Monday, May 18, 2009

Playoff Predictions -- Third Round Part Two

(4) Pittsburgh vs Carolina (6)

That the Penguins-Capitals series went seven games was hardly surprising. What was surprising was the degree in which the Penguins outplayed the Capitals.

The Penguins outplayed the Capitals in at least five of the seven games, and had the edge in shots in all seven. The aggregate shots for the series were 256-180 in favor of Pittsburgh. It's not often that one team manages to outshot the other so decidedly over the course of a best of seven series. When it does happen, there can be no doubt as to which of the two teams was better, especially when the outshooting team is also the victor.

The Capitals had no difficulty at all in outshooting the opposition over the course of the regular season, and the fact that they were so visibly outplayed calls for an explanation. I'm not sure whether Washington played extremely poorly or Pittsburgh played extremely well. It's likely that both are true to some degree. In any event, the Penguins were clearly the better team and deserved to win.

I've written about Boston at length on this blog and frequent readers are well acquainted with my thoughts on the team. At the risk of belaboring the point, I'll keep my comments brief.

The Bruins are a reasonably good team that were eliminated by another reasonably good team. Some people have commented that the Bruins picked the wrong time of year to play their worst hockey, but I don't think that's accurate at all. The Bruins were trading scoring chances with the opposition all season long, yet came out ahead on the basis of strong goaltending and luck. Against Carolina, the Bruins again traded scoring chances in an entertaining and evenly-matched affair. They did not play poorly. The difference was that, unlike during the regular season, they didn't receive an undue share of the bounces. It's as simple as that.

It's funny -- if you asked the average hockey fan which of the two Eastern series' went according to expectation, and which of the two did not, they'd likely identify Pittsburgh-Washington as the former, and Carolina-Boston as the latter. I couldn't disagree more.

As for the series in question, I'm forced to go with the Penguins. The Penguins, as I mentioned earlier, were very impressive against the Capitals. If they play even somewhat similar against the Hurricanes, they should advance. Carolina, having defeated two quality opponents, is obviously a good team, and I don't anticipate that this will be an easy series for the Penguins. At the same time, I don't see them losing.

Pens in 6.

12 comments:

Sunny Mehta said...

I have to say, I'm scratching my head a little about the Penguins.

ES shot numbers are some of the most persistent stats we have, and the Pens really seem to have defied that and played quite unpredictably.

Check out the Pens' 5-on-5 shot numbers...


2007-08 Regular Season:
.87 ratio (26.6 SF/60, 30.7 SA/60)

2008 Playoffs:
.94 ratio (27.1 SF/60, 28.9 SA/60)

2008-09 Regular Season:
.95 ratio (28.0 SF/60, 29.5 SA/60)

2009 Playoffs:
1.22 ratio (32.7 SF/60, 26.9 SA/60)


Notice how different the '09 playoff numbers are. And also note that they did it largely against a team that, up til that point, was dominant shots-wise!

My main questions are:

A) Why are their '09 Playoff numbers so different than '09 Regular Season?

B) Could we have predicted this somehow? (And if so, what's the pertinent evidence?)


Hopefully Vic and others will chime in too.

Jeff J said...

Since the trade deadline the Pens' shot rates have been well into the black:

http://battleofalberta.blogspot.com/2009/04/playoff-arrows.htmlAt the deadline they demoted Satan and added Guerin and Kunitz. Hard to believe those changes could explain the whole difference. Maybe the tail end of their schedule was soft?


But yeah, I've noticed the very same thing Sunny. Crosby seems to be the biggest culprit in terms of the disparity between shot counts and GF/GA. It's starting to look like a trend. Some theories I've mulled over are:
1/ Crosby's ankle sprain was still bothering him for much of the past year (A friend of mine suffered a bad sprain well before Crosby's and he's still not 100%).
2/ Crosby might be one of those rare birds whose shots for/against counts don't reflect the possession time or chance counts. People have suggested this might have been the case with Gretzky.

