Sunday, April 5, 2009

Estimating 1970s Save Percentages

There has been a lot of work done in putting together hockey box scores, led by the Hockey Summary Project, which has enabled us to look at save percentage numbers for many past seasons. There are complete save percentage records available for the original six period from 1955 to 1967, and official save percentages are available since 1982. The period in the middle, however, is lacking, which makes it difficult to evaluate those particular goalies, including Hall of Famers like Dryden, Esposito, Giacomin, Cheevers, and Parent.

I do have playoff save percentages for every season in that period, so I had the idea of comparing how many shots against each goalie faced compared to league average in the playoffs, and using that number to estimate regular season shots against totals. From there, since we know the actual historical GAA, we can estimate save percentages for the seasons that are missing and thereby arrive at an estimated career save percentage for each goalie.

I tested this out, and it seemed to work well for goalies who consistently qualified for the playoffs, especially those who played on the same team. For example, it estimated Martin Brodeur to have a .913 career save percentage, and he is actually at .914. From the work of one dedicated Flyers fan, the one who is hosting the Hockey Summary Project, we have Bernie Parent's complete stats as a Flyer. His career save percentage in Philadelphia was .917. My estimated save percentage? The same .917.

These numbers probably make the method seem more accurate than it is, but I think the numbers are probably fairly reasonable. The system does not work well for some goalies, mainly those who had a number of seasons where they missed the playoffs or had a few deep playoff runs with one team but played the majority of their careers somewhere else. An example is Gilles Meloche, who played eleven seasons in the non-save percentage era and only made the playoffs in two of them. His non-playoff teams would have been very likely to give up more shots against than his playoff teams did, so his save percentage is probably understated. Roger Crozier is another guy whose results aren't that meaningful, since he played most of his career on bad teams but had a couple of deep playoff runs on good teams which make up the majority of his postseason participation.

Here is the complete list of goalies who faced at least 10,000 (estimated) shots against, sorted by goals over league average:

GoalieSv%LgSv%SASeasonPlayoffGOA
Esposito.909.89028,287537.512.7550.2
Dryden.924.89211,444366.245.2411.4
Parent.917.89918,023324.427.3351.7
B. Smith.894.88319,142210.658.8269.4
Resch.894.88016,688233.614.9248.5
D. Edwards.890.87913,124144.4-3.0141.4
Vachon.899.89322,969137.82.8140.7
Palmateer.892.88210,994109.916.5126.4
Favell.908.89811,899119.02.9121.9
D. Bouchard.889.88318,552111.3-3.8107.5
Gilbert.890.88811,67523.4-8.315.0
G. Smith.900.89916,69016.7-1.914.8
Meloche.885.88523,9240.09.49.4
Cheevers.889.88911,6320.00.00.0
Giacomin.900.90216,795-33.6-24.7-58.3
Crozier.898.90514,178-99.2-6.7-105.9


The lack of parity in the 1970s means that there is a lot of team effect included in these results. All of the guys in the top 5 played on very good teams, and their backups also did well by this measure. That's not to say they weren't good goalies, but of course they played in ideal situations.

I think Rogie Vachon, Mike Palmateer, and Dan Bouchard do pretty well here considering their teammates. If Billy Smith faced 5% easier shot quality than those guys, for example, he would be behind all three of them in career regular season goals above average. That seems very plausible to me, especially when you take into account how well Resch did in New York. On the other hand, Dryden would have needed to face 25% easier shot quality than Vachon to end up at the same level, and it is unlikely that Habs were anywhere near that good. The evidence suggests that Dryden was an example of a great goalie on a great team.

For guys like Gary Smith and Gilles Meloche just to show up at average is a strong result, since I highly doubt the California Golden Seals allowed league average shot quality against. I bet both of them were better than someone like Doug Favell, despite the 100+ goal gap between them. If Smith and Meloche faced shots that were 5% harder than average, they would rank up around the same level as Bouchard, Palmateer and Vachon.

