Who's the best backstop in the biz?

About once a year -– usually the week of my “keeper league” hockey pool -- I pick up a copy of the Hockey News. This former staple in my life has almost entirely been replaced by the Internet.

But my draft is tomorrow, so yesterday I leafed through a friend’s copy of THN. Besides a cover story on the Oilers, the current issue has a couple of features on goaltending, a favourite subject of this former batshit-crazy goalie. Or should that read this still-batshit-crazy former goalie. Either way, you’ve been warned.

Speaking of batshit-crazy, THN columnist Ryan Kennedy came up with a new ranking system he called the “Goaltender Confidence Index”. The formula he concocted, “based on save percentage, goals-against average, save percentage on the penalty-kill, percentage of games in which the goalie was pulled, percentage of games giving up five goals or more, winning percentage and shutout percentage” ranked Montreal’s unproven pair of Carey Price and Jaroslav Halak the best goaltending duo in the NHL. Yeah, that’s what I thought too. Meanwhile, Nashville’s duo of Dan Ellis and Nobody ranked 6th, ahead of viable tandems like NYR’s Henrik Lundqvist and Steve Valiquette (8th) and Minnesota’s Niklas Backstrom and Josh Harding (9th). Vancouver’s own “one-man duo” Roberto Luongo ranked 13th, far behind Ellis. Say what?

Elsewhere in that same issue of THN, however, there was a section called “Inside the Numbers”, where they presented (without analysis, thankfully) stats for a variety of categories over the three years since the lockout. This is a useful time frame, long enough for trends to develop and for the cream to rise to the top.

For goaltenders they listed the top 25 in five different categories: Wins, Shutouts, Total Saves, Save Percentage and Goals Against Average. Not a bad basket of stats, IMO, maybe a little top heavy in counting stats (3) vs. those based in percentages (2), but recognizing both quantity and quality. Whereas Kennedy’s formula consisted of percentages only, and some pretty contrived ones at that. Its inability to factor in volume of work is its undoing in my opinion.

Anyway, seeing those five columns of stats I wondered if they might be useful to rank the NHL’s best goalies since the lockout. 16 goalies made the top 25 in all five categories, so I simply awarded them points for their position on each list, and sorted to the lowest aggregate. (My method was very similar to Kennedy's, a total coincidence since I hadn't read his column yet when I did this little study.) A very crude method to say the least, but crude can be good. I didn’t even bother with half points for ties; e.g. the three guys tied for 5th in Sv% all got 5 points, a slight benefit to them but so what? Presenting results with more significant digits than the method supports is a pet peeve of mine, a common error which I prefer to avoid.

As I anticipated, the cream rose to the top. Here’s the list (a perfect score is 5):

Rank. Goaltender **** W SO Sv Sv% GAA Total
-------------------------------------------
1. Martin Brodeur *** 1. 1. 2. 4. 3. = 11
2. Miikka Kiprusoff * 2. 2. 3. 8. 8. = 23
3. Henrik Lundqvist * 6. 3. 8. 5. 2. = 24
4. Roberto Luongo *** 3. 5. 1. 5. 13 = 27
5. Tomas Vokoun ***** 8. 7. 6. 3. 15 = 38
6. J.-S. Giguere **** 7. 11 12 5. 4. = 39
7. Marty Turco ****** 4. 9. 11 22 5. = 51
8. Manny Legace ***** 10 3. 19 14 7. = 53
9. Cristobal Huet *** 19 7. 22 2. 10 = 60
10.Ryan Miller ****** 5. 21 6. 17 17 = 66
11.Rick DiPietro **** 9. 16 4. 19 24 = 72
12.Evgeni Nabokov *** 10 14 17 24 9. = 74
13.Tim Thomas ******* 17 20 9. 10 22 = 78
14.Martin Gerber **** 12 21 18 18 20 = 89
15.Vesa Toskala ***** 13 16 23 24 14 = 90
16.Martin Biron ***** 19 21 21 14 22 = 97

