Monday, February 7, 2011

Hall of Famers Come In Fours

Looking through the list of goalies in the Hall of Fame sorted by birthday, there seems to be a repeating pattern. Several goalies will be inducted that were all born over a short time period. That will be followed by an extended dry spell, before another cluster of similarly-aged candidates gets enshrined. This is followed by another gap, and the cycle repeats.

This doesn't particularly apply to the oldest goalies in the Hall of Fame (Hugh Lehman, Georges Vezina, Hap Holmes, Clint Benedict and George Hainsworth), all of whom were born between 1885 and 1895. Then again, things were quite different back then in hockey's early days, with several different professional leagues, teams folding and moving all the time, and some goalies having very atypical career curves, often playing many years of amateur hockey before breaking into the professional at an advanced age.

The first group of Hall of Fame goalies who spent the large majority of their careers in the National Hockey League were born around the turn of the 20th century: Roy Worters (1900), Alec Connell (1902), Tiny Thompson (1903), and Charlie Gardiner (1904). After those four, it took a decade to produce the next Hall of Famer (Turk Broda in 1914).

Broda was quickly followed by Brimsek the following year and Bill Durnan the year after that to create the trio of goaltending legends that dominated the NHL in the 1940s. In the decade after Durnan, Chuck Rayner (1920), Johnny Bower (1924) and Harry Lumley (1926) were born. Lumley and especially Bower made most of their Hall of Fame cases after the previous "Big Three" had retired.

Perhaps the best goalie cohort of all was the 1929-1931 group, which includes Terry Sawchuk, Jacques Plante, Gump Worsley and Glenn Hall. In the wake of that dominating quartet there was a gap of eight years until the next Hall of Famer, and even then the next two inductees born (Ed Giacomin in 1939 and Gerry Cheevers in 1940) are both among the weakest goalies enshrined, with Hall of Fame cases largely built on taking advantage of an unbalanced league.

It took 12 years after Hall until the next no-doubt first-ballot Hall of Fame netminder came into the world in the person of Tony Esposito in 1943. Once again, he was quickly followed by a couple of others: Bernie Parent in 1945 and Ken Dryden in 1947, with Billy Smith arriving not far behind in 1950.

The next decade (1951 to 1961) didn't see any goaltenders born who would eventually be considered Hall of Famers, and that is probably not likely to change either with the games played leaders from that period including Mike Liut, Greg Millen, Andy Moog, Kelly Hrudey and Don Beaupre.

With 1962 came Grant Fuhr to break the drought, but 1965 was the real money year for goaltending ability, producing three first-ballot Hall of Famers in Dominik Hasek, Patrick Roy and Ed Belfour. All goalies born in 1965 combined to play a 4,687 games in the NHL, well ahead of any other year on record, and Tom Barrasso may still have an outside chance at making it four Hall of Famers from one birth year.

Yet again the feast and famine pattern looks like it will continue, with Curtis Joseph the only goalie born from 1966 to 1971 that is likely to be even seriously discussed by the Hall of Fame committee, and although some goalies have significant portions of their careers remaining the only guaranteed Hall of Famer born since 1971 remains Martin Brodeur.

What is causing all this? Is it merely that the random allotment of goaltending ability just happened to result in some groupings close together? That's likely part of it, but it's not particularly probable that a similar pattern would have repeated itself essentially four times in a row.

Factors that may have had some impact are the level of league scoring, the size and level of parity in the league, and changing league rules or revolutions in goalie training or techniques. Certain periods seemed to be set up better than others to create Hall of Famers, either because lower scoring levels led to lower GAAs and higher shutout totals or because an unbalanced league made it easier to rack up wins on the top teams.

Beyond that, it seems apparent that opportunity would have played a significant role, especially in smaller leagues with only 6 or 12 starting jobs available. In reality it was likely even more restrictive than that, given the required level of team success typically needed to produce a Hall of Fame career. With two or three teams dominating the standings year after year throughout much of the NHL's history, goalies usually either had to be fortunate enough to be signed by those elite clubs, or else they had to play well enough for long enough on one of the league's bottom-feeders that they were eventually given the chance to don the sweater of a Cup contender.

I'd say that the most likely explanation is that a few elite goalies have a tendency to monopolize awards and gravitate towards the best starting jobs in the league, making it that much harder for the guys coming after them to put together the trophy case and team success that the Hall of Fame has historically required for entry.

