Tuesday, April 29, 2008

Mike Gillis on Goalies

New Vancouver GM Mike Gillis weighed in not just on his goaltender Roberto Luongo ("the best goaltender in hockey"), but also had the following quote about the importance of goalie play: "Whenever you have a goalie, a lot of people will suggest a goalie is worth 50 to 70 percent of your team utility."

This is a particularly peculiar quote, in that the Canucks and Vancouver observers should be in a good position to not overvalue the importance of a goalie. Their team just missed the playoffs despite having the best goalie in hockey. Sure, Luongo closed the season poorly, but why wasn't Vancouver comfortably in a playoff position by then? At the All-Star break, Luongo had a 2.10 GAA, a .925 save percentage, and 6 shutouts, and Vancouver was still only in a playoff position by a single point. And what about last season, where Luongo played at an extremely high level in the playoffs yet his team was still obliterated by Anaheim?

If a goalie is 70% of the team utility, how did Vancouver lose so many games? Based on that assumption, the only rational conclusion would be that they were by far the worst team in the league other than Luongo, and I don't think either the talent on their roster or their performance stats support that assertion.

Gillis is probably just hyping his team and his players to some degree, but there are some that are portraying his hire as a move towards a more statistical approach to the game. If so, I suggest he start checking his equations, or he might end up making some of the same mistakes Nonis did by expecting that his goalie was going to take him places that were simply unreachable without a much greater input from the rest of the team.

Friday, April 25, 2008

Shots and Shot Quality

I had to link to an excellent post at Sisu Hockey looking at shot quality and number of shots.

A couple interesting things: For shots on offence, it is not just the number of shots that matters. The type of shots taken matters a great deal, as goals for correlated just as well with expected goals (which is based on a shot quality calculation) as with simple shot totals.

On defence, however, the correlation between shots allowed and goals was just as high as it was between expected goals against and goals. This means that there is no general rule that teams that allow more shots also allow lower shot quality, or that teams that allow fewer shots allow more difficult chances. In general, bad teams allow more shots and more dangerous scoring chances. So goalies on good teams have a big advantage.

Another interesting stat is that in the first round of this year's playoffs, the team that was expected to score more goals won. There were a few games where a strong or weak performance by a goalie made the difference, but over the course of 7 games or less, it was the shooters' results that carried the day every time. If you switched goalies in all the first round series, probably the same teams would win. Just another thing to think about when someone on TV talks about how this or that series will come down to the goaltending matchup.

Tuesday, April 22, 2008

The Relationship between High-Scoring Teams and "Money Goalies"

What do Billy Smith, Grant Fuhr, Terry Sawchuk, and Gerry Cheevers have in common? Most people will tell you they were winners, they were clutch, they knew how to make the big save, they led their teams to multiple Stanley Cups, right? They didn't care about their personal stats or meaningless games. Ever hear the story about Gerry Cheevers skating right out of his net and leaving it empty because he didn't want to get hit by a Bobby Hull slapshot in a meaningless regular season game? Similar stories are often told about all 4 of them - they didn't care if they let in an early goal but they really bore down when it mattered, and they were always there to make the big save for their team at the important points in the game.

But there is another very obvious link between those goalies, and that is that they all played for some of the greatest and most high-scoring teams of all-time. Teams that finished first overall for years in a row, often outscoring the rest of the league by dozens if not hundreds of goals. The '80s Oilers were the highest-scoring team in history, the Islanders dynasty was not too far behind and was much better defensively as well, the Gordie Howe Red Wings were a dominant offensive team that finished first overall in 8 out of 9 seasons, and '70s Bruins were the highest scoring team of all-time after era adjustments. All of these teams were first overall multiple times, and this was even when the goalies were supposedly not caring about their results and saving up for the postseason. So just imagine how good they should have been in the playoffs.

And yes, those teams were good in the playoffs, and so were the goalies, at least when they had Hall of Famers surrounding them. But what if they didn't? What happened then? Grant Fuhr was 26-29 in the playoffs without Wayne Gretzky (and 63-21 with him). Gerry Cheevers was 33-33 in the playoffs (WHA included) without Bobby Orr (and 27-13 with him). Terry Sawchuk was 20-15 with 3 Stanley Cups during his first five years in Detroit. For the rest of his long career, he was 34-33 in the playoffs with just 1 Cup win in the original six era. Billy Smith went 69-21 in a five-year stretch when the Islanders were at their peak, but for the rest of his career he wasn't even the #1 goalie, outplayed in the playoffs by Chico Resch and Kelly Hrudey and posting a 19-15 record on some very good teams. This, of course, raises the question: who was really the clutch player? Billy Smith, or Denis Potvin? Gerry Cheevers, or Bobby Orr?

