clock menu more-arrow no yes

Filed under:

Off the Charts: Don’t pay special teams percentages much mind

New, comments

There are better ways to evaluate special teams effectiveness.

NEW YORK, NEW YORK - OCTOBER 08: Brent Burns #88 of the San Jose Sharks skates against the New York Islanders at the Barclays Center on October 08, 2018 in the Brooklyn borough of New York City. The islanders shutout the Sharks 4-0.
No more Burns bombs, please
Photo by Bruce Bennett/Getty Images

Last year, the Sharks finished the regular season with the 16-highest power play percentage and the second-best penalty kill percentage. This season, we’re already off to a running start with references to last year’s special teams units.

The problem with referring to just percentages when it comes to special teams is that it misses a lot of what is actually happening. In other words, using percentages alone is not a good way to evaluate the effectiveness of a special teams unit.

Let’s look at an example. Say the Sharks get four power play opportunities one game. They score one goal on those first three opportunities, so their power play percentage on the night is 33 percent (1/3). But, when did they score the goal? Let’s say they score the goal one minute into the third power play. In reality, they scored a goal on just 2.5 power play opportunities, so their true percentage should be something like 40 percent. Now, let’s say that on the fourth power play, the team takes a penalty 30 seconds into the man advantage. The team’s power play percentage falls to 25 percent (1/4) on the night, despite the fact they only had 2.75 full power play opportunities.

If we wanted a more effective measure of special teams percentage, then the Sharks power play would be 36 percent on the evening (1/2.75) vs. the 25 percent they’ll be recorded as having. In one case, the Sharks should be commended for scoring quickly. In the other example, they should be reprimanded for taking a penalty during the man advantage. In either case, their power play percentage drops below what they were truly able to accomplish while they had an extra skater. If a team gets a double minor or a five-minute major power play, and scores on it, they’ll still just receive recognition for scoring on just one opportunity.

Using rate stats, we can put everything on a level playing field. Instead of creating a percentage based on the number of opportunities — which is easily skewed by a shortened power play here and there — we can look at how well they performed per hour of special teams time. Rate stats take into account shortened and elongated power plays and ensures we are rewarding the team for what they did well regardless of how much time they spent on the ice.

Last year’s penalty kill was not nearly as good as the percentage suggested

Last year, we talked about the team’s penalty kill, and how it wasn’t nearly as effective as its second-best kill percentage suggested. Last year, the Sharks allowed the 10th-lowest rate of shots during the 4-on-5 penalty kill, an impressive mark. However, the rate of unblocked shots and expected goals the team allowed was about league average (15th-best, in either case). If you prefer visual representations of that, here is a heatmap of the unblocked shots the team allowed on the penalty kill:

hockeyviz
That mass of purple in front of the goal represents a lot of unblocked shots against
@ineffectivemath

The blob of purple directly in front and a bit to the left of the Sharks’ goalmouth? A whole mass of unblocked shots the team regularly allowed. It’s clear by both the numbers and visualization that the Sharks didn’t do a very good job of limiting dangerous chances during the penalty kill. The team remained near the top of the leaderboard because of their goaltending.

Martin Jones saved 90 percent of all 4-on-5 shots he faced last season. That’s the 53rd-best mark of all 301 goalie seasons of at least 200 minutes at 4-on-5 since 2007-08. Aaron Dell didn’t kill enough penalties last year to qualify for this list, but he also stopped 90 percent of all 4-on-5 shots he faced.

On that same list of 301 goalie seasons, Martin Jones’ 8.38 goals saved above average (GSAA) mark ranks 15th out of all 301 seasons. Dell’s 2.72 figure isn’t as impressive, but it still would have ranked among the top 90 goalies on that same list. This is despite the fact the Sharks allowed an expected 4-on-5 save percentage of 86 percent with Jones on the ice and that of 87.4 percent with Dell on the ice. In Jones’ case, that figure represents the eight-lowest mark of those 301 seasons.

The Sharks 2017-18 penalty kill allowed a historically bad amount of dangerous chances and shots against. It was their goaltenders who provided the second-best kill percentage. Goaltenders are also penalty killers, so the kill percentage can’t be ignored totally. We can put everything together like this: The Sharks 2017-18 penalty kill was middling-to-bad in terms of the shots and chances it allowed, but the team’s goalies played exceptionally well to mitigate that fact.

The Sharks 2017-18 power play was one of the best in the league after some mid-November tweaks

Likewise, the team’s 2017-18 power play percentage belied its actual effectiveness. The Sharks power play finished the season ranked 16th overall in terms of percentage. But the numbers below the percentage tell a different story. Had DeBoer and his benchmen stayed the course early in the season, the Sharks power play percentage might have looked even worse than it did at season’s end. During their November 18 game against the Boston Bruins, the coaching staff made changes to the power play, giving birth to what would become a veritable monster for the rest of the regular season.

After that game, from November 19 onward, San Jose produced the following 5-on-4 marks (league rank):
Shots for per 60 minutes: 118 (3rd)
Unblocked shots for per 60 minutes: 86.9 (3rd)
Expected goals for per 60 minutes: 8.4 (3rd)
Expected shooting percentage on unblocked shots: 9.7 percent (10th)

This is what those metrics look like on a heatmap:

hockeyviz
Lots of shots from the Ovi spot
@ineffectivemath

San Jose sent plenty of one-timed pucks careening toward net from the left dot (the Ovi spot), as well as their fair share of chances from the low slot. Their shooting percentage on those unblocked shots was just south of 10 percent (12th-best mark), and the team’s overall shooting percentage was 14th-highest. The team generated shots and chances among the best teams in the league, they just didn’t finish. Last year, after those changes in November, the Sharks’ power play wasn’t bad. In fact, the man advantage was quite the opposite. It’s fair to say the Sharks had a top-10 power play in the league but were unlucky (or suffered from bad shooting) to not have scored more goals.

If the Sharks would like to rise to a likely top-10 unit, here is the number they should strive for versus where they currently are:

Power Play

  • Shots for per 60 minutes: 100 // 66.1
  • Unblocked shots for per 60 minutes: 74 // 44
  • Expected goals for per 60 minutes: 6.5 // 2.68
  • Expected shooting percentage on unblocked shots: 9 percent // 6.1

Clearly, the newish-look power play has been pitiful this year. Things started to look up during the team’s game against the Islanders, but they have a long way to go before creating any meaningful damage. It’s a wonder they don’t revert back to what was working during last season’s final 65 games.

Penalty Kill

  • Shots against per 60 minutes: 90 // 102
  • Unblocked shots against per 60 minutes: 68 // 78
  • Expected goals against per 60 minutes: 6 and fewer // 11.28
  • Expected save percentage on unblocked shots: .914 // .856

Even more so than the power play, the Sharks’ penalty kill needs a lot of help. They’re getting gashed by opposing power plays, and it is unlikely both goalies pull out elite seasons on the 4-on-5 again.

This season, and every season thereafter, pay attention when someone cites the Sharks penalty kill or power play percentage. Don’t take that at face value — raw percentages alone contain a lot of luck and goaltending in them and aren’t indicative of how the team’s special teams are performing. Just like at 5-on-5, teams that are taking lots of shots and generating tons of chances but not scoring a lot are likely to regress. Head somewhere like Natural Stat Trick or Corsica and look up 5-on-4 or 4-on-5 rates for a better idea of how effective a team has been on either side of the man advantage.

So far in this short season, San Jose has been awful on both sides of the special teams ledger, regardless of what their percentages say. Don’t pay the percentages too much mind, whatever you do.