Red Wings Roster: Glendening versus Andersson

As we wait for training camp to start and the battles for positioning which will ensue, there's one battle which seems to have already been decided at the bottom of the roster that I think has gone the wrong way. I'm not sure how much it's going to matter by the time the season starts, but I think that valuing Luke Glendening over Joakim Andersson is a mistake.

The fans have treated it as a foregone conclusion since the end of the Wings' playoff run. Part of that is because there are people who legitimately think Luke Glendening brings more to the ice than Andersson does and part of it is because the fans know how much Mike Babcock adores Gledening for his "compete level", which he compares to NHL superstars like Sidney Crosby, Patrice Bergeron, and Pavel Datsyuk. With Glendening, it's all about the intangibles.

This is good for him, because tangibly, Joakim Andersson is a better hockey player.

Starting off with the offensive side of the game, Joakim Andersson can best be described as a stone-handed disappointment. He was never much of a point-producing prospect, putting up approximately one point every four games. However, he's practically Datsyukian compared to Glendening's one-in-eight pace. Even at the AHL level, where there was more an expectation to score on both of them (and where Andersson benefited greatly from being the center for Gustav Nyquist and Tomas Tatar), Glendening has produced only half as well as Andersson has.

I know, I know. Their jobs are mostly defensive. I get that. That doesn't mean that offensive production is meaningless. Fans love to criticize players who only produce points while being a defensive liability. Why should one-way players on the other side of the puck get more leeway?

Besides, provided we're looking defensively, Glendening isn't better at that either. Judging by shorthanded play, Andersson was on for a lower goals against rate and for a smaller Corsi against rate than Glendening. The even strength possession difference was stark. At 5-on-5 with the score close, Andersson saw 52.8% of the Corsi events go the Red Wings' way while Glendening lorded over a team-worst 47.3% among forwards who got into at least 40 games. Looking at usage, Glendening's slightly tougher competition and Andersson's tougher zone starts don't adequately cover for these differences.

Despite possession numbers telling a story that Andersson brings more value to the Red Wings, there are some positives that Glendening has over his competition which helps feed into the reasons why he's valued higher. For one, Glendening is a faster skater and three inches shorter. This isn't a backhanded shot at him. Glendening's speed gives him the ability to cover more ice, even with his reach disadvantage. His smaller size (and lower center of gravity) lets him recover from hits a little faster as well. Also, I don't want to discount that it simply makes Glendening look better. Andersson has the gait of a drunken fawn and he hits people with all the grace of Stretch Armstrong. It's honestly more fun to watch Glendening running around hitting people.

Another advantage Glendening held over Andersson last season was luck. Neither player was particularly lucky as far as the traditional measure (PDO). Compared to Andersson though, Glendening is practically a lottery-winner.

Let's take a look at the way player luck factored for forwards (min. 40 games)

Player TOI SF SA SA/Min Normalized GA (sv% regressed 70%) SV% SV% Goals
Gustav Nyquist 730 393 334 0.46 450.21 32.91 94.30% 25.66
Tomas Tatar 839.3 440 359 0.43 420.89 32.66 92.80% 30.30
Daniel Alfredsson 864.9 398 392 0.45 445.98 34.47 92.90% 31.66
Henrik Zetterberg 684.7 390 332 0.48 477.13 36.17 93.40% 31.49
Riley Sheahan 485.4 261 222 0.46 450.04 35.60 92.30% 34.65
Justin Abdelkader 943.9 448 459 0.49 478.50 37.13 92.80% 34.45
Johan Franzen 720 366 354 0.49 483.80 37.40 92.90% 34.35
Darren Helm 495.6 237 228 0.46 452.69 36.62 91.70% 37.57
Drew Miller 885.7 406 396 0.45 439.95 36.25 91.20% 38.72
Luke Glendening 604.9 262 300 0.50 488.01 38.02 92.70% 35.63
Pavel Datsyuk 679 353 301 0.44 436.21 37.91 89.70% 44.93
Joakim Andersson 696.7 332 316 0.45 446.31 38.52 89.90% 45.08
Daniel Cleary 551.1 240 288 0.52 514.23 41.14 92.00% 41.14
Todd Bertuzzi 676.8 294 332 0.49 482.70 43.10 88.90% 53.58

What we're looking at here is a normalized chart showing the effects of variance on a player's output. The TOI and shots for/against show the true figures for each player on ice at 5-on-5 this season. What I've done is normalized the rates of shots against for 12 minutes per game over an 82 game season to show the expected rates for goals against. This pretends that every player had even ice time and every factor which led to their shots for/against rates stayed steady (again, Andersson played a bit more defensive minutes while Glendening played slightly tougher competition... like 3/4th line instead of simply 4th line competition). This list is sorted by the amount of expected shots against multiplied by the opponents' actual shooting percentage while that player was on the ice to get the goals that player would have seen against under normalized ice time. Here you can see what a ghastly difference that a near 3% change in save percentage can make over how many goals a player will give up. Andersson saw the puck get behind his goalie much more often and that hurts severely.