To estimate the difference in save percentage from the best teams to the worst, we can look at the save percentages we do have. There are save percentages available from the 1970-71 season, for example, the fourth season after expansion, which gives some time for player movement between teams as well as a few years for the expansion teams to start recruiting and developing talent. The average save percentage for goalies on Original Six teams: .911. Average save percentage for goalies on expansion teams: .898. Even if we assume that the O6 goalies were 10% better than the goalies on expansion teams, which is almost certainly well overdoing it, that implies that the shot quality against was 5% easier in Toronto, Montreal, Boston, New York, Detroit and Chicago. I'd bet the true shot quality difference in those early years was probably closer to 10%.

I'm still not sure exactly how to rank Tony Esposito. His numbers are very good, at or near the top in every scenario, and yet I still feel that he was only the 3rd best goalie of his generation. Esposito is similar to Martin Brodeur in terms of playing a lot of games every year behind a mostly good defence.

The two guys who appear to have some explaining to do are Hall of Famers Gerry Cheevers and Ed Giacomin. How does Cheevers end up exactly at league average in save percentage in both the regular season and the playoffs if he was such a great goalie? And presumably Hall of Famers should not be below average in the regular season and dreadful in the playoffs like Giacomin was? I think Cheevers is in the Hall of Fame because of Bobby Orr and because Gilles Gilbert's bad playoff play made Cheevers look clutch, while Giacomin is in the Hall of Fame because of the New York media and fortunate timing (playing on an original six team right after expansion he grabbed a couple of First Team All-Star awards before the emergence of Dryden and Parent).

I think evaluating goalie play in the 1970s is all about figuring out the appropriate team adjustments, because there was so little parity around the league. It is pretty clear that Esposito, Parent and Dryden were the three best goalies of the decade (they also ranked 1-2-3 in my measure of GAA vs. backup goalies). I think it is still unclear who was the best of the next tier of starters. It's not fair to simply give that credit to whoever happened to play in Philadelphia or Long Island. I think there were a number of overlooked goalies in the next tier of solid starters who simply never had the fortune of playing with talented teammates (Meloche, Palmateer, Bouchard, Gary Smith, etc.).

When we are finally able to put together the complete save percentage record, it will be interesting to compare these estimates and hopefully break down team effects in a little more detail so we can figure out who comes next after the big 3 in the ranking of best goalies of the 1970s.

10 comments:

overpass said...

Good post. I completely agree that team effects are very important to evaluating this generation of goaltenders. For example, I've always thought that Ken Dryden must be the most difficult star player in NHL history to evaluate, at least in the modern era. While his SV% were good, if there was ever a team that allowed low quality shots you'd expect it to be the Montreal Canadiens of the 1970s.

If you wanted to look into team effect some more, you could try looking at what happens to the (estimated) SV% of goalies like Rogie Vachon who go went from the "good teams" to the "bad teams". It's likely the sample sizes would be too small to conclude anything with confidence.

I just realized in clicking around hockey-reference that full shot data and SV% are available for the WHA. To add to a couple of names mentioned here, I thought it was interesting to see that Gerry Cheevers consistently had very good SV% in the WHA. Bernie Parent's SV% was very ordinary in his one year there. Jacques Plante had a league-average SV% at age 46.

Speaking of Plante, what did his numbers look like by your estimation method? What aboutt Glenn Hall?

James Benesh said...

Really good stuff. I've been playing with those playoff numbers for a year now, and I'm surprised I never tried to approximate regular season numbers with them before. I am not surprised at all that Dryden, Esposito and Parent came out on top.

The Contrarian Goaltender said...

Speaking of Plante, what did his numbers look like by your estimation method? What aboutt Glenn Hall?

Both Plante and Hall did very well.

Plante: .919 sv%, .904 avg
Hall: .919 sv%, .908 avg

It's likely the sample sizes would be too small to conclude anything with confidence.

I looked in some detail at the late 1970s. It seems like goalies really moved around on the bottom-feeding teams, but it's still pretty tough to assess teams like the Islanders and Habs. I can estimate it, but like you say sample size is the issue.

