With Cory Schneider's new contract (outlined by John here), I decided it's time we figure out just what happened last year. I pointed out in my last article that Marty's goaltender performance may have hurt us statistically in our playoff run since he was such a fall off from Cory's production. A few were perturbed by this comparably small portion of the piece since Marty did in fact win more games than Cory. This was a perplexing to many Devil's fans. In particular we scored more for Marty on a consistent basis. In this article we will introduce the problem, provide some interesting stats on the Devil's performance in with each in net, and then request directions to head from you, the readers, so we can solve it together. First I'd like to mention a few things:
1. All data was retrieved from ExtraSkater, the data combines the game-by-game info with the Goalie game logs
2. The 2 games Marty got pulled in are being treated as outliers and not counted for simplicity sake.
3. There may be several reasons why this happened, and there may be no reasons, that is what the investigation is for.
The next section will outline some of the basic troubling stats.
Below is a chart of some of the fundamental differences in the two goalies performances and the team's performance when either was on the ice.
Differences in goalies:
Cory Schneider gave up 0.32 less goals per game than Marty. Cory's 2.02 GAA (remember the pulled games dont coun't for him either) is the best of any goalie who played more than half their teams starts and 3rd best among 30-game starters. Marty's. Brodeur's was 17th among qualified keepers and would be much lower if they pulled games were included. The save perentages are also uneven though not to the same degree. The save percentage was a differences of just over 1 percent.
Differences in production:
Other than the fact that Marty had three more wins despite his worse personal stats and having played 4 less games, the team on the ice seems to have performed differently. Very differently. The Devil's scored 2.81 goals on
25.43 26.7 shots with Brodeur in net. The shot total stayed the same, but they scored about 1 less goal per game when Schneider was goalie. This is the problem at hand that requires investigation
Finding an Explanation: Attempt #1
Attempt #1: Is my first try at explaining this. My hypothesis is that it is a combination of the fact that they played a different distribution of Home/Away games (presumably Marty played more at home), and furthermore, that the quality of the teams affected the outcomes (presumably Schneider's competition were worse and/or allowed more goals against).
The Devils perform better at home. This is obvious from their records presented on NHL.com. Some of the more savvy Devil's fans may have noticed DeBoer's preference to start Marty at home. Perhaps they were aware it was his last season, perhaps it was part of their plan. Regardless, the Home/Away distribution is as follows.
There is a noticeable difference in which goalie played what type of game. If expressed as a plus/minus, Schneider would be a minus-5 and Brodeur a plus-3 for a difference of 8. That can affect statistics quite a bit but probably not to the degree we've observed.
Quality of Competition
To investigate this statistically, I chose three stats to inquire about. I took all of the opponents in the 80 games counted (we dropped 2 in the beginning) and averages their Year-end point totals, their GAAs, and the CAAs (Corsi against averages). This will judge where the team ended in the standings, how many goals they were used to giving up, and how many shots the were used to seeing attempted against. Below are the results.
|Opp Pts||Opp GAA||Opp CAA|
Oy. That went terribly. So, apparently, the teams Brodeur faced ended the season having scored, on average, about 5 and a half points better per team than Schneider's opponents. Thats a large enough margain to be reasonably justified in saying Brodeur faced better teams.
Not only were the teams that Schneider played against worse, but they were specifically worse at stopping offense. They let in more goals per game (albeit by a slim 0.05 bargain) and over one extra shot per game. Remember that these stats are strictly with reference to the opponents year-end production and not how they played in the one game against us.
So not only did was our hypothesis wrong as we failed in our first attempt to explain the difference, we actually made it seem less likely that this could happen.
Help me please...
This article is meant to be brief as it is an introduction to a larger study. So here is where you guys come in. I'm going to list a couple options below and I'd like you to vote on what I should investigate. I will choose one or two of the highest voted selections and then one of my own to evaluate and report the findings in a future article (hopefully next week's). So as mentioned before, here are some possibilities for investigation into why it seems like the Devil's played so much better for Brodeur.
*Corrections in chart credit to Smelph's comments.