Analytics Quarter-Season Review, Part 1: Goaltending
<em>As we move closer to the passing of the torch, how are Pekka and Juuse doing? And how do they figure into the team's recent slide?</em>
This is part 1 of my quarter-season review of the Nashville Predators in 2019-20. Twenty games is considered the usual watermark where statistics tend to be more representative of the team as a whole, and we will begin this series with the position that has seen highs and lows already: the goaltenders.
Coming into the 2019-20 NHL season, Nashville fans were waiting in anticipation to see how changes to the team—a summer removed from a disappointing first-round playoff exit—would change the product on the ice. Many have criticized General Manager David Poile's desire last summer to keep most of the team together, and this year, Poile made some changes.
The team's vaunted defense corps lost All-Star and face of the NHL P.K. Subban in a trade to New Jersey, with the expectation that rookie Dante Fabbro could more or less fill his shoes. In his place, we saw the arrival of Matt Duchene as a free agent, and there was speculation that this could be the most talented forward group the team has ever had—so much so that, for once in the team's history, they might have too many good centers.
But that was not the case for the goaltenders. Pekka Rinne had already signed what was likely his final contract, and his heir apparent, Juuse Saros, was going to continue to see more starts. Many consider Nashville to have one of the best goaltending tandems in the league, and coming into this year seemed to be no different. And while Connor Ingram was brought to the team as a free agent this summer, most knew that in the immediate future, he'd spend his time in Milwaukee—hopefully to one day serve as Saros's backup once Rinne's time had ended.
Goaltending was the last thing on most fans'—and analysts'—minds. But the puck dropped, and things changed.
At even strength, Pekka Rinne comes in at 25th in the NHL (minimum 50 unblocked shot attempts against) with a save percentage of 92.1%, while Juuse Saros has the sixth-worst (57th) overall at 88.6% (the bottom of the list includes recently-assigned New Jersey goaltender Cory Schneider, both San Jose goalies Martin Jones and Aaron Dell, and, surprisingly, Sergei Bobrovsky). Using the expected goals model from Evolving Wild, Rinne and Saros don’t fare much better when using Goals Saved Above Expected (GSAx) - Pekka ranked 45th with -3.47 (he has given up about 3 and a half goals more than expected) while Saros comes in at 54th with -5.5.
The above chart is sorted to show the worst GSAx at the top, but you can their performance is suffering in other places as well. However, one difference stands out: Expected Goals Against/60 (xGA/60)—Rinne faces the fourth-fewest xG against in the NHL, while Saros faces the 13th most. It would appear that there’s something different happening in front of the two netminders. Let’s explore that further.
Laundry and Juuse Saros - Hung Out To Dry
In a couple of Twitter threads I posted Monday, I noticed this difference while looking at Micah Blake McCurdy’s isolate maps—showing where shots are being taken more or less than league average the scoring threat that results. For defensive measures, a lower threat value is good.
And not to bring up an old argument - one I'm still not sure about analytically - but this is also a bit curious pic.twitter.com/uSNx4UDz09— Bryan Bastin (@projpatsummitt) November 18, 2019
Juuse Saros is facing shots against at near league-average, while Rinne is getting what appears to be pretty elite shot suppression. The biggest differences? Opponents are getting many, many more shots in the crease against Saros than Rinne.
In order to look at this phenomenon and put it into context with previous seasons, I used Natural Stat Trick to pull data for both goalies for the previous four seasons. I created a few charts to visualize their performances, and to keep things concise, I’ll list a few important points for each.
Chart 1: Expected Goals Against vs Shots on Goal Against
- In the last four seasons, the most xGA Rinne has faced (shot quality) is 2.19 xGA/60 last season. This season, Juuse Saros has faced 2.21 xGA/60, by far the least he’s faced in the previous four seasons. Rinne is seeing much less quality against than Saros.
- Pekka Rinne has faced fewer shots on goal than Saros in 3 out of the last 4 seasons—oddly enough, his Vezina-winning campaign in 2017-18 saw Rinne see more shots than his backup. That same season, Saros saw the highest number and highest quality of shots of any of the previous seasons. In general, Rinne is seeing fewer shots against than Juuse Saros./
Chart 2: Rebounds Against and Rush Attempts vs Save Percentage
- One of the weaknesses Juuse Saros has had compared to his mentor is controlling rebound attempts against—Pekka Rinne is much more likely to prevent the puck going back out for second chances. In fact, Rinne has stayed between 2.4 and 2.9 rebounds against per 60 for the last four years, while Saros has ranged from 2.6 all the way to 3.9. This season, however, Saros has only allowed 2.47 rebounds against, yet his save percentage has plummeted. Shots off the rebound contribute heavily to expected goals so despite seeing higher xGA than Rinne, Saros is still controlling rebounds at a career-best level.
