The analytical community relies heavily on shot differential (Fenwick or Corsi) for gauging team success because shot differential has been shown to correlate strongly with puck possession and offensive zone time and to be a good predictor of future goal differential. One obstacle that we face in this analysis is that teams play conservatively with a lead and allow the opponents to outshoot them. The community historically has focused on results when the score is tied to remove these score effects, but at Broad Street Hockey this year we have begun correcting for score effects with Score-Adjusted Fenwick (SAF).
The biggest benefit of making this adjustment is that we get a much larger sample size -- on average, the score is tied for only 18 even strength minutes per game, whereas Score-Adjusted Fenwick uses over 42 minutes per game. The result of the larger sample size is that more accurate predictions can be made after fewer games.
It has previously been shown that for predicting playoff results, the period of time after the trade deadline is particularly important, so I'm interested in looking at how teams have performed since then. Honestly, I'm nervous about reading a lot into 12-14 games using any metric, but this is where the larger sample size of Score-Adjusted Fenwick should be most useful so let's see what it tells us.
Here are the Score-Adjusted Fenwick totals since the trade deadline (through games on 3/25):
|Team||Post-deadline SAF||Pre-deadline SAF||Change|
Let's get an important caveat in before we dig into this. Like I said before, even though SAF uses more data than Fenwick Tied, a ~12-14 game sample is still small enough to make me nervous. It covers as much gameplay as a ~30-35 game sample of Fenwick Tied, but over the course of a month things like strength of schedule and travel could play a significant role.
Still, as we head towards the playoffs, it's interesting to note which teams are playing well. Here's what jumps out at me in this data:
- The Kings are scorching hot. In their last five games (through 3/25), they outshot Anaheim, Nashville, San Jose, St Louis, and Boston by a collective 167-98 at 5-on-5. They're looking like an awfully scary opponent for the Blues or Canucks.
- Some bubble teams dominate our list of most-improved. Washington, Ottawa, and Los Angeles are making a good push for the bottom playoff spots. On the other hand, Calgary is a sad tale of small sample sizes. They have the single biggest improvement in shot differential, and a 6-4-4 record to show for it, in no small part because they went 0-3 in the shootout (and are 3-9 for the year).
- Toronto is making a great charge at a high draft pick. They've actually edged out Minnesota for worst score adjusted Fenwick since the trade deadline.
- The Flyers also haven't been so great lately. We're feeling pretty good about their 10-3-1 record, but we should all be well aware of just how much of that is the result of Ilya Bryzgalov's hot streak, and just how unsustainable it is. He won't have a .952 save percentage at 5-on-5 in the long run, and they won't kill 92.3% of the penalties they face either. Right now, Bryzgalov's performance is masking some puck-possession deficiencies.
- Pittsburgh has the second-best SAF since the deadline (54.4%). Florida is 23rd since the deadline (47.9%). If Bryzgalov's hot streak causes the Flyers to fend off New Jersey for the #5 seed so they can face Pittsburgh, and Bryzgalov then gets blamed for a playoff loss in the first round, that would be like rain on your wedding day.
I had to make one tweak to the numbers to produce this data. When I first published the Score-Adjusted Fenwick, I explained that I would prefer to use a weighted average based on the team's actual spent in each game state -- if a team was up by two goals a lot, then I'd want to put more emphasis on how they performed when up by two. The numbers here are all taken from behindthenet.ca, which unfortunately has a bug in how it reports the time spent in each game state, so I relied on league averages for the weighting instead.
This didn't seem like a big deal until I went to produce numbers based on results since the trade deadline and found that the calculations for the Blues blew up because they've never trailed by two goals in this 13-game period, so I didn't have a Fenwick-when-trailing-by-two score for them to average in. Uh-oh.
If we want to use SAF at the beginning of a year, we'll have to get this ironed out -- hopefully the behindthenet.ca TOI numbers get fixed in the off-season. But for now, I'm using the Blues' Fenwick-when-trailing-by-two score for the year, and doing the same for a few other teams that had fewer than 30 minutes in a certain game state.