Guys like Smith or Cheevers are the goalie equivalent of Claude Lemieux or Esa Tikkanen, players that weren't the best in the league but played for good teams and had playoff success. Lemieux or Tikkanen have little chance to make the Hall of Fame, but goalies are evaluated by a totally different (and in my view, ridiculous) standard. Gerry Cheevers never won a major individual award, never made a season-ending All-Star team, and never finished higher than 6th in GAA. But since he played a few seasons with the most talented hockey player in history on one of the highest scoring teams of all-time, he developed enough of a reputation to be voted into the Hall of Fame.

So if those guys were mostly created by their teams, what does an actual money goalie look like? I'd go with Turk Broda as a much better playoff performer than any of the 4 listed above. Broda only once played for a team that finished first in the league in the regular season. His teams rarely led the league in goals for or goals against. Yet he won 4 Stanley Cups, and his winning percentage went from .562 in the regular season to .606 in the playoffs. It doesn't look like Broda was just being carried along by his teams. I don't mean to say either that anyone who played on a great team is just along for the ride. Someone like Ken Dryden, for example, is difficult to fault since he won the Cup every single year that Montreal had a dominant team, the team lost when he wasn't in net, and he has the 1971 Cup run where he beat three better teams in the playoffs. He is certainly different than someone like, say, Fuhr, who only won anything in Edmonton and watched the Oilers win a lot of playoff games with Andy Moog or Bill Ranford in net.

The moral of the story is that perceptions are very biased. People see winners, and they like to make up explanations for their success. If a goalie wins a lot of games or Cups, the reasoning is that they must be doing something right, so they must be "clutch" and making the "big save" and so on. The simpler, and more accurate, explanation can often be found in the goals scored column, something that goalies can do almost nothing about. If we take that account, maybe we can finally stop calling goalies "clutch" just because they happened to be the guy in the net while their Hall of Fame teammates blew away the rest of the league.

Friday, April 18, 2008

Is Brodeur Tired?

With the New Jersey Devils trailing the New York Rangers 3-1 in their playoff series, and Martin Brodeur having given up a pair of soft game-winning goals, there is some talk going around that he is simply tired from playing so many games this season.

The same things were being said last year, when Vincent Lecavalier had his way with New Jersey and Brodeur for most of their first round series and the Senators made short work of the Devils in round two.

But does this explanation make any sense? This is a guy who has played 75+ games nearly every season for the last dozen years. Why is he suddenly tired?

Let's see how Brodeur usually does late in the season. If he is tired, he should be suffering a gradual decline as the year goes on. I looked at his stats for all regular season games in April, and, surprise, Brodeur has been consistently excellent to close out the year. His regular season stat line in April since the lockout goes 15-1-0, .934, 1.92. His 2007-08 performance was right in line: 3-1-0, .923, 1.98. Not too much evidence of tiredness there.

So to those claiming that Brodeur is fatigued, it appears that it didn't affect him while the season was still underway, and just happened to show up, by pure coincidence, exactly about the same time Brodeur started facing Lecavalier, Jagr, and Heatley.

New Jersey is no longer an elite team. Their team systems and development models just can't make up for the fact that they have lost too much talent over the last few years - Stevens, Niedermayer, Rafalski, Gomez, et al. They have continued to have regular season success, and even the current edition of the Devils is still able to beat Atlanta and Toronto in November almost every time with their efficient play. But when they get into the playoffs against the elite teams they just aren't able to handle them, and Martin Brodeur hasn't shown recently that he is able to be a major difference maker in the second season. Nothing wrong with that, really, a goalie can only do so much. Brodeur went 5-11-3 against the Rangers, Canadiens and Penguins this season, so it really shouldn't be too much of a surprise that his team would face an uphill climb to make a deep playoff run.

