Damien Cox Has No Clue

In a recent column on ESPN, long-time Toronto Star columnist and regular ESPN contributor Damien Cox proved once and for all that he should be banned for life from using, let alone analyzing, any statistical measures. In fact, it could be argued that he should be banned from using numbers of any kind.

For when it comes to numbers, Damien Cox has no clue.

I wonder if words can suitably express how terrible this column is. Let us find out!

The column is entitled "Only wins should be the measure of a netminder". Interesting. Controversial. Everybody knows that hockey is a team sport, so to suggest that wins and only wins should determine who is a good and who is a bad goaltender is a peculiar argument. This sentiment is reminiscent of the opinion of countless old-school "baseball men" who believe that pitchers are nothing more than the sum of their wins and losses.

What arguments does Damien present towards making his case that only wins should be the measure of a netminder? Contradictory, irrelevant, illogical, and untrue arguments:

For years, it was all about goals-against average, the statistic that was supposed to tell you all you needed to know about an NHL goaltender.

Indeed, for more than five decades, they awarded the Vezina Trophy, named in honor of the famed Montreal puckstopper Georges Vezina, strictly on the basis of which goaltender or goaltenders helped their team give up fewer goals than any other team.

That changed in 1981, when the league dreamed up the Jennings Trophy to honor the goalies with the best GAA, leaving the Vezina for the masked man adjudged "to be the best at his position" in the NHL.

It is true that for many years now, goals-against average (GAA) has undoubtedly been the most widely-used statistic for ranking goaltender performance (much to my chagrin). For some reason, the NHL and national media decided that GAA was the most important criteria in ranking goalies, but as far as I know, nobody ever claimed that GAA was ever "all you needed to know about an NHL goaltender.”

I’ve argued for and against certain goalies being better than others since I was a small kid, and from what I remember, most people were quite aware that GAA wasn’t the whole story.

How to judge that particular issue, of course, has long been a matter of some debate, perhaps no more so than when the netminding blip that was Jim Carey of the Washington Capitals scooped up the silverware in 1996.

I wish I knew what made Damien Cox call the 1996 Jim Carey Vezina win a blip. I happen to agree, but I wonder if we agree for the same reasons. See, Carey finished a respectable 2nd in wins, the category that means the most to Cox.

aside: Chris Osgood had 4 more Wins in 21 fewer games, a better GAA and a better save percentage – it was those darn 9 Carey shutouts that pushed the vote in his favour, I’m sure of it.

second aside: Carey was also lucky that the Buffalo Sabres stunk that season, because the great Dominik Hasek had won the trophy in 1994, 1995, and won it again in 1997, 1998, 1999 and 2001. Funny he knew how to win in all of those years but somehow forgot how to do so in 1995-1996!

In recent years, the wisest of the wise have decided that, like size, it isn't G.A.A. that matters most. After all, if a goalie faces five shots per game and lets in two, a flashy G.A.A. of 2.00 might not be telling you all you need to know.

Instead, save percentage, something those of us who watched the game before the Original 30 never even knew existed, has become the sexy measuring stick for goaltenders.

Cox’s fears begin to bubble to the surface. Somebody who was confident in their own knowledge and not ignorant about the meaning behind numbers might write the preceding paragraph like this instead:

In recent years, more and more people have become aware of GAA’s imperfections. After all, if a goalie faces five shots per game and lets in two, a flashy GAA of 2.00 is probably not an accurate representation of how the netminder performed.

Instead, save percentage has finally arrived as the more popular measuring stick for goaltenders.

Of course anybody who has been reading Damien Cox for as long as I have recognizes that he is terrified of what he doesn’t understand.

Thus, here he lashes out by subtly mocking people who prefer save percentage as a measuring stick as “the wisest of the wise”, implying that they believe themselves to be far smarter than the old guard, of which he is a member (never mind that in most cases they probably are smarter).

Thus, here he claims that save percentage “has become the sexy measuring stick”; the word “sexy” meant to imply that save percentage is about style over substance.

Thus, here he betrays his own ignorance by admitting that save percentage is a number that “those of us who watched the game before the Original 30 never even knew existed”. That is akin to pointing out that, back when he was a living, breathing pre-schooler, he didn’t know that oxygen existed.

For years, Cox has been watching great goaltending performances (breathing air) without any awareness of save percentage (oxygen). Of course his lack of knowledge is irrelevant to the big picture, which is that save percentage has existed for as long as hockey’s existed. We only needed to record the proper data and calculate it.

Like on-base percentage in baseball, save percentage has come to be seen as the true measure of what a goaltender is accomplishing every night.

