Digging into Shea Weber’s poor Relative Corsi

I heard something on the Buffalo Sabres’ broadcast during a game between the Sabres and Washington Capitals that caught my ear. Buffalo’s ever-entertaining color analyst Rob Ray had a tidbit about the Caps’ head coach Barry Trotz and analytics.

The story went that Trotz had been approached by Nashville’s numbers guy during the previous season about cutting back Shea Weber’s ice time. Ray chuckled at the notion and mentioned Trotz did the same. Then there was something about the eye test.

Here’s the thing: The eye test is pretty definitive on Shea Weber. He’s awesome.

Weber is everything you dream a No. 1 defenseman will be. He’s huge, can skate, is intimidating and has a shot that nobody on earth wants to step in front of.

But his Corsi number stinks.

This season, when Weber is on the ice, the Predators are taking 50.1% of the total shot attempts. When he’s off the ice, 54.5%. With a brutal -4.4% Relative Corsi, it is easy to understand why a numbers person might raise his hand and ask if someone else should be taking the No. 1 minutes.

After hearing Ray’s anecdote, I decided to dig into his Corsi number.

Evaluating a player based on his Corsi rating is like trying to find your cell phone by having a friend call it. At first, you hear it, but still don’t see it. You don’t usually find the phone until you’ve thrown all the pillows off the couch.

With most players, we know which pillows our phone often falls underneath. We lift them up one at a time: What is his ice time? How often does he start in the offensive zone? What is his quality of competition? Who is he paired or on a line with?

By that time, you’ve usually found it. You can draw a decent conclusion about how a coach feels about his player by those four numbers. Say Player X plays 24 minutes per game, starts in the offensive zone 50% of the time, faces top competition and is paired with another good player. Hazaa, he’s a top pair defenseman.

Now – if his Corsi is good, we figure he’s a star, if it’s bad, he’s not a true top pair defenseman.

The formula generally points us toward the ringing phone.

However, we have to go a little farther to find out why Weber’s numbers don’t match up with his Corsi.

Here goes….

Weber plays a ridiculous amount of minutes with Paul Gaustad, who has a 10.9% Offensive Zone Start percentage. Consider that most fourth line “shutdown” forwards/defensive zone faceoff specialists are somewhere in the 40% range.

Most of these fourth line faceoff men also face mediocre competition at best. Gaustad, however, takes on the opponents’ top centermen pretty often. The only forwards (who aren’t Gaustad’s linemates) on the Predators with higher Quality of Competition (TOI) are James Neal, Filip Forsberg and Mike Ribeiro. And those fellas all start in the Offensive Zone at rates above 64%.

Guess who else faces the toughest competition? You guessed it: Shea Weber.

As you can see, it isn’t close. Weber and partner Roman Josi get the hardest assignments by far.

Weber 9

Quality of Competition shouldn’t be the thing that drops Weber’s Corsi. For most of the elite defenseman in the league, what makes them elite is their ability to continue to tilt the ice despite getting those difficult assignments.

But the difference is those Offensive Zone Starts and who Weber is taking them with: Gaustad.

Having to face top lines in the Dzone so often alongside a fourth liner has an effect on the Predators’ ace defenseman’s Corsi. In fact, when he plays with anyone else, his Corsi is terrific.

Weber 1


Weber 3

Surprise, Surprise, he gets way more offensive zone time with those other players, too.

Weber 2


Starting in the defensive zone as often as he does means it is more difficult to get pucks to the other end of the ice to create shots. What makes that also difficult is that Gaustad is not an offensively skilled player. If you take a defensive zone faceoff and get the puck to Sidney Crosby, he will likely carry it into the other end and get a shot. Weber’s centerman (and other fourth line mates as well) have the assignment to simply get the puck out of the defensive zone and allow the Preds to change. That has a huge effect on Corsi For Per 60 Minutes

Weber 4

Your next question should be: Is this any different from any of the other top defenseman in the NHL?

Looking at a few of the NHL’s top Corsi defenseman, and likely Norris Trophy candidates, Drew Doughty, Johnny Boychuk and Kris Letang, the answer is yes.

Weber 6


Weber 7


Weber 8

Those three don’t have anything even close to a Gaustad to drag their Corsi % into the depths of the Earth.