When well-known players don't deliver in big situations, people want to either blame them or make up excuses. Brodeur already has the reputation of being a superstar clutch winner, so it isn't easy to point the finger. The usual way to reconcile reality to reputation, therefore, is to claim he was "tired" or make some other excuse. I think a simpler and more accurate solution is that this year, just as last year, Martin Brodeur is playing against a superior opponent in the playoffs, which is something that has been pretty rare throughout his career. He has little margin for error, and is not playing at the very top of his game. The result is that he is getting scored on and his team is losing. No further excuses or explanations needed.

Tuesday, April 15, 2008

Patrick Roy Never Singlehandedly Won Anything

It's playoff time, which means it is time to cue the cliches like "you need goaltending to win" and "best goaltender wins" and "he needs to come up with some timely saves" and so on. The ultimate honour, of course, is to be called "clutch" and to be given credit for singlehandedly winning a series or even a Cup for their team.

But does that ever actually happen? Most hockey fans will point to Patrick Roy, in 1986 and 1993 as an example of a goalie that won it all by himself. Unfortunately, this view is quite wrong. Patrick Roy was another goalie that showed that the goalies get too much credit for the play of their teams. He was the beneficiary of some powerhouse Montreal teams, and in his entire career he never played on a weak team with the exception of 1994-95. That doesn't mean he was not a great goalie, just that he never singlehandedly won anything despite what his fans repeatedly claim.

The Canadiens were consistently outstanding for nearly Roy's entire career in Montreal. Take a look at their year-by-year goal differentials: +47, +50, +36, +60, +97, +54, +24, +60, +46, +35. In the decade from 1984-85 to 1993-94, Montreal outscored their opponents by 509 goals over 808 games. That was the second best mark in the league behind only Calgary. Here are the top 5 teams in the league over that time period ('84-85 to '93-94, W-L-T-Pts-Goal Diff):

1. Calgary: 433-274-101, 967 pts, +590
2. Montreal, 430-274-104, 964 pts, +509
3. Boston, 412-294-102, 926 pts, +334
4. Washington, 413-312-83, 909 pts, +327
5. Edmonton, 399-314-95, 893 pts, +292

Calgary and Montreal were the class of the league. So is it surprising that Montreal won 2 Cups? Not really. After a decade of being a top-5 team they should be expected to win a Cup or two.

Now the fact that the Canadiens had Roy certainly has some bearing on that impressive record. But goalies played less back then, and about 40% of Montreal's games had someone other than St. Patrick in the net. In those games, which featured mostly weak goalies like the notorious Andre "Red Light" Racicot, Montreal went 170-128-45. This was not up to their usual standards to be sure, but is still roughly equivalent to Washington's record in the table above, showing that Montreal was a top team even without Roy.

When you look a little deeper at the 1986 and 1993 squads, it becomes clear that they were both very impressive teams that underachieved during the regular season but showed their strengths in the postseason. The 1986 Habs had the third best goal differential in the league, and appear to simply have been a victim of bad luck in close games as well as the tough Adams Division, where every team was .500 or better. Montreal went 15-15-2 against their divisional opponents, and 25-18-5 against everyone else. They may not have been overloaded with goalscorers beyond Mats Naslund and Bobby Smith, but they had a lot of excellent two-way forwards like Guy Carbonneau and Bob Gainey, and a strong blueline led by Larry Robinson and Chris Chelios.

It was a similar story in 1993. In fact as late as March 13, 1993, the Montreal Canadiens were actually in first place in the entire NHL. Then they hit a late season swoon, and were edged out by both Boston and Quebec for the division title. The 1993 version was a similar mold - a few gifted scorers (Damphousse, Bellows, Lebeau), mixed with a large group of strong two-way forwards like Muller and Keane and a solid group of defencemen.

The two teams were both lucky enough to avoid the top teams come playoff time, a significant factor in their success. In the two runs combined, Montreal never played a single team with more regular season wins than them, and they were probably the best team in all 8 matchups.

Another overhyped achievement is Roy's overtime record in the '93 run. People will often refer to the 10 straight OT wins right up alongside Roy's 4 Cups and 3 Conn Smythes as proof of his greatest-of-all-time status. In those 10 games combined, Roy played a total of 96 minutes of shutout hockey in OT. A valuable contribution, and a noteworthy one to be sure, but not such a singularly impressive accomplishment that it should automatically crown him as the greatest ever. There are a number of goalies that have strung together 2 or more shutouts in a row at key times in the playoffs, some of them even doing it against better opponents, yet without even a small fraction of the hype.