Now as first-order approximations go, that looks like a pretty damn good list to me. No surprise to see Martin Brodeur at the top, of course; hockey’s version of Albert Pujols records excellent-to-outstanding numbers across the statistical spectrum year after year after year. In this particular study -- in which reputation counts for nothing, results everything -- Marty’s not living on past glories, only accomplishments since the lockout are even considered. Brodeur is the only guy to rank in the top 5 in each category; otherwise only Kipper and Lundqvist managed to crack the top 10 across the board, while only Luongo made top 5 in as many as 4 of the 5 categories.

Of course it’s naïve to consider each of these categories to be of equal value. All of them measure team success as well as individual success in different mixtures. (There is much more to be said about team effects on goalie performance than I can discuss in this post, but suffice to say they are substantial. But for now I’m just letting the raw numbers speak for themselves.)

Surely, though, Sv% is a better rating of a goalie’s true performance than, say, shutouts. So I decided to do a weighted average, scoring the stats themselves in order of significance. My first system was 5-4-3-2-1 for Sv%, GAA, Wins, SO, and Saves, respectively. Where the original system was weighted 60/40 for counting stats, this weighted system flipped it to 60/40 in favour of percentage-based stats. In other words, quality first, then quantity.

For all that I now valued one stat five times as much as another, the results didn’t change a whole lot. Here’s the Top 10 (a perfect score is 15):

1. Brodeur 39
2. Lundqvist 65
3. Kiprusoff 85
4. Luongo 97
5. Giguere 106
6. Vokoun 118
7. Huet 143
8. Legace 153
9. Turco 171
10.Miller 216

Lundqvist, a three-time Vezina “finalist” in his first three years in the league, slips comfortably into second ahead of Kipper but still far behind the guy who actually won the last two Vezina Trophies (and 4 of the last 5). Giguere squeezes into 5th ahead of Vokoun. Huet, who has a tremendous save percentage, moves up two spots while Turco (whose specialty is getting the puck moving in the other direction) slides down.

There are those who argue that the only truly individual goalie stat is Sv%. I don’t agree -- it too depends on team play -- but perhaps the goalie has more control over that one column than any of the others. So I reassessed the weights to 10-4-3-2-1, thus giving Sv% a 50% impact on the total score, and percentages in general a 70/30 preference. And still, the results didn’t change much. Here again is the Top 10 (a perfect score is 20):

1. Brodeur 59
2. Lundqvist 90
3. Luongo 122
4. Kiprusoff 125
5. Giguere 131
6. Vokoun 133
7. Huet 153
8. Legace 223
9. Turco 281
10.Thomas 288

Luongo scrapes into the top three ahead of Kipper, while Thomas noses ahead of Miller for 10th. Huet, 2nd in the NHL is Sv% in 2005-08, remains in 7th but is more readily grouped with the guys above him on the list than below. And above them all Marty Brodeur is home and cooled out as the top statistical goalie in the game.

You might have noticed in the first table that none of the listed guys finished first in either Sv% or GAA. The mystery man is Niklas Backstrom, who led the league 2005-08 in both categories, at .923 and 2.17 respcetively. Those totals were achieved over just two years, as Backstrom came to the NHL in 2006. Thus he is deficient in counting stats. I calculated a third, pro-rated season for Backstrom, which of course maintained his excellent percentages while theoretically bumping him into 7th in shutouts and 12th in Wins. But he remained completely out of the top 25 in Saves, so still didn’t meet my intial criteria. Backstrom’s a good goalie, but to consider him the best in the league over the past three years based on fewer than 100 GP and fewer than 2500 saves would be a stretch.

As for the Oilers, while Roloson, Garon and even Conklin appeared on some of the Top 25 lists, they didn't play enough, or well enough, over the full three years to qualify among the elite.

Enough about goalies, already! It's time to get my nose back into my McKeen's to prepare for tomorrow.