One thing that is worth mentioning is that Gaustad’s overall Corsi % is a shade higher without Weber than with. The explanation for that seems pretty simple: Weber is always out vs. top competition. If Gaustad is away from Weber, that probably means he’s facing off with a lower line, making it easier to suppress shots and get them.

Notice that before Gaustad’s arrival, Weber still was out for mostly defensive minutes, but had better possession stats.

Weber 10

In 2011-12, his best season, the defensive zone starts were spread over the four centers almost equally. Compare that to his usage this year with Gaustad

Weber 11

Corsi is a tricky bird sometimes. When something doesn’t make sense compared to a player’s reputation, it doesn’t always mean the coach/broadcaster is wrong. It might mean there are abnormal factors at play that are throwing off the numbers.

Is it a sure thing that the Dzone faceoffs with Gaustad are 100 percent of the answer to the Shea Weber Corsi Problem? No. But considering Weber’s past success with Corsi, it seems likely that it isn’t a Weber problem, it’s a usage problem.

Ring, ring.


Photo via Wiki Commons



12 thoughts on “Digging into Shea Weber’s poor Relative Corsi

  1. Hey Matthew, if Josi is Weber’s most comment partner, what do Josi’s WOWYs look like? Based on this article I’d assume they are similar based on similar usage.

  2. Good work though I would be careful attributing this to zone starts as there is very little evidence that zone starts have a major impact on a players statistics. Maybe 1-2% change in CF% at most.

    My belief is what you are seeing is more due to a playing style choice, and more specifically the choice not to push for offensive chances. The reason I suggest this is because the primary reason Weber w/ Gaustad has a bad Corsi is due to a major drop in CF/60 (as you pointed out) and only a minor increase in CA/60.

    If zone starts were a primary reason for change in CF% one would think that the impact to CA/60 and CF/60 would be roughly equal (but opposite). By that I mean the opportunity for one team to generate more shot attempts due to offensive zone starts should be roughly equal to the loss of opportunity for the other team to generate more shot attempts due to fewer offensive zone starts.

    This doesn’t appear to be the case with Gaustad or many other “defense first” players. Jay McClement had a terrible Corsi for the Leafs, but if you just looked at his CA60 it was reasonably good (for the Leafs anyway). His CF60 was terrible though. Same for Mike Komisarek during his time with the Leafs and you’ll find the same with many players who primarily play a defensive role.

    Now Gaustad’s (and his line mates) relative lack of offensive skill may be a factor in the decreased CF60 as well but I believe when your primary objective is to not give up a goal there may be a cognitive choice to maintain control of the puck rather than taking a shot (which often leads to your opponent gaining puck control). This, to an extent, also happens with score effects too. Also, if you read my posts on how coaching changes have resulted in significant changes to various players CF% you’ll get an idea how significantly a player can influence their stats solely through coaching/playing style changes.

    I believe the major reason for Weber’s poor CF% when playing with Gaustad is not the zone starts but that when Gaustad is on the ice the major objective of the Predators is to not give up a goal while scoring a goal is given relatively low priority.

    • Hey David.

      Thanks for reading the article. The point you make is true and one I touched on a little in the article. Something worth digging into the video would be to see how they play after a Gaustad with with Weber on the ice.

      I do find it hard to believe a 13% offensive zone start percentage wouldn’t have an effect on Corsi. Having almost exclusive faceoffs in the defensive zone is some thing that I can’t remember seeing before. Common sense would say that constantly going against first or second lines under those circumstances would drag down Corsi with teams getting shots off faceoffs etc that much Vs. Gaustad line never getting those attempts.

      Appreciate the feedback! It’s something worth digging more. Just wanted to get the ball rolling

      • Zone starts do have an effect, just not all that much, or as much as one might think. I do have ZS adjusted WOWY available. My zone start adjusted data ignore the 10 seconds after n offensive/defensive face off which is where the majority of the benefit takes place. Let’s take the past 2 seasons as an example to increase the sample size.

        Gaustad with Weber 5v5: 43.0 CF%, 18.3 OZone%
        Gaustad with Weber ignoring 10s after a zone face off: 45.1 CF%

        18.3% is an extreme ZS bias but it only impacted their CF% by -2.1%. This season the defensive zone start bias was more significant and resulted in a -2.9% impact (43.6% vs 40.7%). This is about as big an impact as you will ever see.

        Weber with Ribiero is what you get when Weber is playing an offensive role (56.1 CF% this season with 70.88 CF60, 55.53 CA60). Weber with Gaustad is what you get when Weber is in a defensive role (40.7 CF%, 42.47 CF60, 61.98 CA60).