First of all, on-base percentage in baseball is seen as one of the most important measures of what a hitter is accomplishing, not as “the true measure” of those accomplishments. Similarly, save percentage in hockey is merely the single best representation of how a goalie is performing, not the full and inescapable truth. Damien Cox is refuting a misrepresentation of his opponents’ arguments. Did somebody say “straw man?”

In addition, note the presence of the words “every night” at the end of the sentence. Statistics such as a percentage are far more powerful when they draw from a large sample size of data. In a single hockey game we can expect any of a wide range of results; it would be unwise to automatically rely on any particular statistic to rate a goalie’s one-game performance. Cox inserts the words “every night” because he wants to present a scenario in which save percentage, as a measuring stick, is as close to ineffective as possible. Unfortunately by doing so, he presents his preferred statistic, wins, in a similar but even more unfavourable light. Can anybody describe how well a goalie played last night if they were told only that the goalie’s team had lost? Didn’t think so.

The problem with this number, of course, is that it doesn't take into account the quality or difficulty of shots a goalie faces. Just how many he stops out of how many he faces.

And the problem with ‘wins’ as a number, of course, is that it doesn’t take into account the quality or difficulty of shots a goalie faces OR how many he stops out of how many he faces.

It’s rare to witness somebody argue that, in order to best understand something, you should remove some of the known explanation to make room for more unknown and more unexplained. Imagine if the medical profession worked like that! “Uh, I’m not sure all of this research into the immune system explains everything we need to know about the health of this individual. I think that instead of measuring white blood cell counts and antibodies and all of that nonsense, we should just see if the patient pulls through or not to rate their overall health.”

Well, in the new NHL, it may be time to simplify again. As in, just wins, baby.

Victories are what matter the most, and perhaps should be the decisive issue when it comes to passing on the Vezina legacy.

Like save percentage, of course, wins weren't really a big deal for the majority of the NHL's existence. That was a baseball thing, and perhaps it was just generally thought awarding a "win" to a goaltender was giving him just a little too much credit for a team effort.

Victories are always what mattered most, to all players. That’s the central goal in sports: to win! It’s just that, as Cox mentions, awarding a win to a goalie is to give too much credit for a team effort. Does he not agree with that?

Wins and losses stats crept into the game in the 1980s, however, and began to make more sense as a measure of goaltending excellence as minding the twine tent became as crucial to the outcome of a hockey game as pitching was in baseball.

In fact, by the mid- to late 1990s, it could be argued that goaltenders had usurped pitchers as the most single dominant position in any team sport.

Getting confused yet? Wins became important in the high-scoring 1980s. But then in the mid-to late 1990s goaltenders were at their most influential, which implies that in a lower-scoring environment the goalies have more impact on winning and losing than they do in a higher-scoring environment, like this season. So wouldn’t it be wise to avoid using wins to measure a goalie’s worth in the current scoring climate?

With the advent of the "new" NHL, however, that's changing again. In fact, a defenseman who can actually move a body out of the slot area may soon become the MVP of the league.

Or could it be possible that scorers might again become kings of the jungle as they were in the days when a fellow like Bernie Nicholls could pot 70 goals in a single season?

So good defensemen are about to become the players most responsible for winning and losing? Or is it to be the high-scoring forwards? I thought that wins were supposed to measure how well a goalie was playing and yet here Cox seems to be arguing that wins and losses are largely out of a goalie’s control, since goalies are at the mercy of how well their teammates play on offense and defense! Maybe now is a good time to suggest to him that we need a statistic that measures how well a goalie handles what is in his control. You know, like stopping as high a percentage of shots as he can, for instance.

If only we had such a statistic!

Which brings us back to wins. Given what we've seen in the opening month of NHL activity -- and holy smokes, hasn't it been fun? -- it may well be the only number that truly matters is the number of wins a goaltender records while in goal.

That's not just because the GAA and save percentage numbers seem inflated and out of whack, particularly for veteran goalies like Martin Brodeur, Marty Turco and Nikolai Khabibulin.

It's just that winning the game, or figuring out how to avoid complete and utter embarrassment in the face of nightly offensive onslaughts, is quite likely becoming the only issue that matters.

To Cox, save percentage numbers seem “out of whack” because a handful of well-regarded goalies have poor save percentages this season. Martin Brodeur, for example, has a save percentage of .883, good for 30th in the league. Marty Turco’s save percentage is .872 (37th in the league). Khabibulin is ranked dead last (42nd) in the league for save percentage at .847. Instead of chalking up those numbers to poor performance, however, Cox suggests that the numbers themselves are at fault.

