Andrew MacDonald has been a lightning rod in the comments discussion of late. He has godawful possession #s, not justified by his zone starts, but faces top competition. Now that top competition has done better against MacDonald both this season and last than they have against other D men.
Certain people on this site have taken that to mean that AMac just isn't capable of being a top pair D Man, not that he can't handle lesser competition. Perhaps he can handle lesser competition while the top competition severely outmatches him!
So, to answer this question, I've decided to try and look at how AMac has fared this year and last year against competition of various levels. To do this, I went to stats.hockeyanalysis.com and to AMac's page. They include on the bottom AMac's performance vs forwards of opponents and how those opponents do otherwise. This allowed me to essentially separate AMac's performance into performance against players of various levels of quality. How well would AMac do against the Poor, Mediocre and merely good players instead of the elite?*
*One issue with this level of analysis is that I'm not measuring the difficulty of facing various lines, but the individual players. So if an opponent mixed his lines randomly for a game AMac played, AMac might face one elite player + 4 scrubs - and this analysis will treat this as AMac vs an elite player and vs two scrub forwards, even though it's not quite either of those levels of competition. In reality, this should be a tiny effect since it's unlikely most teams will be showing up with scrambled lines.
For those who are new to Corsi, it essentially is just a measurement of shot attempt differential - it counts here the percentage of shot attempts (including shots on goal, missed shots, and blocked shots) that are directed at the opponents' net while AMac is on the ice. We've found it to be an incredibly good proxy for % of time in the offensive zone, or as we call it for short: "Possession." Above 50% is good, since that means you're spending more time in the offensive zone than in the defensive zone.
I divided opponents up into four categories:
Poor Opponents (sub 47% Corsi players):
Mediocre Opponents (Corsi between 47% and 49.5%)
Good Opponents (Corsi between 49.5% and 53%)
Elite Opponents (Corsi above 53%)
Here's last year:
Amac vs Poor Opponents: Average Corsi of these opponents: 44% (Terrible). Actual vs AMac: 47%
Amac vs Mediocre Opponents: Average Corsi of these opponents: 48.5% (Meh): Actual vs AMac: 53.2%
AMac vs Good Opponents: Average Corsi of these opponents: 51.1% (Good): Actual vs AMac: 54%
AMac vs Elite Opponents: Average Corsi of these opponents: 56.5% (elite): Actual vs AMac: 58.2%
AMac indeed did fare poorly vs top competition - 58.2% Corsi is incredibly great play by those opponents and horrible play by the Isles. And good opponents similarly outperformed their average performances vs Amac - instead of just barely outperforming the Isles as they usually would have, they strongly won the possession battle vs AMac.
And here's the problem: even BAD opponents are outperforming their averages against Amac! Mediocre players, who usually lose the possession battle, performed nearly as well as the good opponents and clearly won the battle to get into our own zone. And the Crappy opponents - who were dominated on average last year, still outplayed their average (and thus were only bad not godawful) vs Andrew MacDonald.
AMac played only about 41% of his time vs Poor or Mediocre opponents. Unfortunately, he didn't play well vs these guys either, which is why his #s look so poor.
Here's this year:
Amac vs Poor Opponents: Average Corsi of these opponents: 43.5% (Terrible). Actual vs AMac: 54.5%
Amac vs Mediocre Opponents: Average Corsi of these opponents: 48.3% (Meh): Actual vs AMac: 59.0%
AMac vs Good Opponents: Average Corsi of these opponents: 51.2% (Good): Actual vs AMac: 59.7%
AMac vs Elite Opponents: Average Corsi of these opponents: 56.7% (elite): Actual vs AMac: 59.9%
This is even worse - Even bad opponents are dominating the ice with AMac on the ice! Presumably the play of poor and mediocre players should get not so dominant (these are bad players after all) but still - again this poses strong evidence that AMac's performance isn't due to not being a first pair defender as much as AMac kind of hasn't been a good defender at all this year.
------
Another Methodology:
Olivier Bouchard (@Oli_Bou) is a Habs analyst and tracker of Habs games. I mentioned this work on twitter, and he revealed to me he had his own methodology which does a similar thing. What he does is look at performance of a player when the opponent has various #s of top 6 forwards or top 3 D on the ice. So minutes vs 0-1 top 9 players are easy, vs 2-3 are decent difficulty, vs 4+ are very hard.
For AMac this year, he had the following #s:
Against 0-1 top6F: 42% zone starts, 44% Corsi
Against 2+: 50% ZS, 43% Corsi.
In other words, AMac is facing weaker competition only when they're starting in our own zone this year, but he's being strongly outplayed by these tough opponents (no 42% zone starts does not come close to justifying 44% corsi - the adjustment is at most 1-2% for that zone-start level.) And against top competition, AMac is facing even zone starts but getting purely dominated. Basically what we see above.
Conclusion:
The problem of course here is that AMac plays so infrequently with anyone other than Travis Hamonic, it's hard to be sure he's the problem and not Hamonic - the sample sizes with other pairs are too small. But that pairing has been godawful vs all sorts of competition.