Patrick Roy had very good save percentage numbers, a substantially better winning record than his backup goalies, and some excellent playoff performances. However, he was definitely advantaged by the teams he played on, which were nearly always dominant, even in Montreal, and even in 1986 and 1993. Therefore, the label he often receives of having carried a weak team to victory is undeserved. In the final reckoning, given his opportunities and talented teammates, Roy probably won about what he should have.

Friday, April 11, 2008

Do Chokers Exist?

Goalies are very quick to get labelled based on their playoff performance, usually much more quickly than other players. It is not unusual to read, for instance, that Martin Gerber has a weak playoff record, despite only playing a career total of 241 minutes before 2008, most of it battling sickness. But are these labels really justified? Can every goalie be neatly divided into the camp of choker or clutch?

My view is that nobody is a choker. Play long enough, and everyone will end up simply playing at their skill level. To make it to the NHL, all goalies have been in dozens of pressure-filled situations. If there was a goalie that couldn't handle the pressure, he would wash out in junior or college or the minor leagues when his performance suffered under the pressure of playoffs or a playoff race or the knowledge that there were professional scouts in the stands. Much is made of how the game changes in the playoffs, but it doesn't change at all for goalies - the shots come in and you try to stop them, regardless of the number of penalties called or the amount of hitting.

For example, take Marty Turco. He had a good playoff year, then two bad ones in a row. People started questioning his playoff abilities. Then what happened last year? 1.29 GAA and .952 save percentage in a 7 game series. He wasn't a choker, he just was overdue for some good performances. Even the best goalies have bad playoffs and good playoffs; everyone remembers Patrick Roy in 1986, 1993, and 2001, but he also had 1987, 1991, and 1998. Just because one guy had a bad playoffs or two already doesn't mean he is mentally weak, it just means that most likely there are some good times still on the way. Turco has a spotty playoff record, and his first round opponent Giguere has a great one, but does that really have any bearing at all on what is going to happen between them over the next 2 weeks? After game one, it sure doesn't look like it.

To try to test my position, I am looking for goalies who had substantially worse records in the playoffs than in the regular season, over a significantly large number of games. This should not simply be a function of the weakness of their teams or the strength of their opponents, but a clear deterioration in their play. If you have any suggestions, or other evidence that relates to the topic of "clutchness", please post it in the comment thread.

Thursday, April 10, 2008

Recent Form

As the playoffs begin, here is another stats breakdown, with recent form (post All-Star game stats) for the goalies on the top contenders:

Giguere: 14-3-2, 1.61, .939, 2
Backstrom: 15-4-6, 2.06, .926, 2
Turco: 13-10-2, 2.19, .914, 2
Nabokov: 19-6-2, 2.12, .905, 1
Hasek: 10-3-1, 2.20, .901, 2

Fleury: 10-2-1, 1.53, .947, 2
Price: 15-5-0, 2.26, .931, 3
Brodeur: 18-11-4, 2.19, .921, 0
Lundqvist: 15-6-5, 2.12, .919, 3
Huet: 15-6-1, 2.31, .918, 3

Interestingly, the Eastern goalies have done substantially better over the last few months. The Western Conference has been better this season, but it looks like the defensive play in the East, especially among the top teams, has been improving, which could indicate a competitive Stanley Cup Final series.

Tuesday, April 8, 2008

Records vs. Good Teams

As we move into the playoffs, everyone is looking for indicators of which team is likely to win. There is also a lot of discussion about goalies, as this is the time of year when goalies are most overrated in terms of their impact. Some lazy analysts just look at the starting goalies and assume that the better one will win, failing to take into account team offence, team defence, special teams, etc., which is about 75-80% of what will make up the final result.

However, one thing that is interesting to look at is to see how each team and goalie did against actual playoff opponents. For the contending teams, it is perhaps even more important to see how well they did against the other top teams. I looked at how the top teams in each conference did against the other top teams from their same conference, to try to anticipate who might be likely to go to the finals. The teams selected were Montreal, Pittsburgh, New Jersey, and the New York Rangers from the East, and Detroit, San Jose, Dallas, Anaheim and Minnesota from the West.

