Is it a bigger accomplishment to receive a (C+) in a very, very difficult class in school... or receive an (A-) in a trivially easier one?
In an effort to run some numbers on NYR goaltenders Henrik Lundqvist & Cam Talbot, we here at SatherOnWaivers have accidentally stumbled upon a whole new method of analyzing & evaluating an NHL goaltender’s performance in a season. Initially we wanted to judge Talbot’s performance in net this season versus Lundqvist, particularly this season since filling in full-time for Lundqvist since his injury. So here’s what we did:
- We went through each net minder’s seasonal game log, and noted the amount of ice time each goaltender logged against each NHL team.
- Then, we calculated each goalie's respective opponent’s # of NHL standings points, weighted in average of each goalie's specific icetimes against each opponent
- We took each goaltender’s Goals-Saved-Above-Average and divided it by number of hours of ice time. If you’ve never heard of this stat before, you can read about it here.
- In the name of relativity and perspective, we also found the same data for both goalies in the 2013-14 NHL season as well
- Finally, we procured a weighted average of how many standings points in an opponent each goalie was facing in the average-60 minute game.
Let’s say it’s the 2013-14 NHL Season, we want to compare a few goalies, and the NHL league-wide average standings points per team was 92.233. Goalie A played 60 minutes against the Buffalo Sabres, who were the worst team in the NHL (52 points). Let’s say in those 60 minutes, Goalie A had a GSAA of 3.00, meaning he theoretically stopped 3 more goals than the baseline-average NHL goalie would have stopped himself. Now Goalie B, on the other hand, also played 60 minutes that season, except played the Anaheim Ducks, who were the #1 team in the West (116 points). Yet in that game, Goalie B had a GSAA of 1.50.
Goalie A stopped 3 goals-above-average against a terrible team, while Goalie B stopped 1.5 goals-above-average against an elite team. So which performance was empirically better?
That was what we were trying to offer a remedy to, but instead created a brand new goaltending analytic: The Ludas Rating. The Ludas Rating offers itself as a measure of inflation/deflation for performance numbers, taking into account what quality of opponent those numbers were generated from in the first place.
In other words, the Ludas Rating attempts to tackle “how many standings points was Hank or Cam facing in the average game? … And even then, how does one compare different performances against varying quality of opponent?”
To solve this example...
- 3.0 GSAA in 60 minutes
- Average NHL Standings points of Average Competition: 52
- Average NHL Standings points per NHL team: 92.233
52 / 92.233 = 0.5638
3.00 x 0.5638 = 1.691
1.691 “adjusted" GSAA per hour, or a “Ludas Rating” of 1.691
- 1.5 GSAA in 60 minutes
- Average NHL standings points of Average Competition: 116
- Average NHL Standings points per NHL team: 92.233
116/92.233 = 1.2577
1.50 x 1.2577 = 1.8865
1.8865 “adjusted” GSAA per hour, or a “Ludas Rating of 1.8865
… So it would seem Goalie B has a higher/better Ludas Rating, as his traditional numbers were not as good as those of Goalie A, but his quality of competition was much higher, thus making his 1.5 GSAA/hr vs ANH adjustably better than the 3.00 GSAA/hr vs BUF.
So now that we’ve explained what the Ludas Rating is, let’s return to the original subjects of Cam Talbot & Henrik Lundqvist. Here’s their relative stats from last year, the 2013-14 NHL season:
And just like the introductory example into the Ludas Rating, we see Talbot’s .7558 GSAA/hr seems vastly superior to Lundqvist’s .1963. But, taking into account Talbot was playing weaker competition (83.536 standings points/opponent) than Lundqvist (91.716 standings points/opponent), the Ludas Rating adjusts Talbot’s GSAA/hr to be a lower rate. While Lundqvist technically has his lowered too (playing opponents less than 1% below-average), not nearly as much as Talbot’s.
Like other analytical metrics into interpreting data, the Ludas Rating is not perfect. A franchise playing against Lundqvist in October may be a very different team than the NHL Standings will reflect 6 months later. Goalies must face 328 shots in an 82-game season to qualify, as their GSAA is not tabulated under a lesser sample size. This is a quality-of-competition adjustor for GSAA-per-hour, using relative NHL Standings Points as a measuring stick to judge opposition’s “quality.”
So what about their Ludas Ratings in the current 2014-15 NHL season?
*Since the season is ongoing, and teams have not completed all their games, the metric of figuring the Average NHL Standings Points per Team is “projected”, unlike prior seasons fully in the books. That is to say, if a team has 60 Standings Points in 60 games so far, they will be treated as though they are an 82-Standings Points team as that is their current mathematical projection.
Once again we see Cam Talbot has faced overall-lower competition (avg 88.9 standings points per opponent) than King Henrik (avg 94.668 standings points per opponent).
In both seasons, the Ludas Rating has reduced Talbot’s impressive numbers to a more realistic amount, as he has not been facing quality competition. Yet it is worth reminding: Talbot, an undrafted net minder who played college hockey at University of Alabama-Huntsville, who stumbled into the NHL following an abrupt mid-season retirement from Martin Biron in 2013… has undeniably played above-average in his time in New York.
Are we expecting Talbot to play as good, or better than, Lundqvist? Probably not. Are we expecting Talbot to play above-average in the NHL? Of course. These studies show that, while his sharp career numbers are impressive, a little zip comes off the fastball when we hold Talbot accountable for playing sub-average teams for the most part. Overall, it’s hard to truly feel Cam Talbot is not performing satisfactorily... just not as well as Lundqvist.
* Please note: The data required to calculate a goaltender's Ludas Rating is very obnoxious and tedious. We are working on an algorithm to run smooth tabulations of various goaltenders' Ludas Rating, as to expand upon the metric, and provide more relative insight into goaltending leaguewide and in seasons past. We will post new data as it comes in, we ask you to be patient as this is a new equation, and currently takes a long time to configure. We look forward to your feedback!