You’d think that, to bolster his case that save percentage is bad and wins are good, his example would include goalies with poor save percentages and a high number of wins, but you’d be wrong. Brodeur is 4-5 in 9 games this season, Turco is 4-4 in 9 games and Khabibulin is 3-8 in 11 games. These three goalies have combined for 11 wins in 29 games. Doesn’t Cox think therefore that ‘wins’ totals are also “out of whack”?

If the only issue that matters (for a goalie) is “winning the game, or figuring out how to avoid complete and utter embarrassment in the face of nightly offensive onslaughts”, Cox’s own examples are failing miserably.

Take the case of 40-year-old Eddie Belfour of the Toronto Maple Leafs, for instance. The Eagle's personal stats have been taking a pounding this season. As of Thursday, his GAA was at 3.37, hardly a Belfour-type figure, and his save percentage was .889, well below the .900 mark, under which goalies historically assumed to be performing as if the eye holes in their masks were taped over.

If by “historically” Damien Cox means during the low-scoring mid-1990s to 2003-2004, then his arbitrary selection of .900 happens to be reasonable. However in the high-scoring 1980s, a save percentage of .900 was not only highly impressive, but rare. In 1986-1987, for example, Ron Hextall was the only starting goaltender to have a save percentage over .900 (.902). The next season, Patrick Roy led the league with a save percentage of .900. I was young at the time, but I don’t remember any eye holes being taped over.

On Monday, Belfour's stats weren't enhanced, as he gave up another four goals, making it 13 goals allowed in three games.

But two of those were Leafs wins, including Monday's victory over Boston in which Belfour blocked 43 of 47 Boston shots in regulation, turned away another six in overtime, then blanked Glen Murray, Joe Thornton and Patrice Bergeron in succession to win the shootout (Eric Lindros scored for the Leafs).

All of that, interestingly, overshadowed the mistakes Belfour had made on the night, including a whiff on Murray's very stoppable second-period wrist shot and a screw-up on playing the puck behind his net that led to a goal by Brad Boyes. On that play, Belfour went to play the puck behind the end line outside the trapezoid markings, realized at the last moment that he couldn't, and then was helpless to stop Boyes from firing it home.

The new goaltending gear, particularly the downsized catching mitt, has been giving Belfour fits. As well, he was likely the best goalie at playing the puck outside his crease besides Brodeur, so the restrictions in that area have affected him more than most.

"One minute you're a slug, the next minute you're a hero, so you don't know what to think," Belfour said. "I felt bad about the [Murray] goal that snuck under my glove and then that stupid play where I went out to play the puck and I got caught in no-man's land. You just feel terrible for the guys and you obviously want to do whatever you can to help out."

At the end of the night, Belfour got the big "W," and a Leafs team that has been decidedly leaky in the defensive zone had been saved by its goaltending.

When you dry your eyes, let me know and we can try to tackle the insanity of the preceding sob story together. OK, to support his case that wins are the ultimate measure for how good a goalie is playing, Cox gives us a show-stopper of an example with the Boston game where Belfour stopped 49 out of 53 shots. Folks, I present to you ... Mr. Save Percentage!! *applause*:

49/53 = .925.

One of Belfour’s best games of the season, and Damien Cox’s example of why save percentage is bad and wins are good is a match where Belfour’s save percentage was .925.

With enemies like that, save percentage doesn’t need friends.

See, figuring out the new NHL seems largely about looking at things in a different way and tossing out old assumptions.

A defenseman who can't bang bodies in his own end might be seen as far more valuable now because he can create in the enemy zone more freely. Forwards who lack brawn or speed might be more useful if they can produce on the power play.

And goaltenders, who before might have been seen as wanting if their GAA was over 2.50 and their save percentage less than .915, should now be viewed more generously, with more of a focus on the number of points their teams generate in the standings.

Because scoring is up, Cox supports viewing goalie stats more generously than before. But ignoring a goalie’s individual contribution to the team (namely how adept they are at saving shots) and instead giving them full credit or full blame when the team wins or loses doesn’t represent generosity.

If the NHL scoring climate has changed, it will be reflected in league averages. For instance, if average save percentage is .899 this season instead of, say, .909 the year before (hypothetical example), then that suggests we should adjust our views as to what a good save percentage is for this season, not that we need to throw away the save percentage statistic and switch to a more crude measure of individual success!

These fellows are, after all, also dealing with new equipment sizes, with streamlined jerseys still on the way, and veteran goalies would tell you they are working harder these days while facing more difficult shots and elongated offensive-zone possessions by enemy teams on a nightly basis.