I think Denis Herron is pretty illustrative, though, of the difference between bad, average, and good:

K.C. ('75-'76): 15-52-15, 3.96
Pitt ('77-'79): 57-55-32, 3.35
Mont ('80-'82): 43-18-17, 2.80

I imagine the save percentage differences would be relatively similar.

I just realized in clicking around hockey-reference that full shot data and SV% are available for the WHA. To add to a couple of names mentioned here, I thought it was interesting to see that Gerry Cheevers consistently had very good SV% in the WHA. Bernie Parent's SV% was very ordinary in his one year there. Jacques Plante had a league-average SV% at age 46.

Sure, but there are team effects in that league as well. It does look like Cheevers was one of the best goalies in the WHA, but the fact that his backup had an .896 save percentage over 4 seasons suggests that his team was better than most.

I also think Parent might have been pretty good in the WHA despite his save percentage results, when you look at his teammates.

Parent: 33-28-0, 3.61, .886
Backups: 5-12-0, 4.39, .871

Next season, post-Parent:
Goalies: 27-50-1, 4.28, .875

Bruce said...

Excellent work, CG.

There are save percentages available from the 1970-71 season, for example, the fourth season after expansion, which gives some time for player movement between teams as well as a few years for the expansion teams to start recruiting and developing talent.

True for 1967 expansionists, but 1970-71 was itself an expansion year, which always skewed league results away from parity. The newcomers were weak sisters, and the wholeleague would be weaker due to dilution of talent. A deep, talented team such as the Boston Bruins were primed to take full advantage.

I remember well this game from that season where Joe Daley was given first star honours in the Boston Garden despite allowing 8 (eight) goals and posting a Sv% of "just" .889. The first ever Buffalo-at-Boston game, and it wasn't a fair fight.

Given the 1970s contained no fewer than three expansions, one contraction, and one merger, your comments about parity are absolutely relevant. It didn't exist. There were plenty of "NHL" games which had all the tension and suspense of a Canada-Kazakhstan game, where you expect one team to outshoot by a 2-1 margin and outscore by more than that. Percentages are skewed, and not to be trusted.

Which leads us inevitably to Ken Dryden. Any study that concludes that Kenny D. was the best goalie on the best team is entirely in line with my appreciation of that era. I was no Habs fan to say the least, but I respected them as the best organization in the game top-to-bottom: the best GM, best coach, best forwards, best defence, best goaltending, best farm team. They had it all, and Ken Dryden was absolutely no exception.

Dryden was notable for rising to the occasion. There would be games where Montreal would come out half asleep and the other guys would fire 10 shots in the first 10 minutes and only be ahead 1-0. The Montreal "faithful" would boo the team anytime it fell behind -- I heard it two or three diferent times in 1976-77, when they posted a home record of 33-1-6 !! -- but by the third period the score would be 5-2 Habs and Dryden would be leaning on his stick and the star pickers would be talking about the Flower and Big Bird and the Little M and whoever "led" the comeback.

The other thing Dryden was absolutely famous for was facing zero shots for an hour or so of real time and then suddenly being confronted with a turnover and a difficult chance and he'd rise to the occasion and the other guys would deflate even further. Danny Gallivan and Dick Irvin used to mention that a LOT.

Your GOA totals confirm that longevity is on Esposito's side, whereas Dryden and Parent were all about peak value. One way of measuring peak value which will make you scream goes like this: Dryden 6, Parent 2, Esposito 0, the ordering of which happens to agree with my ranking of the best goalies of the 1970s. I would still rank Tony 0 a little ahead of Gerry Cheevers, who like Parent won 2 Cups but did so as a good goalie on a great team. He was Chris Osgood with flair and a memorable mask.

Cheevers' legend was made when Boston couldn't win the Cup after he bolted to the WHA in 1972, but he was badly outplayed by Dryden in 1971 when Boston was the favourite, and didn't come close to returning the favour in 1977, 1978 or 1979. One of the reasons Montreal ALWAYS beat Boston was that they ALWAYS had better goaltending.

As for Ed Giacomin, he is a fraud. :D

The Contrarian Goaltender said...