One might expect Brodeur and Nabokov, as the two Vezina frontrunners, to have the best numbers here. In fact, it is the complete opposite, which may not potentially bode well for their teams in the playoffs.

Henrik Lundqvist: 15-3-2, 1.77, .926, 3
Marc-Andre Fleury: 4-4-2, 2.42, .923, 2
Carey Price: 4-1-0, 2.45, .919, 1
Martin Brodeur: 5-11-3, 2.58, .904, 1

Dominik Hasek: 7-2-0, 1.88, .912, 2
Marty Turco: 11-6-3, 2.36, .909, 2
J.S. Giguere: 9-9-1, 2.44, .906, 1
Niklas Backstrom: 2-5-3, 2.92, .905, 0
Evgeni Nabokov: 11-8-2, 2.51, .897, 0

Wednesday, April 2, 2008

The Playoff Benchmark

In the current MVP debate, one of the most hotly debated points is whether or not Alexander Ovechkin deserves to win the MVP if his team misses the playoffs. This is quite common, to claim that someone is not really valuable unless their team wins. This is illogical, though, because Ovechkin's goalscoring dominance would be hugely beneficial to any team in the league, from Detroit to Los Angeles. The Caps are where they are in the standings mainly because of Ovechkin's teammates and coaches. If Bruce Boudreau was the coach and Cristobal Huet the goalie for the entire season, Washington would probably be the division champions even if Ovechkin only scored 40 goals. So picking Malkin or Iginla as MVP if Washington is eliminated is basically saying that Ovechkin is less valuable because his goalies and defence are worse, and where's the logic in that?

A great goalie will make more saves than an average goalie on any team. What is peculiar is that his contribution is valued much more highly on certain mediocre teams just because it made the difference between a playoff seeding and an early golf season. This, of course, has nothing to do with the goalie himself, merely the cumulative effort of his play and his teammates.

Let's say there is a goalie so good that he saves two goals per game compared to a lesser netminder. This is a vast exaggeration, as the margins between NHL goalies are very narrow and probably around a tenth of that, but it is just an illustration. Even with that overwhelming contribution, sometimes the team around him will play so poorly that they will lose even despite the extra goals he saves. Sometimes the team will be good enough that they would have won even if he never made those stops. And sometimes the goalie is the difference maker, turning a 4-3 loss into a 3-2 win.

In all three scenarios, the goalie's contribution to the team is EXACTLY THE SAME. He saved two goals for his team. The difference between winning or losing was simply in what the rest of the team added to the goalie's contribution. Therefore, it makes no sense at all to award extra bonus points when his teammates happened to play very well, or even when they were just good enough to allow him to be the difference maker. Forget "clutch" and "making the big save" and all that, the reason he lost one game and won another has nothing to do with his play and everything to do with the rest of the team in the team sport that they were playing as a team.

That is why Roberto Luongo never won in Florida (and why Tomas Vokoun isn't winning there now) because there were too many shots against and not enough goals for. On some of his best nights he may have been turning 5-2 losses into 3-2 losses, but according to the "wins are the only thing that matters" mindset he never got any credit for anything. Jose Theodore in 2002 was pretty similar, but his teammates were a little better, so he helped the Habs go from narrowly losing to narrowly winning. Different starting baseline, so therefore different result, and he ended up with the league MVP trophy. Theodore in 2002 wasn't better or more valuable than, say, Luongo in 2004, the Canadiens were just better than the Panthers.

Making the playoffs is obviously desirable, and results in additional revenues and exposure for the franchise. That doesn't mean that any effort that falls short of the playoffs is worthless, or that a player is better or worse because his team happens to be in a tight playoff race. The playoff cut line, arbitrarily drawn at 8 teams by the NHL playoff system, should not work as an irrational magnifier where all efforts are worthless up until they cross that line, at which point they become tremendously valuable.

Tomas Vokoun or Ilja Bryzgalov aren't any less valuable this year just because their teams miss the playoffs, just as Martin Gerber and Cam Ward aren't more valuable because they and their teammates might be fortunate to experience a likely very short postseason run. Vancouver could finish 8th or 9th and it wouldn't change the fact that Roberto Luongo is one of the best and most valuable goalies in the league.

So whether Washington wins or loses their next two games, vote Ovechkin for MVP.