Poor babies. Of course, ALL GOALIES ARE IN THE SAME SITUATION with respect to these issues. Arrrgh.

How does he not see that none of his points persuade as to why wins are better than save percentage? The preceding paragraph wasn’t the least bit relevant! Is he getting paid by the word? Actually, is he really getting paid at all for this crap?

So, if we put wins ahead of everything else, as of today, Detroit's Manny Legace (10 wins) and Tomas Vokoun of the Nashville Predators (seven wins) would be the favorites to win the Vezina. Just behind them would be Montreal's Jose Theodore, Florida Panthers puckstopper Roberto Luongo and Ryan Miller of the Buffalo Sabres.

Sigh. Manny Legace leads the league in wins. He also happens to be 5th in save percentage, and second in save percentage behind Lundqvist out of the full-time starters only. Vokoun is 3rd in wins. He is 9th overall in save percentage or 5th out of the full-time starters. Roberto Luongo’s career save percentage is .920 for pity’s sake and he’s at .922 this season – he’s a perennial Vezina Trophy contender precisely because he regularly stops a higher percentage of shots than most other goalies. It certainly isn't because Florida wins too many games. Ryan Miller has 6 wins which is tied for 4th as I type, is ranked 12th overall in save percentage or 7th overall among starting goalies.

Theodore has been struggling a bit to stop pucks (.883 save percentage) but has 8 wins. I guess Cox finally has his proof! Or, more likely, he's stumbled across an exception, in this early part of the season.

When you sort NHL goaltenders by save percentage, you notice right away a correlation between high save percentage and winning percentage (wins/games played) and Won-Lost record. To illustrate, the top-10 goalies as sorted by save percentage have a combined record of 57-20! Gosh, you stop a lot of pucks and good things seem to happen; you win a lot of games (unless you are Dwayne Roloson) and you don’t lose very many games (unless you are Roberto Luongo or Dwayne Roloson).

Call it the Grant Fuhr standard. In the 1987-88 season, Fuhr appeared in 75 games for the Edmonton Oilers, registering a GAA of 3.43 and a save percentage of .881.

Nothing special, right?

Except Fuhr won 40 games, and ultimately, helped the Oilers capture the Cup. For his efforts, Fuhr was awarded the Vezina for the first and only time in his spectacular career.

He was, the voters understood, better than the numbers.

Ahhh, the good old late 1980s, where men were men, goalies didn’t worry about stopping pucks but instead focused on winning games, and the Oilers scored 22% more goals than the league average.

Fuhr’s reputation is that he would frequently let in soft goals because he knew that the Oilers would score a lot, and then he would shut the door when the game was on the line. Guess what? In 1987-88, Grant Fuhr’s GAA of 3.43 was well below the league average of 3.62, and his save percentage was slightly above the league average of .880. In other words, he was an above-average goalie for a great team. The voters understood that, and given that there weren’t too many credible challengers who played in nearly as many games as Fuhr that season I am not at all surprised that he won.

I don’t doubt that his league-leading 40 wins played a huge role in his award, much to Damien Cox’s glee. But note that Fuhr played in 75 games compared to 45-55 games for most other starters. Imagine if future Hall-of-Famer Patrick Roy had played in 75 games that year instead of 45; he would have won approximately 38 games (on the reasonable assumption that he could have kept up his .511 winning percentage for 30 extra games on the 1st place Canadiens) and the voters would have chosen him and his 2.90 GAA and .900 save percentage, perhaps in a landslide.

So if you are Damien Cox, feeling threatened by numbers you don’t understand, you:

• point out the imperfections in an imperfect, but good, statistic.
• misrepresent your opponents’ opinion of the value of that imperfect, but good, statistic.
• argue against the misrepresentation that you created.
• champion a much less sophisticated measure instead, a statistic which happens to share the same imperfections that the better yardstick does but has many additional and more serious flaws.
• give examples which not only do not prove your point, but actually lend support to the statistical measure you are arguing against.
• pepper the column with completely irrelevant information.
• mention Grant Fuhr; he's your "wins, baby" trump card.
  

That was, quite simply, one of the worst sports columns I’ve ever read.

**************************

(thanks to Greg for the link)

UPDATE (November 3, 2005):

Declan over at CAtO beat me to the punch yesterday. Go read his fantastic (and far more concise) take on this same Damien Cox column.

Declan: "Scout's Honour" I did not read yours before writing mine, despite some obvious similarities. Just coincidental!