Bruce: Thanks for the comments. I trust the numbers a lot, as you well know, but for this period I really don't. I don't think I can rule out the possibility that any of the guys on the list above were the 4th best of the era, other than most likely Giacomin, Crozier, and the Favell/Gilbert types who were usually the #2 goalies on very good teams.

I have an open question to you or anyone who was watching hockey back then: Do you remember any of the "bad-team" goalies (Meloche, Gary Smith, etc.), and whether there were any of them that stood out? Also, were there any middle-of-the-road teams that played very good defence, i.e. that era's equivalent of the Minnesota Wild? It looks to me like the Atlanta Flames might have fit that description, and I don't want to make the mistake of equating a low win total with a high shot quality against.

Bruce said...

Now back to our regularly-scheduled programming.

I trust the numbers a lot, as you well know, but for this period I really don't.

It's worth bearing in mind that there were 6 NHL teams in 1966-67, and eight seasons later in 1974-75 there were 32 "major league" teams. I think any attempt to derive reliable mathematical models from those years will simply confirm what we already know ... chaos. Data is useful, but handle with care.

I knew we were in serious trouble when the marquee said tonight's opponent was Rosaire Paiement and the Chicago Cougars. That said, at that point I was just happy Edmonton had a team in any significant league, but even on my most optimistic days it was pretty hard to visualize the Stanley Cup that was less than a decade away! Turnaround was a merger away, but the game sure bottomed out there, right around the time the B.S. Bullies were winning the Cup.

Statman said...

Poor Joe Daley facing 72 shots... if he was chosen 1st star, despite letting in 8 goals, it was out of sympathy!

Which is more indicative of his play that game?

GAA: 8.00
or
SV%: .889


League avg GAA: 3.12
[Avg GF of Boston 70-71: 399/78 = 5.13]
League avg SV% in 70-71: .903
[Avg SV% of Bos opponents in Bos-Opponent games: .874]



[Daley has/had a sports card store in Wpg, right across the street from the site of the deceased Arena.]

Bruce said...

Statman: By all accounts he was chosen 1st star cuz he deserved it. Received a standing ovation from the Bruins crowd. Daley made 40 stops through 40 minutes and kept the score at 2-2, until the Bruins poured another 30 on him in the third. You can be sure they were largely high-quality shots; those '70-71 Bruins were going for records and didn't let up in a game like this one. Obviously, given the 30 shots.

The only regular season game in my lifetime that featured more shots at one goalie was also in Boston Garden. This was The Ron Tugnutt Game, in which Tugger stopped 70 of 73 shots to preserve a 3-3 tie for the badly-overmatched Nordiques. I saw just the OT of that game (it was bonus coverage at the end of another game broadcast), in which Boston outshot Quebec 12-0. It was Tugnutt against the world. The last shot in the dying seconds was a right-on-target screamer from Ray Bourque off a feed from Adam Oates with Cam Neely practically sitting on Tugnutt and totally obscuring his vision, yet somehow the netminder snaked out his glove and made a mind-boggling grab. One of the greatest saves I've ever seen, and in an era of no loser point Tugnutt deserved nothing less. That time too the Boston crowd gave him a standing ovation at game's end, and the Boston players chosen as second and third stars actually went over and congratulated him. It gave me the chills to watch it.

Just don't get me started on Sam LoPresti ...

Martin said...

Don't know if you want a comment 3 years later, but.

I grew up in Montreal and though I was a kid I still remember Dryden's rookie year. I remember the crowd Booing if we went behind, and the scandal in the press when the Habs lost (which happened 8-10 times!) But Bruce told you about Dryden.

I remember feeling so sorry for Gilles Meloche. OMG, when the Habs played him we would get something like 50 shots against him... But made save after save on some AMAZING shots. It was so frustrating. Sometimes it would take 2 periods to get a goal past him.

The times I saw him play he seemed like a top goalie. Really great goalie on a terrible team.

Host PPH said...

It is quite impressive that they have recorded that amount of data even for that period of time.