Wednesday, September 29, 2010

The 1993 Canadiens in OT

Anyone who followed the 1993 NHL playoffs remembers that the Montreal Canadiens had an exceptional record in overtime. The Habs were 10-1 that year in games that went past 60 minutes, and even that one loss tends to get forgotten since it came in the Habs' first playoff game. As a result, the more oft-cited statistic is "Montreal won 10 straight playoff overtime games in 1993".

The journalistic narrative that year was that Patrick Roy singlehandedly won the Cup. Given that Roy didn't score any of Montreal's 10 overtime goals, any reasonable person would have to conclude that there was more to it than that. Roy's contribution was surely significant, but how much of the OT streak was a result of his play, and how much was earned by the shooters in front of him?

The Hockey Summary Project now has the 1993 numbers posted, which means that we now have access to shot data that might help answer that question.

In 11 overtime sessions in 1993 Montreal was outshot 66-59, yet scored 10 goals to their opposition's 1. That's a 16.9% shooting percentage and a .985 save percentage. Needless to say, that is a remarkable run (it is not often you see a PDO number of 1.15, even in a very small sample).

The league average save percentage in overtime in 1993 was .907. Montreal wasn't really affecting that average much since the total save percentage in overtime periods involving the Canadiens was .912, it was just heavily skewed in the Habs' favour.

If they had the league average OT shooting and save percentages Montreal would have been expected to score 5.5 goals and allow 6 in overtime. That means they outperformed their expected goal differential by 9.5 goals, of which 4.5 were the shooters outperforming and 5.0 was the goalie outperforming.

That gives approximately a 50-50 split in total contribution between the goalie and the shooters. Without question St. Patrick was great that spring, but even the best goalies need help if they want to win anything. Contrary to popular myth, Montreal had a very good team in front of him, and the Habs were able to increase their odds by managing to avoid the league's top teams in their playoff bracket. It was still a very heavy dose of good fortune that Montreal was able to be that opportunistic in those high-leverage situations, as without the Canadiens' great record in close games they probably would not have won the Stanley Cup.

Monday, September 27, 2010

What To Expect From: Jimmy Howard

If I had to bet on one starting goalie from last year posting a lower save percentage in 2010-11 than he did in 2009-10, I'd pick Jimmy Howard.

All the signs are there for a regression. First of all, Howard outperformed on special teams. He had a very good save percentage on the penalty kill (.905, 4th among starting goalies), and he faced a low percentage of shots against on the PK to begin with (16.0%). He only let in one shorthanded goal on 50 shots against while Detroit was on the power play. Put all that together, and the result is that Howard's even strength save percentage (.925) was nearly equal to his overall save percentage (.924), something that is rare and generally not sustainable in the long run.

For any goalie with essentially one NHL season under his belt, it's best to look at their minor league performance to see if there is a track record of success. Howard played four full seasons in the AHL from 2005-06 to 2008-09, where he compiled a .911 save percentage on 5,324 shots. That's not bad, but it's nothing that suggests future NHL stardom either. Howard's backups combined for .895 on 4,171 shots, so perhaps there is some evidence that Grand Rapids was not the best defensive team or took a lot of penalties or has a miserly official scorer, but I still don't think Howard's minor league performance is at a sufficient level to foreshadow a future elite NHL starter.

None of Howard's AHL seasons were more than .005 above or below that .911. There's not a clear improvement trend in his numbers, which means that I wouldn't put it at all out of the realm of possibility that he might be a similar goalie in skill level now to what he was the age of 23 or 24. His ascension probably had as much or more to do with spots opening up ahead of him in the organization as with the development in his own game.

It's pretty obvious that the Red Wings themselves didn't think Howard was anything special until very recently. Why else would they have signed Ty Conklin in the summer of 2008 to back up Chris Osgood during the 2008-09 season? Howard was 24 years old with three full seasons in Grand Rapids under his belt at that point, and his own team still didn't rate him as good enough to be an NHL backup. Having said that, the Red Wings are known for their patience in developing prospects, so perhaps it was entirely a matter of maximizing Howard's playing time, but regardless he didn't exactly force his way into the NHL either.

I'm not willing to make the argument that there are strong team effects boosting Howard's performance without a lot more data, but on the other hand I don't rate playing behind Nicklas Lidstrom, Pavel Datsyuk and Henrik Zetterberg to be one of the league's toughest assignments either.

