL, why would he fare better when tasked with more minutes and tougher matchups?

Spoiler alert: he didn’t.

Among regular Flyers defensemen, Andrew MacDonald ranked seventh in score-adjusted Corsi For percentage. And for those who argue that Corsi doesn’t treat MacDonald fairly because he blocks a high amount of shots, MacDonald finished dead last on the defense in Fenwick percentage, which removes blocks from the equation.

Others have argued that play-driving metrics miss something when it comes to evaluating his value, and that plus/minus is the better way to go. But in 2016-17, MacDonald actually underperformed his Corsi and Fenwick in terms of Goals For percentage, posting a 42.68% rate that ranked him sixth on the Flyers defense.

Most concerning, however, was his impact on talented young defensemen Provorov and Gostisbehere. MacDonald received almost 88% of his 5v5 minutes this season paired with one of the two young pieces, and his presence on their pairing appears to have negatively impacted their play-driving ability, in a dramatic way.

In their minutes away from Andrew MacDonald, both Provorov and Gostisbehere drove play to strong degrees. But with MacDonald, they ranged from passable (Gostisbehere) to flat out bad (Provorov). And it’s not like the Goals For percentages were much better — Provorov had a 44.9 GF% with MacDonald, while Gostisbehere was at 37.5%. The Flyers’ two most important blueliners were clearly, measurably worse when they played alongside MacDonald at 5v5.

The issues don’t end there. MacDonald had a negative penalty differential at 5v5 (took 10, drew five). His on-ice shot and goal prevention metrics on the penalty kill were the worst on the team among PKers with at least 100 minutes at 4v5 this season. His Failed Defensive Zone Exit rate at 5v5 was third-highest on the defense at 27.08 percent.

Looking for bright spots in Andrew MacDonald’s statistical resume from 2016-17 is like trying to find a good cheesesteak outside of the Philly area: they exist, but you have to scour the globe and even when you do find a good one, you can’t shake the feeling that you’re subconsciously grading it on a curve.

His 5v5 scoring metrics (0.75 Points/60) were solid, ranking him 90th among defensemen with at least 300 minutes. However, as he ranked 7th among Flyers blueliners in Primary Shot Contributions/60 (ahead of only Nick Schultz), it’s not as if he was directly driving a ton of offense, making those decent scoring metrics look a bit fluky.

His on-ice Goals Against per 60 of 2.15 is perfectly solid as well, and ranked second-lowest on the Philadelphia defense. That seems impressive, until you remember that goal differential is far more important than straight goal prevention, and the Flyers averaged just 1.65 Goals For per 60 with MacDonald on the ice, easily putting his overall Goals For percentage in negative range.

Finally, MacDonald did perform better in weighted shot metrics (Expected Goals) versus raw ones (Corsi and Fenwick). His 48.30% score-adjusted xG and -1.10% xG Relative easily top his 47.50% score-adjusted Corsi and -3.89% Corsi Rel. But it’s not as if his results in xG are actually good — they’re just not horrific like his performance in terms of Corsi. They’re more in the realm of “passable third pair guy” rather than “press box seat warmer,” which is still far from the usage that MacDonald actually received.

This simply is not the resume of an above-average NHL defenseman.

What could explain the usage of MacDonald?

Let’s be clear — while I’m in total disagreement regarding the Flyers’ current usage of Andrew MacDonald, this does not mean that I believe the organization is “stupid” or “anti-stats.” Ron Hextall has employed intelligent analysts in his organization since being hired as Flyers GM, and I am certain that he is open to the metrics that we discuss on BSH every day. As for Dave Hakstol, in my conversations with him after games and at practices, it’s obvious that he knows his hockey and has even let it slip on occasion that he looks at team-level shot metrics at the very least.

But the gap in the valuation of MacDonald by public stats and by the current Flyers organization is enormous. The numbers imply that he’s a barely-passable third pair NHL defenseman, while Philadelphia sent on the ice him for 20 minutes each night, usually against top lines and top power play units. What is causing this discrepancy? And just as important, what changed in the team’s evaluation of MacDonald from 2015 (when he was sent to the AHL) and 2016 (when he was viewed as an indispensable member of the defense)?

One possible explanation is that the coaching staff was essentially fooled by an unsustainably high PDO that MacDonald posted after returning from his final healthy scratch of November. Over his next 13 games — which included the winning streak — MacDonald posted an 5v5 on-ice Goals For percentage of 68.94%, an obviously-stellar rate. But it was a hollow number, as the Flyers controlled just 47.44% of the shot attempts during that span. His performance was being propped up by a whopping 105.28 PDO, especially helped by the fact that Flyers goalies stopped 97.15% of the shots on goal while MacDonald was the ice during that period.

Remember, this was the first time that Provorov and MacDonald were paired together, and they put together one heck of a first impression, functioning as a positive goal differential-driving top pair during the 10-game winning streak. It’s not hard to imagine Hakstol getting a great early read on the viability of the duo, and locking into it as a necessity even after the goal-based metrics of the pair tanked, chasing results that were never going to recur.

Explanation No. 2 is that the coaching staff and organization are primarily using weighted metrics to judge MacDonald’s play. I noted back in September that MacDonald’s xG rates graded out nearly as poor as his Corsi percentages. But much of that is driven by a frightful -6.60% Expected Goals Relative that he posted back in 2013-14. In his first three full seasons as a Flyer, however, MacDonald grades out passably.

It’s not like these metrics are especially impressive, or even qualify as good. But they’re not terrible, and a far cry from his Corsi metrics, which are among the league’s worst. It’s easy to imagine Hextall and Hakstol looking at these numbers (or similar, private metrics) and thinking that MacDonald is a perfectly fine third pair defenseman on a good team.