I don't rate goalies based on traditional accomplishments like Vezina Trophies, postseason All-Star selections and Stanley Cups, but many people do, including apparently most of the Hall of Fame voting committee. In that type of evaluation method, I think it is important to consider the strength of a goalie's teammates and the quality of goaltending peers he was competing against, and to consider not just how many times a goalie won an award but also how many times he got close to winning. After all, the best goalie in a particular season and the best goalie in the league are not always the same thing. There are additional useful ways to verify how well a goalie was rated by his contemporaries, if you consider that to be an important piece of information, such as looking at observer accounts and other primary sources, salary/trade history, international selections, coach and player polls, etc.

That said, I still maintain that a careful analysis of historical statistical performance remains the best method to properly evaluate the accomplishments of a netminder, with care taken to adjust for team factors during the parts of the league history that were more unbalanced than others.

11 comments:

Anonymous said...

A bit offtopic, but the NFL's Ben "Chris Osgood" Roethlisberger came up short last night after his defense wasn't able to bail him out this time. Catch any of the game?

The Contrarian Goaltender said...

Yeah, I was cheering for the Packers and was glad Roethlisberger lost too. I'm not sure whether it's fair to compare him to Osgood, but if Big Ben had picked up his third ring I'm sure there would be a lot of people comparing him favourably to much better quarterbacks, that's for sure.

Anonymous said...

uh.. yea... what greeeaatt article trying to depict brodeur as a fraud... welllllll done.. NOT! nice try tho.
considering you only mention his name once.... way to go. you're so cool! :)

Jonathan said...

Fat Ben's individual stats are far above average. He is not the same as Chris Osgood.

The problem with NFL stats is not only the typical "proven winner" fallacy, but also that individual stats are so messed up--especially for quarterbacks--that an efficient passer is often rated behind a guy who compiled loads of TD passes and yards.

TD passes are grossly overweighted by the average fan or telepundit.
QB rating is a horrendously stupid statistic. A QB who goes 10-for-10 and zero passing yards get a rating of 73.8. A guy who just throws ten incompletions gets something in the fifties. A TD pass is essentially the same thing as 100 yards per QB rating, which is nuts.

I can go on and on, but the fact is that people in the NFL just don't know how to evaluate QBs.

As in hockey, people look at the wrong numbers: counting stats instead of efficiency/averages. Because Peyton Manning throws 70% of the time and Fat Ben only throws 50% of the time, people assume his stats are just not up there with Peyton Manning. Yet this year, Fat Ben was arguably one of the five best QBs in the league--based on yards/attempt, INT ratio, and sack ratio, and things like that.

Jonathan said...

http://wp.advancednflstats.com/playerstats.php?pos=QB

for some info on NFL stats. In my opinion, the best "bottom line" stats are EPA, EPA/play, and AYPA. They don't account for a few team effects and certainly don't account for strength of schedule, which is a bigger factor in football. But they're solid baseline estimates.

overpass said...

As in hockey, people look at the wrong numbers: counting stats instead of efficiency/averages.

I haven't been paying attention: has the usage/efficiency tradeoff issue been settled for quarterbacks?

Anonymous said...

@Jonathan

I agree that QB rating is stupid, and efficiency stats are more important, however unlike with hockey goalies, some counting stats are meaningful for QBs and efficiency stats can be a bit misleading.

In the example you give, you compare Manning throwing 70% of the time to Big Ben throwing only 50% of the time. Since throwing the ball is more effective then running in terms of YPA (A good RB averages 5 yards per carry, while top QBs average over 8 yards per throw attempt), why don't teams just throw the ball every play? Well obviously because the more a team throws it, the more their opponents are going to expect a pass and adjust their defense accordingly. Hence I think it's fair to say that for a fixed QB, throwing efficiency is negatively correlated with throwing attempts at least in terms of throwing attempts as a percentage of total downs.

This year Big Ben averaged 8.2 yards per attempt on 32.4 attempts per game. Payton averaged 6.9 yards and 42.4 attempts per game. Pittsburgh was 17th in the league averaging 4.1 yards per rush attempt while Indy was a dismal 3.8 yards per rush attempt, good for 25th in the league. So if Pitt averaged twice as many yards every time Big Ben chucked it up compared to handing it off to Mendenhall, why would you only put the ball in Ben 's hands half the time? Because the team doesn't trust him with a heavy work load. Obviously they want the number of attempts that maximize their chances of winning, and for Pitt, this means only letting Ben throw it 50% of the time.