- In the last two seasons, Saros and Rinne have faced more rush attempts against than they have previously—Rinne facing on average 1.6 rush attempts/60, while Saros faced about 2.2 rush attempts/60. While rebounds are something at least partially controlled by the goalie themselves, team defense is responsible for giving up rush attempts against, and Saros is facing relatively more than Rinne in both of the past seasons./
Chart 3: Save Percentage by Shot Danger
- Low-, Medium- and High-Danger shots are an alternate way of denoting expected goals—they use largely the same criteria, but divide shots into buckets instead of spreading them out along a continuum—and it shouldn’t surprise you that Juuse Saros has seen a severe drop in high-danger save percentage this season (10 percent less than last season). While his overall save percentage has dropped about four percent from the average of the last four seasons, Saros is still managing decent save percentages in low- and medium-danger shots, but is struggling with the high volume of high-danger shots he faces.
- Pekka Rinne, on the other hand, is doing better than last season in terms of high-danger save percentage, saving 87.1%. It is the same with low-danger shots (97.9%) and his overall save percentage is only down about 0.8% from his average in the past four seasons. What’s incredibly odd is his medium-danger save percentage - a pretty bad 83.6%, less than what he saves on high-danger shots on average. Pekka Rinne is saving most shots at only slightly reduced levels compared to his average over the last four years, but medium-danger saves have seen a severe reduction./
Chart 4: Shot Distance vs Goal Distance
- Pekka Rinne is seeing the same average shot distance against as he did last season and 2016-17, but a much further average goal distance, something that oddly lines up with his Vezina season. Still, it is concerning that shots are coming from a similar distance, but goals are coming from farther out than last season.
- Keeping in mind that the distances between places on the chart are relatively small, Saros in other seasons has seen similar average shot distances, but goal distances have crept closer. This season, however, Saros is seeing closer shots and goal distances against, which supports the theory that the defense is allowing opponents to shoot and score closer to the net./
Shot Angle Save Percentage by Shot Distance and Number of Shots
Preface: These are new charts I’m working on as I attempt to learn to code in Python - Tableau doesn’t have much of an option to replicate this, and I’m still evaluating it to see just how informative it is. Leave me feedback and tell me what you think.
Pekka Rinne, as we saw in previous graphs, is seeing shots averaging between 35-45 feet away, and a lot of goals are coming from those angles between Rinne’s 10 and 2 o’clock (good save percentage is blue, bad is red). It is interesting to see that Pekka is giving up more goals from directly in front and from the left side from distance.
Just another way to visualize the difference between defense in front of Rinne and Saros here. Juuse Saros is letting in goals from different angles, a little bit more from his right side. Still, Saros seems to be doing better as the team defense pushes shots farther away.
Early this season, people were ready to give Rinne the Vezina—and early this season, they weren’t far from the truth. But we’re seeing a true tale of two goalies. Rinne has regressed and is letting in shots from further away, because the defense in front of him is forcing a lot of shots from low-danger areas. With the team’s shooting percentage coming crashing back down to Earth, no longer can he win games due to the team scoring four or more goals per game.
Juuse Saros, on the other hand, is facing much more pressure. The team is allowing opponents to take shots from closer to the net, leading to more high-danger chances, and it’s hard to make those saves. Still, he’s well below his regular average on high-danger chances, so while he’s cleaned up his rebounds, there’s still work to be done.
The Nashville Predators are doing a decent job of reducing quality chances against for the most part, but they are still playing differently in front of Saros than they are with Pekka Rinne. Despite the losses, Saros has started to improve (as he usually does) after a rocky start, but Rinne’s tendency to let in mid-range shots is something to be concerned about.
It feels like tradition, at this point, that this team goes as the goaltending goes, but it’s not fair to put all the blame in net. The Predators can’t get away from a system that encourages a lot of long-distance, low-quality shots, and when shooting luck runs out, well, we see what we’ve been seeing the last few weeks.