Tuesday, November 18, 2014

The Oilers Shooting Percentage and Why Calgary Stinks

There seems to be a lot of consternation in the Oilers world about how the local Ice Stick Game team is as bad as last year. If you go by the historical standings from a year ago, the Oilers were 22 games played with a record of 5-15-2 for 12 points (29th), with a -29 goal differential (30th). Fast forward to today, and the Oilers have played 18 games, gone 6-10-2 for 14 points (27th), with a -16 goal differential (29th). In other words, marginally better, 4 games in hand with a 2 point lead is a small improvement. But we are talking about comparing shit prepared by a fine French chef versus shit that a horse's ass has air dropped onto a grassy field: it's still shit.

The thing is, you know where the Oilers are right now. They aren't in a good place, but like I said, you knew that. So as a fan, it is more important to determine where this team is going. Is it a fiery oil tanker capsizing at sea, with a McDavid mermaid swimming up to inspect the wreckage? Or are the blue and orange pulling themselves out of a oily, dark abyss to finally see the light? One obvious way is to look at points. It stands to reason that if a team is bad at picking up points in the first 18 games, they will probably be lousy at doing that the rest of the season. It's not incorrect logic per se, but it can be statistically inferred that Corsi Tied is a better predictor of future success than winning percentage (credit to the awesome JLikens, RIP):

For three whole seasons, for every segment of games, the Corsi Tied (that's the T) is a better predictor of future success than the teams current win percentage! With that in mind, the Oilers are currently ranked 17th in the league with a Corsi Close of 49.9%. Corsi close is like corsi tied but also allows for either team being up by one, which expands the number of events to be counted and is a little more reliable. The Oilers last year were 43.4%, which was 28th in the league. So a couple things to take from this:
  1. By shot metrics - which we've established will predict their future success with more accuracy - the Oilers are a vastly improved squad (+6.5% is a monumental shift in this regard).
  2. By shot metrics the Oilers are probably a better team than their points percentage indicates; they should be performing at about the middle of the bell curve. The current 17th overall team is Ottawa, who are 8-5-4 for 20 points, with a +2 goal differential.
Of course the Oilers aren't 17th in the league. They are an abysmal 27th. Which begs the question: why aren't they seeing the results that the shot metrics are predicting? One reason is simple enough: the Oilers have had an awful PDO. PDO is simply the addition of the teams shooting percentage with the teams save percentage. Aside from a few truly terrible and tremendous teams, this number tends to converge on 100%. The Oilers are 29th in the league with a PDO of 97.03. Only Carolina (96.66) is worse. The Oilers goaltending is 28th (90.07), and their shooting 24th (6.96).

I'm not going to get into goaltending (and the team save percentage), simply because a) it's been poor - and it may not improve substantially, and b) your eyes would bleed if I included any more numbers than I already have. What I will look at, is what will happen if the Oilers shooting percentage rebounded from 24th in the league to a more middling number.

Strap on your pocket protectors people, we are about to deep dive into some math.

5v5 Scoring, normalized to League Average Shooting%

To do the back of the envelope math: the Oilers are currently shooting at 6.96% 5v5, and if we correct it to league average (8% typically, the Oilers averaged 7.83 over their last two seasons), and multiply the 15% increase (8/6.96) to EV goal scoring, we have the Oilers at (1.15 * 2.08 = ) 2.39 EV goals, boosting their overall number to 2.75. That would be good enough to 12th in the league - essentially doubling their offensive rank.

5v4 Scoring, normalized to League Average Shooting%

It gets a little more interesting when you look at the 5v4 scoring for the Oilers. It turns out shooting percentage is extremely volatile, even over long stretches of the season, but I can use the 12.8% league average to do some (hopeful and really imprecise) work to decode the Oilers current PP woes. First, the Oilers are shooting the puck at about 49.9 shots per 60 minutes, and their shooting percentage is a 27th ranked 8.05, which gives them a rate of 4.02 goals/60. Now let's adjust that to league average: 12.8 / 8.05 = 59.0% increase = 4.02 * 1.59 = 6.39 GF/60 5v4. Since the Oilers have received 104:36 of 5v4 time, which is 9.7% of their total ice (104.6/60*18), that should increase their overall scoring to (6.39 * 0.097 + 2.39 = ) 3.01 GF/60! That would be 6th in the league, so double the previous offensive rank again.

There are a number of (probably insane) extrapolations here:
  1. The Oilers will finish with 10% of their ice on the PP - last year they had about 8.7% of their time on the PP.
  2. Power play shooting percentage is very volatile, and teams can have long stretches with lousy numbers (although 8.05 is extremely low, a two year average of the last two years the lowest is 9.15 by Carolina).

If you haven't swallowed your own tongue by now, we all need to pay attention to one primary thing: shooting percentage tends to move towards league averages, shot rates do not. I didn't manipulate shot rates (repeatable), only regressed the Oilers shooting percentages to mean, and the Oilers went from a bottom feeder on offence to a top flight team behind such juggernauts as Calgary and the Islanders (heavy sarcasm there).

So the idea here is this: the Oilers will not likely remain with the second worst save percentage in the league. How unlikely? If it remained the same (90.07), it would be the lowest number except for one team in the last 5 years, the '12-'13 Flames. Assuming the shooting percentages correct somewhat, the Oilers will be demonstrably better in putting the biscuit in the basket.

If you are wondering why I included the Flames in the title, I'll let my tweet speak for itself:

They are essentially the opposite case of the Oilers: sky-high PDO, rock bottom Corsi For Close. If you want to take a guess as to the two teams trajectories, expect a gentle uneven slope slightly upwards for the Oilers - glowing praise - and a spiral into the flaming pits of hockey hell for Calgary. At least, that's what my slide ruler told me.


  1. Your math is all fine and dandy but guess what, we are the worst team in the west while the lames are 1 point behind the ducks for tops. I understand that you are saying we should be better and the lames should be worse because of math but in the end it comes down to wins vs losses, that's the only math that we should care about. When the losses pile up it doesn't help me ease the pain in knowing oh wait we are better then the losses indicate. In the end this is another losing season, Oil will not make the playoffs and if anything we are cheering for another high pick.