Excluding the 2009-10 regular season, Howard's career NHL record is .912 on 614 shots. That's a tiny sample, but I think it's still probably more representative of Howard's true skill. Puck Prospectus' VUKOTA has Howard projected at a .914 save percentage next year. Guys who have great years are likely to regress somewhat the following year, that's just basic sports statistics. If they don't have an established track record of success either, then the indicator lights are flashing even more strongly. I'd say Howard is more likely to have a below-average save percentage (say, .905-.910) than he is to match his .924 this coming season.

There is of course some small chance that Howard either legitimately became great or his run of luck continues and he remains near the top of the save percentage leaderboard, but it's certainly not the way to bet. It will be interesting to see whether his performance over the next couple of seasons indicates that he is anything special or just another guy at the NHL level.

Thursday, September 23, 2010

What to Expect From Steve Mason

The Columbus Blue Jackets and Steve Mason recently came to an agreement on a two-year extension for the 2011-12 and 2012-13 seasons, at an annual cap hit of $2.9 million.

This deal doesn't make sense to me. First of all, the number seems high relative to Mason's comparables, particularly Carey Price's cap hit of $2.75 million. Secondly, and probably more importantly, I don't think the Blue Jackets really know yet what they have in Mason, who followed his good but overrated 2008-09 season with a pretty weak 2009-10.

Steve Mason's career save percentage in the NHL, regular season and playoffs combined, is .907. That's a below average number in today's scoring environment. I will cut him some slack because he did break into the league at a very young age, and it is entirely possible that he will continue to develop into a goalie with significantly better numbers down the road. But that's why I think this season is pretty important in terms of pegging Mason. It's his third year in the NHL, he's 22, and by all accounts he worked hard on his game over the summer. If I was running the Blue Jackets, I'd tell Mason to prove to me that he deserves to get paid, rather than giving him a sweet deal coming off of a down season. Maybe Mason breaks out and it costs me an extra $500K per season to buy his remaining RFA years, but in the cap era I think it's better to avoid costly mistakes than to pay market value for guys who deserve it.

A lot of people talk about consistency in goaltending, and often it is debatable whether they are actually talking about the variance in the goalie's performance or whether they are simply criticizing or complimenting the goalie's ability. So far in his career, I think it is quite fair to say that Steve Mason has been inconsistent. Kent Wilson at The Score put together an interesting graph of the game-by-game results so far in Mason's career. In 120 career starts, Mason has 15 shutouts and 13 games with an .800 save percentage or worse. That's a shutout percentage of 13% and an awful outing rate of 11%, which are both well above the averages of 6.5% and 7.7% respectively.

I've heard a few explanations given for Mason's results last season. Some questioned his conditioning, others his mental toughness. Teams were shooting high glove on him with great success, which suggests that he needs to work on his technique. Probably a lot of it was simply higher-than-normal random variance resulting from a small sample size. Once again, to me that's a reason to be cautious. If you focus on his shutouts and great games and the way he broke into the league by storm in late 2008, it's probably easy to think that Mason just needs to fix a few things in his game and he is headed for greatness, but that's an overly optimistic viewpoint.

Mason is particularly interesting because he's a guy that has been rated highly by the scouts throughout his career, often perhaps higher than his numbers would suggest he actually deserved. He got drafted in the third round despite not playing very much as a 17-year old in the OHL, he started for the Canadian world junior team ahead of Jonathan Bernier, he was invited to the Team Canada Olympic camp last summer, and he won the Calder and was nominated for the Vezina in 2009. His new contract was mostly based on projection, which again likely relied heavily on input from the team's scouts. Mason is big and moves well, he looks like a butterfly goalie should look, but unless he's actually stopping the puck at a high rate that doesn't translate into helping his team win hockey games.

Columbus is making a bet that they didn't need to make (or at least one they didn't need to make right now), and they likely parted with more money than they needed to, given comparable contracts. I'll be interested to see how it works out for them. I think we'll know a more about Steve Mason after this season, but as of right now I'm not convinced he's an above-average NHL goalie, either now or in the near future. It's a pretty safe bet to expect him to rebound from last year, but I'd still be a bit surprised if he repeated his rookie season mark of .916, especially with Ken Hitchcock no longer behind the bench.

Tuesday, September 21, 2010

Ken Dryden's Backups

One of the great benefits of the Hockey Summary Project has been filling in the gaps in the save percentage record going all the way back to 1952-53. That greatly expands the potential analytical work that can be done on hockey goalies throughout history before the NHL began officially tracking shots and saves in 1983-84.

