clock menu more-arrow no yes mobile

Filed under:

The Devils are Playing Well: How to Assess Acheivement Early in the Season

The Devils are 1-1-1. That doesn’t look too great on the face, but some stats indicate we’ve actually played very well. Others don't. I look at some of both in this article.

New Jersey Devils v Florida Panthers Photo by Mike Ehrmann/Getty Images

(Disclaimer: Topic was written before win against Ducks and people on social media were already calling the season a lost cause. So this was written to quell those fears. Edits have been inserted on 10/19 to keep the content current as of publish date)

It’s always hard to assess teams so early in the season. A binary response of a win or loss always lends itself to overreation in either direction. Had Hall put in the OT winner in the first game, or had Cammalleri and/or Greene’s goals counted in the second game, we’d all be singing a drastically different tune. But, as of this article, se sit at 0-1-1 and we are looking for answers.

H-R’s Simple Rating System (SRS)

Early in the season, I look at a stat from Hockey-Reference called SRS. Originally made for football, SRS is a simple rating system that takes into account strength of schedule and goal differential and nothing else. A primer from creater, Doug Drinen, can be found here.

According to SRS, the Devils are — as of Tuesday morning -- the 3rd best team in the league. The reason for this is almost completely closed system that is our schedule. We’ve lost two games to two teams who both won their other game which was against a Detroit team that just obliterated a previously 2-0 Senators team. So the logic goes:

Ottawa (3.57, 5th) = good (2-0 before playing Detroit) Edit: and beat Coyotes by 3 to go to 3-1 on the season

Detroit (7.23, 4th) = clearly better (beat Ottawa by a lot)

Florida (9.78, 1st) and Tampa (9.28 2nd) = Really f***in good (both beat Detroit by multiple goals)

Devils (3rd 8.56) = slightly worse than Florida and Tampa (only lost by one), but better than Detroit (Detroit lost to Florida and Tampa by more than we did).

No, you’re not misreading that no there’s not a typo. All these are the top 5 teams in the NHL according to this statistic. It’s not coincidental though, it’s computational. If you really get into the weeds (if you don’t want to, skip to the end of the paragraph), you’ll see this particular segment works out like a tower more than a Rock, Paper, Scissor game. FLA and TB had wins supported by DET beating OTT. OTT was legitimized when its first two victims, MTL and TOR, both won all other games. MTL and TOR both beat a team that won their other games as well. And by this point, we are already 4 degrees of separation away from something that could be considered an objectively bad team at this point in the seaon. All of which adds up to the Devils being the 3rd best team in the NHL.

(Edit: After last night’s games, the Devils are 5th in the NHL, having fallen behind MTL and DET)

This extreme dependence on the outcome of games that you were not actually a part of is one of the many contentious elements of SRS. This is why you should not view it as the Bible of Hockey Power Rankings. It is however, superior to goal differential and point percentage.


The other thing to look at is PDO (save percentage plus shooting percentage — commonly dubbed the “luck” statistic). I’m going to take a slightly altered view here, because I have an issue with PDO. Individual players and individual goalies have percentages that have independent means. However, when you look at 20 + skaters, enough noise is introduced to make it seem random. This is why people call it “lucky,” — because on a macro level, it regresses to the mean. That is less true with goalies though -- especially on teams that rely on one player heavily. That goalie’s save percentage, not the league average, should be viewed as the regression point in these cases.

The reason I bring this up is that the Devils have a PDO of 98.3 which is not low enough to say we are due to improve. However, that number is buoyed by a SV% of .926 which, while well above league average, is almost identical ot Schneider’s career average of .927. Now that may go down a little due to randomness and due to other goalies not named Cory playing games, but it is far from shocking.

Compare that to shot percentage, a statistic that irrefutably regresses to the mean. The Devils stand at 5.2 — last in the league by a healthy margin. The league average is 10.7 and there’s only one other team with below 6 (Capitals — 5.8). The Devils Sh% by year has been 8.38, 8.82, 8.78, 7.3, 9.59, 8.14, 8.95, 8.81, 9.08, 5.17, with the 5.17 being this year. Our average is 8.82 and we’ve only had below 8 once. If we assume our team average of 8.82, we would be projected for 5.1 goals this year and we would be a positive-differential team.

(Edit: S% is now 5.8 — 3rd worst in the league, Sv% is 0.935 — 4th best, PDO is 99.6 — average)

Now those statistics all support the fact that the Devils were a better team than we felt they were after the first two games. But, you can also find stats like those from Corsica which show that we are the 7th worst 5v5 Corsi team, 7th worst expected Goal ratio team, and the 5th worst scoring chance ratio team. So maybe we are worse than our record implies

I’m not exactly sure, and that’s sort of the point. There are a lot of different stats you can look at early in the season that are decidedly better than Win%, but when all is said and done, the amount we don’t know is much larger than the amount we know because the sample size is so small. We absolutely do not have enough evidence yet to figure out if this team is good or bad. Feel free to continue prognosticating away though. I know I will.