I don't put much stock into either, myself. It's a hard thing to reconcile.

Sunny Mehta said...

Yeah, Jeff, definitely if asked I would theorize something pretty similar: added players at the trade deadline, got players back from injury (Gonchar, Crosby sort of, etc), changed coaches, etc.

But still, they're posting REALLY different numbers here, and the team doesn't look all that different to my eye. So I'm hunting for more specific answers. (If player transactions were responsible, who specifically?) (If coaching change is responsible, what did he do specifically?)

And I too noticed that since Feb 1st the Pens' shot numbers have been trending up. (I think I even commented about it on this blog.) But I wonder, is there actually historical evidence in trends leading up to the playoffs having predictive value?

JLikens said...

I'm of the opinion that the coaching change has had a large affect upon the change in shot ratio. There are other factors, to be sure, but I think that there's some evidence that it's one of the main causal factors -- perhaps the primary one.

While I'm not certain as to the specifics of Therrien's coaching style, it does seem to have some negative effect on team shot ratio.

Consider this:

From 1999-00 until 2003-04, there wasn't much to choose between the five Canadiens teams that played during that time span in terms of talent and personnel.

The Canadiens employed three different coaches during the timeframe in question: Vigneault, Therrien, and Julien.

Here are the shot ratios for each coach:

Vigneault: 0.99 (102 GC, 26.8 SF/G. 27.2 SA/G)

Therrien: 0.8 (190 GC, 24.9 SF/G, 31.2 SA/G)

Julien: 0.94 (118 GC, 27.9 SF/G, 29.5 SA/G)

That's a fairly sizable difference considering that, on paper, team quality didn't really vary among the three coaches.

Sunny Mehta said...

Wow, that's interesting!

Bruce said...

Crosby might be one of those rare birds whose shots for/against counts don't reflect the possession time or chance counts. People have suggested this might have been the case with Gretzky.

Yeah, Oilers had some interesting shot count numbers during the Gretzky years. Quite the disparity between the regular season and playoffs, too. (These are all shots, not just at evens.)
Season : Reg. / Post
-------------------------
1983-84: 0.99 / 1.19
1984-85: 0.99 / 1.15
1985-86: 0.95 / 1.03
1986-87: 0.99 / 1.23
1987-88: 0.99 / 1.09
.

The Oilers' edge was in shooting percentage. Check out these crude PDO #'s (all situations) for those same 5 regular seasons, in which they never once collectively outshot their opposition:

Season : Sh % + Sv % = PDO #
----------------------------------
1983-84: .171 + .883 = 1.054
1984-85: .155 + .887 = 1.042
1985-86: .161 + .889 = 1.050
1986-87: .158 + .881 = 1.039
1987-88: .155 + .880 = 1.035
.

In the playoffs they had an outshooting advantage and largely maintained their edge in percentages.

Playoff: Sh % + Sv % = PDO #
----------------------------------
1983-84: .131 + .907 = 1.038
1984-85: .163 + .895 = 1.058
1985-86: .132 + .903 = 1.035
1986-87: .129 + .900 = 1.029
1987-88: .169 + .883 = 1.052
.

PDO's are slightly inflated cuz stats at my ready disposal include ENGF but not ENGA. Still the trend is pretty clear, Oilers had fantastic margins on percentages whether they were outshooting or not. Typically their defensive numbers, both raw GA and Sv%, were around league average, and the big edge came on offence which should surprise no one. Team shooting percentages over 15%, year after year. Astonishing.

Sunny Mehta said...

Bruce,

Those numbers definitely come across as gaudy.

I do wonder, though, how gaudy they would be if we saw the shot ratios and percentages broken down by ES, PP, and PK. Also it'd be interesting to compare them to league average for those seasons.

I gotta imagine that all the league average percentages were pretty damn different during those years than they are now. Also, I'd be really curious to see how the Oilers' shot ratios and shooting percentages compared to the league at ES. (Just 'cause the three game states are such different beasts, so it might be misleading to see their overall S% if it's largely being driven by special teams, particularly since there seems to be evidence that percentage disparity can be somewhat non-random on special teams.)