        Have a look at Boyd Gordon of the Oilers. His zone start usage is pretty similar but his overall CF% is 42.4% but when ignoring the 10 seconds after a zone face off it only jumps to 43.6% for just a 1.2% difference. He too has a terrible CF60 while his CA60 is a little closer to typical for an Oiler forward.

  3. Pingback: Zone Starts, Corsi, and the Percentages - HockeyAnalysis.com

  4. Very interesting and insightful article.
    I love the fact the NHL is finally embracing some advanced stats but this article accurately points out that a player can’t be broken down by advanced stats alone, and that some measure of “eye test” can still validate a player’s value.
    Being relatively new to advanced stats and corsi myself, I found Weber’s stats a little disconcerting so this article explaining away some of that gives me some food for thought.
    Well done! Thanks for the solid read!

  5. I’m late to the party, but I think you missed by comparing him to players on other teams than players on his own team.

    Ellis has more favorable numbers with the same C’s often with worse zone starts than Weber.

    I wrote this the other day at the comments section at Lowetide.ca :

    Here’s NSH Vollman chart for their D: http://somekindofninja.com/nhl/usage.php?f1=2014_s&f2=5v5&f3=&f5=NSH&f4=D&f7=20-&bubbleType=corsiRel&yAxis=qoc&update-filters=Update+Results

    Ellis gets a massive advantage in zone starts, but the QC is similar to Weber’s.

    As it turns out, most of ozone disparity due to the dzone starts taken with Gaustad’s line.

    Weber TOI w/ Gaustad 348 min, 12.7%OZS
    Ellis takes his turn at those starts, but not as many minutes: 99:52 min, 8.2% OZS
    So Ellis doesn’t play as much with Gaustad, but gets the brutal starts with him too (even worse)

    One way to tease out some of the difference of the difference in impact between Weber and Ellis is to look at the with each of the 4 main NAS C’s in terms of CF% WOWY and OZS%.

    With Weber 411 min – 56.5% CF – 65.5% OZS – CF% w/o Weber 56.8%
    With Ellis 288 min – 60.5%CF – 70.2% OZS – CF% w/o Ellis 55.6%

    With Weber 348 min – 40.2%CF – 12.7%OZS – CF% w/o Weber 44.6%
    With Ellis 99min – 48.0%CF – 8.2% OZS – CF% w/o Ellis 41.1%

    With Weber 310min – 50.3% CF – 50.4% OZS – CF% w/o Weber 49.7%
    With Ellis 136min – 52.2% CF – 43.9% OZS – CF% w/o Ellis 51.8%

    With Weber 250 min – 48.3% CF – 55.6% OZS – CF% w/o Weber 57.5%
    With Ellis 178 min – 54.2% CF – 53.9% OZS – CF% w/o Ellis 54.3%

    All 4 of NSH’s C’s do better with Ellis than with Weber.

    That includes Gaustad and his awful starts. He’s actually much better with Ellis with the same awful ZS, its just that Weber gets more of them.
    Fisher does better with Ellis even though they start in the OZ 6% more with Weber than Ellis.

    I don’t know if we can contribute all the results on just Weber or Ellis, their partners (Josi and Ekholm respectively) are on the ice too.

    What this does show me is that Ellis/Ekholm get better possession results in the same or tougher spots than Weber/Josi with the same team mates.

    In some spots significantly better.

  6. Great article that really helps understand Weber’s situation. However, is this not common for several top defenders? I understand that Gaustad plays a ridiculous amount of defensive zone starts, but take for example Malhotra with Subban or Glendenning with Kronwall.

    Malhotra: 33.3%
    Subban: 53.2%
    Together: 40.7%

    Glendenning: 46.9%
    Kronwall: 53.1%
    Together: 39.9%

    Weber plays quite a bit more with Gaustad than Subban or Kronwall with their respective face-off specialists, but the dramatic drop in Weber’s dCorsi is quite interesting. Definitely something in analytics to keep looking deeper at.

    Are there other defenders such as Weber that have a similar situation?

  7. Pingback: March 18, 2015 |

  8. Pingback: NHL Defensemen Who Should Be Trusted Less - TSS

  9. Pingback: The Few Positives of the P.K. Subban Trade

Leave a Reply

Your email address will not be published. Required fields are marked *