And this is where we get into a key front in the “old school vs. new school” debate, specifically the question of how to best evaluate defensive talent. The new school tends to value on-ice shot differential (since it predicts future goal differential, which wins games) and as a result, shot volume creation and suppression to judge a blueliner’s effectiveness. The old school often leans towards goal and chance prevention.

When MacDonald has been on the ice at 5v5 over the past three years, Philadelphia has given up a lot of shots, and they haven’t created much, either. However, the quality of the shots they have allowed has been fairly low, at least relative to MacDonald’s teammates. Let’s take a look at the average xG value of each shot attempt taken by an opponent while Flyers defensemen have been on the ice since 2014-15. That should help us quantify which defenseman’s “shot quality against” has been lowest.

There’s Andrew MacDonald, right there in third place. There appears to be some statistical basis to the theory long pushed by MacDonald defenders that he keeps shots to the outside and suppresses shot quality. Hockey lifers really might be accurately picking up on an ability to limit dangerous chances, and that very well could play into why people around the game seem to like Andrew MacDonald so much.

But even if MacDonald can suppress shot quality (a questionable assertion, because all this shows is that he’s done it in the past, not that he will in the future), that still leaves one major issue with his game: namely, the fact that his conservative style results in lots of time spent in the defensive zone, and too little time spent on the attack.

Over time, heavy raw volume (even at slightly decreased quality) will result in goals against, and all that time spent chasing the puck rather than moving it up ice and actually creating offense makes it very difficult to finish with a positive goal differential. Last year, for example, the Flyers averaged 52.95 shot attempts per 60 at 5v5 with MacDonald on the ice, versus 61.71 with Radko Gudas and 60.78 with Gostisbehere. The discrepancy in offensive production between MacDonald and his peers is dramatic.

That’s why MacDonald’s on-ice xG percentage over the past three years remains under 50 percent, even with the boost for solid defensive zone chance prevention. At some point, the team needs to score goals to win games. And that’s also why it’s a questionable strategy to continue to place MacDonald alongside the dynamic young defensemen who seemingly are most qualified from a skillset standpoint to help the Flyers score goals. Is the “low shot volume for, high shot volume but low shot quality against” style really the ideal for puck-movers like Ghost and Provorov?

Still, in a year that saw the Flyers goaltenders struggle through the vast majority of the season, maybe the calculation was that MacDonald has an ability to suppress chances, and that’s what the team needed most on its top pair. It’s not a strategy I’d endorse, but I could at least understand the thought process.

What comes next for MacDonald

There remains a lingering belief among those who feel Andrew MacDonald receives too much criticism from fans that the bulk of the anger is a result of his massive contract, and not especially poor on-ice play. I’ve always disagreed with that stance. It’s obviously true that MacDonald will never live up to the $30 million that he received, but other Flyers players have counted as overpaid (Matt Read and Mark Streit being obvious examples) yet never received the same vitriol that MacDonald has.

The anger, I believe, is mostly a testament to MacDonald’s awful performance by advanced metrics. Over the course of his NHL career, he has failed to drive play at 5v5, or establish himself as a dangerous offensive weapon from the backend. Players like Read (he can drive play) and Streit (he can still score) at least brought some obvious positive attributes to the table. MacDonald, on the other hand, has failed to stand out in any way.

He’s not an especially strong puck-mover, as his Controlled Defensive Zone Exit rates split the difference between Brandon Manning and Radko Gudas. He’s unspectacular in the offensive zone. He’s notoriously passive in his neutral zone defense. MacDonald’s biggest problem is that he’s not above-average at any one thing, and also measurably poor at quite a few. The end result is a player who consistently causes his team to lose the raw territorial battle at 5v5.

Yet this season, MacDonald was used like a borderline top pair defenseman. Maybe it was a good string of luck in November and December that turned the organization around on him. Maybe it was his track record of delivering decent chance suppression results in recent years. Or maybe the team just has a soft spot for MacDonald, impressed by his positive attitude after being sent down to the AHL, and still unwilling to fully admit that members of the organization — some of which remain strongly affiliated with the club — misevaluated him back in 2014.

If MacDonald was spending the bulk of his time on the third pairing in a 2016-17 Nick Schultz-type role, he would not be the object of so much analysis. But in two straight years, the Flyers have essentially stapled MacDonald to their most promising young defensive prospect, and appeared happy with the results despite the young gun’s metrics falling off a cliff in the aftermath. This simply cannot continue.

I doubt that putting Gostisbehere and Provorov with MacDonald is severely hurting their respective development curves, but it is making it more difficult to quickly evaluate exactly what the Flyers have in these young players. Provorov sure looked like a future #1, but most advanced metrics told the tale of an underwhelming rookie season. Gostisbehere received similar skepticism in his first year from the mathematically-inclined due to the MacDonald effect. What if this becomes a trend moving forward, in that the Flyers believe that MacDonald is the perfect complement to the newest rookie blueliner to make it to the NHL? Do we really have to wait until Year 2 to know for sure if Robert Hagg is truly a poor play-driver? Travis Sanheim? Phil Myers?

At some point, major decisions will need to be made regarding the kids and their career paths, and I worry about a scenario in which a promising young defenseman falls down the organizational depth chart solely because he spent lots of time with a veteran that the coaching staff mistakenly believes is not a massive drag on all of his teammates.

If Andrew MacDonald is destined to be a Philadelphia Flyer for the remainder of his contract, the best move is to use him in a role befitting his past results. Unfortunately, in 2016-17, that did not happen, and the team suffered for it.

---

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.