I got the idea to run the numbers on Montreal backup goalies in the 1970s after following along with the terrific work being done by Black Dog Hates Skunks on the 1972 Summit Series. The scoring chance numbers being compiled there make a pretty good case that Canada was the superior team, but that perhaps the biggest reason the series ended up being close was lackluster goaltending by Ken Dryden.

It shouldn't be too surprising that it would be easier than normal to play goalie on a team that had Larry Robinson, Serge Savard and Guy Lapointe on defence in an unbalanced league further ravaged by player losses to the WHA. The question is how much? Here are the numbers:

Denis Herron:
Montreal: .901 on 2178 SA
Elsewhere: .889 on 9412 SA

Phil Myre:
Montreal: .904 on 1432 SA
Elsewhere: .881 on 11317 SA

Rogie Vachon:
Montreal: .912 on 5391 SA
Elsewhere: .890 on 16221 SA

Michel Plasse:
Montreal: .888 on 866 SA
Elsewhere: .880 on 8015 SA

Wayne Thomas:
Montreal: .906 on 1424 SA
Elsewhere: .886 on 5563 SA

Bunny Larocque:
Montreal: .894 on 5860 SA
Elsewhere: .861 on 1829 SA

Average save percentage for Dryden backups in Montreal: .902
Average save percentage for Dryden backups elsewhere: .885

I don't think anybody should be surprised by the confirmation that being the Habs' netminder in the 1970s was a pretty sweet gig.

That second number is actually even lower if you weight the other goalies' save percentages based on how much they played in Montreal, although Larocque's numbers have a big effect there since he makes up such a large part of the sample and has a comparatively low amount of playing time outside Montreal.

It's possible that the difference between some of these goalies is somewhat exaggerated as they would have been more likely to face expansion teams as a backup in Montreal and more likely to face the league's best teams (including the Canadiens themselves) as a starter in Pittsburgh or Kansas City or wherever else they played. It's not always an apples-to-apples comparison either because some of these guys were on their way up or their way down when they went through Montreal. Still it's a six goalie sample that supports what we already know: that the Canadiens had such a strong defence. On top of that, the Habs also did not take many penalties, which would be a further benefit for the team's goalies.

If we adjust Dryden's numbers based on the above difference, his career .921 becomes .907. That remains a very good number, given that the league average over that period was .893, but that certainly puts Dryden in the conversation with Tony Esposito (.912 over the same stretch) and Bernie Parent (.914 in the same seasons on a much more heavily penalized team) for the title of the best goalie of the 1970s.

Thursday, September 16, 2010

The Era of Bargain Goalies

After the past offseason, I feel like I should be changing my handle to The Not-Quite-So-Contrarian Goaltender. A number of NHL teams came around to what stat guys have been preaching for a while now, that you shouldn't commit big bucks to the position because there is little margin in goaltending these days and the supply of decent goalies currently exceeds the demand.

Given the economic realities and the nature of the game at the moment, it is simply the smart team-building move to avoid committing big cash to any goalie that hasn't already demonstrated beyond reasonable doubt that he is a difference-maker. That's not great news for the financial plans of goaltenders or prospects aspiring to make the NHL one day (perhaps leading to a lot more Dan Ellis Problems), but that's just the way it is. That message is spreading around the league, from the Hawks walking away from Antti Niemi to the Sharks cutting ties with Evgeni Nabokov to the Canadiens choosing to trade rather than pay their playoff hero Jaroslav Halak.

I've seen some dismiss this as merely the typical rush to copycat the most recent Stanley Cup winner, but I'd certainly dispute that argument. The signs of a goalie glut have been there for a while now, and it's been 7 seasons since a team won a Cup with a Hall of Fame goalie. The five starting goalies with championship rings since 2006 have an average post-lockout save percentage of .906. The league average over that period has been .907.

Average goaltending is good enough these days for a team that has quality in the rest of their roster. Given that, it's not at all surprising that the market price of goalies has been dropping in the salary cap era. The blueprint for many teams is to invest in the guys up front to try to assemble a lineup that can outchance the opposition. That alone should be enough to get into the playoffs and maybe even win a round or two against a weaker opponent, and from that point it's just a matter of crossing their fingers and hoping to get the hot goaltending and/or shooting luck needed to get their hands on the Cup.

I've been busy over the summer working on the Hockey Prospectus annual and other projects, but should be back on a regular posting schedule with training camps starting up around the league.