Indeed, despite Ben ranking fairly high in many statistical categories, he is only 42nd all time in terms of attempts/game, way behind contemporaries like Brees, Manning or Brady. This is despite the fact that only once since 2004 have the Steelers had a running game that was in the top 10 in terms of yards per attempt.

My point is that looking at efficiency numbers for a QB doesn't tell the whole story like it does (more or less) for goalies. When I see a QB with high efficiency numbers but low pass attempts, I question why that QB isn't passing more.

Agent Orange said...

"When I see a QB with high efficiency numbers but low pass attempts, I question why that QB isn't passing more."

Thats like saying "This goalie has a high save % but only started 50 games. Why didn't he play more?"

There are too many things out of a QBs control that can limit his number of pass attempts.

1) Good running back
2) A coach who "establishes the run"
3) A good defense.

Pitts allowed 14.5 ppg last season compared to Indy's 24.45. Pittsburgh had more leads which allowed for grind it out time with the run.

Big Ben's higher yards/attempt is also going to further limit the number of attempts he is required to make (less plays required to drive the field).

Now I don't think Ben is a better QB than Peyton but to use attempts/game as your measuring stick is just silly.

"The problem with NFL stats is not only the typical "proven winner" fallacy, but also that individual stats are so messed up--especially for quarterbacks--that an efficient passer is often rated behind a guy who compiled loads of TD passes and yards."

I'll need some examples here. If you are arguing that a 20/10 TD/INT guy will be rated higher than a 25/20 TD/INT then I would argue thats how it should be.

Based on his number of attempts Peyton Manning would need to throw 81 TDs in a season to CAP out that portion of the passer rating formula.

There are 4 portions of the NFL passer formula.

Completion %
Yards/Attempt
TD/Attempt
INT/Attempt

Between Ben and Peyton last season the effect on QB rating of the first 2 portions was about 1.1 overall rating.

80% of the difference is between TD/INT ratio. Ben's ratio of 3.4 was much more efficient than Peyton's 1.94.

My major issue with QB rating is it doesn't include other things that QBs do that help/hurt their team, such as rushing yards, rushing TDs and fumbles lost.

Anonymous said...

That would be why the actual name of the term is "Passer Rating", not "QB Rating".

Agent Orange said...

Semantics much? Its been referred to as QB rating in every comment that talked about it.

Everyone who follows knows that the "passer rating" is intended to rate the quarterback position. Why aren't sacks included in the passer rating?

If Vick or Freeman lose the pocket and cut out to a 8 yard gain that isn't good/efficient quarterbacking? But if a Brady or Manning slide around in the pocket and complete a 5 yard pass that is?

Jonathan said...

Well, QB rating does tend to correlate rather closely with a QB's ability to pass. However, there are some obvious flaws. I don't want to take credit for this, so I'll post the link:

Here's a sneak preview: "weighted yards per attempt with a bonus of 20 yards for each completion, an additional 80 yards for each touchdown, and a 100-yard penalty for each interception. So two completions for ten yards each are worth the same as one completion for forty yards and one incompletion. A ninety-yard pass play from goal line to the opponent ten is worth the same as a ten yard TD pass..."

I can go on, but the link explains it much better. In the case of Ben Roethlisberger, nobody notices that he has had a borderline-elite season because he only has 17 TD passes. Fortunately, the apple never falls too far from the tree. Despite its obvious areas of improvement, QB rating does seems to get pretty close to actual QB performance.

It's like saying that GAA correlates with sv%. It does, but I know which one I'd rather look at. I'd also rather look at y/a and Int rate as my baseline stat, then make adjustments based on sacks, fumbles, comp%, etc.

As far as pass attempts/game, I see your point--it's easier to pass efficiently when you run 55% of the time, instead of 30% of the time. However, I would also argue that the "run to set up the pass" effect is overrated. I've read stat studies indicating that teams generally run too often, and pass too often in short goal-to-go situations. Like a boxer who jabs too often, they spend too much time trying to randomize, and not enough time just going for the most damaging plays.

Anecdotally, I've heard people say that the N.O. Saints explosive offense in 2009 was primarily due to their running game. This year, some people said that the Packers were in trouble against the Steelers because their offense was one-dimensional. Fortunately, others predicted that the Pack would just ignore the run and proceed to cut up the Steeler secondary. As fate would have it, Aaron Rodgers had a brilliant statline, despite numerous drops from his receivers.