So far, nobody has been able to effectively measure how much difference in save percentage a player can drive. If Glendening is good at making his goalies save almost 93% of shots, what makes Andersson bad at it? Is Pavel Datsyuk the absolute worst forward on the team in terms of preventing shot quality? I don't think so.

To at least make an attempt at figuring luck here, we'll go to the SV% Regressed column for expected goals. What this does is take the player's on-ice save percentage and get it 70% closer to the team average of 92%. Essentially, we're pretending for argument's sake that luck drives 70% of a player's effect on save percentage. Considering none of the people smarter than I am with advanced stats have proven that a player has ANY repeatable effect on save percentage, I'm throwing Glendening a decently-sized bone in saying it's not all luck here. This lowers his advantage over Andersson in assumed goals against down to 0.5 goals over an 82-game season (If you regress it 100%, Andersson is better than Glendening by 2.34 goals by virtue of how many fewer shots the opposition gets with him on the ice).

So now that we've got the defensive side, let's look at the offensive differences.

Player TOI SF SA SF/Min Normalized GF (sh% regressed 70%) SH% SH% Goals
Henrik Zetterberg 684.7 390 332 0.57 560.48 50.16 10.70% 59.97
Riley Sheahan 485.4 261 222 0.54 529.10 45.77 9.70% 51.32
Pavel Datsyuk 679 353 301 0.52 511.56 43.48 9.20% 47.06
Tomas Tatar 839.3 440 359 0.52 515.86 43.07 8.70% 44.88
Justin Abdelkader 943.9 448 459 0.47 467.03 39.84 9.30% 43.43
Johan Franzen 720 366 354 0.51 500.20 41.17 8.30% 41.52
Gustav Nyquist 730 393 334 0.54 529.74 42.80 7.80% 41.32
Todd Bertuzzi 676.8 294 332 0.43 427.45 36.20 9.10% 38.90
Darren Helm 495.6 237 228 0.48 470.56 38.44 8.10% 38.12
Daniel Alfredsson 864.9 398 392 0.46 452.81 35.36 6.90% 31.24
Drew Miller 885.7 406 396 0.46 451.06 34.69 6.50% 29.32
Joakim Andersson 696.7 332 316 0.48 468.91 34.37 5.30% 24.85
Luke Glendening 604.9 262 300 0.43 426.20 31.50 5.50% 23.44
Daniel Cleary 551.1 240 288 0.44 428.52 29.35 3.70% 15.86

Here, we don't even have to regress to the team's shooting percentage (8.2% at 5-on-5) to get an advantage for Andersson. This is even when factoring in that the team shot worse with Joker on the ice than they did with Glendening. If you DO end up choosing to say that 70% of a player's effect on his team's shooting percentage is luck, Andersson's advantage approaches three goals per season.

So let's look at the differences now.

Player TOI SF SA Effective G Differential Expected (regressed) G Differential
Henrik Zetterberg 684.7 390 332 28.48 14.00
Tomas Tatar 839.3 440 359 14.58 10.41
Riley Sheahan 485.4 261 222 16.67 10.17
Gustav Nyquist 730 393 334 15.66 9.89
Pavel Datsyuk 679 353 301 2.13 5.58
Johan Franzen 720 366 354 7.17 3.77
Justin Abdelkader 943.9 448 459 8.98 2.71
Darren Helm 495.6 237 228 0.54 1.82
Daniel Alfredsson 864.9 398 392 -0.42 0.89
Drew Miller 885.7 406 396 -9.40 -1.57
Joakim Andersson 696.7 332 316 -20.23 -4.15
Luke Glendening 604.9 262 300 -12.18 -6.52
Todd Bertuzzi 676.8 294 332 -14.68 -6.90
Daniel Cleary 551.1 240 288 -25.28 -11.78

The Effective Goal Differential shows what normalizing the shots, but keeping the same shooting and save percentages would do, while the expected regressed differential corrects for luck. We can see that normalizing playing time across all forwards and holding all other factors constant, Glendening would have been 8 goals better than Andersson last season in terms of effectiveness on even strength. If we correct for luck however, the advantage disappears completely, instead tilting to Andersson's side (we can also see how brutally awful Dan Cleary was last season, but that's not the point of this comparison).

Last Defense

There is one more argument about Glendening left. This says that Glendening can handle and even frustrate a team's best players, allowing his teammates to get better matchups. Glendening is essentially a sacrificial lamb. Looking at the relationship between Glendening's Corsi, CorsiRel, and ice time, the Wings has a 51.4% Corsi Rating at 5-on-5 in games featuring Luke Glendening. In games not featuring Luke Glendening, the Red Wings' Corsi Rating was... (drumroll)...51.4%

If he makes the team better, it's not showing in a tangible manner.


It's not difficult to see why fans prefer Luke Glendening to Joakim Andersson. His play looks better and the team got better results with him on the ice. The coach raves about him and, let's face it: Luke Glendening looks like Tom Hardy while Joakim Andersson looks like a guy who went to my high school who everybody called "Fart Knocker."

Objectively, the Red Wings are probably better with neither player on the ice regularly, but when you dig into the underlying figures and correct for the things which a player can actually control, Joakim Andersson is a better choice.

[Thanks to for much of the data in this post]