"In the regular season, the Devils gain a lot of marginal goal differential through special teams. You look at their shot attempts at even strength this season - their ratio is basically league average. But their ratio on special teams (i.e. PP shot attempts/PK shot attempts allowed) is amazingly good. That's mostly because they perennially take so few penalties, and partly because their PK is good. Either way, if you buy into the idea that the impact of special teams is lessened in the playoffs, and the impact of even strength is heightened, well that downgrades the Devils from a very good team to a slightly above average team."
We had not researched this in depth, but it strikes us as true. When we went to examine the 2006-07 New Jersey Devils, the number that jumped out at us was the Devils' extreme discipline - the Devils took 271 minor penalties that season, compared to 398 as league average. They drew 367, a difference of 96 power plays. This sounds all well and good, but let us look at the penalty differentials of the other playoff teams in the Eastern Conference versus non-playoff teams:
Playoff Teams: +12
Non-Playoff Teams: -27.4
If we leave out the anomalous Islanders, who were -85 but dragged themselves into the playoffs, the non-Devils playoff teams were +28 in penalty kill and power play differential.
We are going to try something unconventional here, using Bill James' log5 method to estimate one-game winning percentages in baseball. We are going to translate this into power plays drawn/penalties taken. The Devils, for instance, had a 57% 'winning' percentage if we look at (Power Plays/Total Special Teams Situations). So if they played the rotten Florida Panthers, who were -106 and had a percentage of 43%, we would expect the Devils to have a 64% Power Play/Total Special Teams percentage. However, playing the Tampa Bay Lightning, who had a 55% PP/TST, would reduce the Devils' power play expectation to 52% of all special teams situations. These are small potatoes - although it is interesting that New Jersey's playoff power plays for/against was 52%, when its expected value was around 54%.
Digression: We expected that the playoffs would feature better penalty-killing teams, so we used the lazy man's way of adding percentages to come up with these 'averages' of penalty kills:
Playoff Teams: 81.67%
Non-Playoff Teams: 82.2%
Strange. We suspect had we not been lazy that the percentages would be around equal.
Would that we could locate the 2006-07 even strength numbers - we cannot seem to locate the numbers we need to do this properly, with shots on goal (which ferret out a lot of variance). We can, however, express New Jersey's even strength goals for/against as a percentage at even strength:
New Jersey: 50%
None of these numbers are really getting at why New Jersey may have gotten thrashed, until again we return to the backup goalie question. Let us now consider the question in reverse: if Scott Clemmensen, the Devils' backup in 06-07, played as much as the average Eastern Conference playoff team backup, how many more goals would the Devils have let in? Let us assume that his numbers are as expected, despite the tiny sample size.
Clemmensen would start 22 games and give up 69 goals with his .889 - Brodeur would start 60 games and give up 141. This gives us 210 goals against for New Jersey instead of 187, scuttling their already miniscule goal differential into a negative. The only edge they would have in this instance is their defense and ability to avoid penalty trouble.
The Devils' insistence on playing Brodeur a great deal in 2006-07 masked their deficiencies at even strength, where even despite him they were a breakeven team. In the playoffs they get a reduced advantage from their ability to draw penalties and avoid power plays against because most teams that take a great deal of penalties are out of the playoffs.