Is there anywhere we can find stats from those years broken down by ES/PP/PK?

Jeff J said...

Great catch on Therrien, jlikens.

Bruce said...

SunnyMehta: I always enjoy your comments, don't think I've ever responded to one directly.

I gotta imagine that all the league average percentages were pretty damn different during those years than they are now.Certainly there has been a huge change. Wade Gullison has recently posted these updated league average numbers to the files section of the Hockey Analysis Group:
Season : Sh % + Sv % = PDO #
1983-84: .129 + .873 = 1.002
1984-85: .127 + .874 = 1.001
1985-86: .128 + .874 = 1.002
1986-87: .122 + .880 = 1.002
1987-88: .122 + .880 = 1.002
---
2003-04: .092 + .911 = 1.003
2005-06: .101 + .901 = 1.002
2006-07: .097 + .905 = 1.002
2007-08: .094 + .909 = 1.003
2008-09: .094 + .908 = 1.002
.

... indicating that for the 5-year segments under review shooting percentage has dropped 3 full percentage points, from ~12.5% to ~9.5%.

The "bonus" PDO of ~.002 (and growing) is due to empty net goals counted for the shooter but not against the goalie. This value has been trending upwards as the dead puck era has delivered fewer normal goals and has resulted in more close games where pulling the goalie is de rigeur.

To expand on that, in 1983-88 a total of 541 ENG were scored in 4200 GP, 0.128 per GP or 1.68% of all goals. Whereas in 2003-09, 1014 ENG were scored in 6150 GP, 0.165 per GP and a whopping 2.94% of all goals.

I do wonder, though, how gaudy they would be if we saw the shot ratios and percentages broken down by ES, PP, and PK.

While I agree it would be interesting to have a breakdown by game situation, to be frank sometimes I think the Staterati put a little bit too much stock into situational stuff. Reductionism has its place to be sure, but holism works fine too, especially when it's the only data you got. How does the team do in the whole-game context across 60 minutes? How do they compare to other teams? If the difference is substantial, it can't just be hand-waved away to powerplays, especially given we do have data on powerplay opportunities and goals and percentages and pretty much everything except shots.

Certainly special teams shots can be hugely different in a single game, especially one where one team gets 8 powerplays and the other team 1. But over the course of a season those numbers fall into a fairly narrow range, with very few teams having a per game differential of as much as +/- 1 powerplay opportunity. So much (though certainly not all) of it will wash out over the longer term.

A good example can be seen in the above tables, in which marginal shooting percentages rose markedly in the post-lockout season, due in large part to a league-wide increase in powerplay goals of almost 50%. Yet shooting percentage as a whole went up by less than 1%. All those powerplays had an effect, I'd call it a Big effect on shooting and save percentages, but it's nonetheless constrained.

Since ~75% of the game is played at even strength, that situation is going to be very heavily represented in any data which includes all situations. While lesser teams might fluctuate around the mean, it is impossible to dominate in a category overall without dominating at evens.

The Oilers of that era were a particuarly exceptional example of domination in a particular category, namely shooting efficiency. Read on.

Bruce said...

Also it'd be interesting to compare them to league average for those seasons. Yes it would be, and is, interesting. The shots and percentage numbers are summarized on each team's season summary page on Hockey-Reference.com. Given the brute-force methodology involved, I studied just one season, 1986-87.

The Oilers scored 70 powerplay goals and allowed 74 (the league average was 72), so they were hardly fattening up the percentages with the man advantage. They were just beginning their offensive decline, scoring "just" 372 goals after five seasons of 400+.

"Gaudy" barely begins to describe the Oilers' advantage in shooting percentage. Team ** SOG *** G **** Sh%
-----------------------------------
EDM *** 2357 ** 372 ** 0.158
PHI *** 2409 ** 310 ** 0.129
LAK *** 2473 ** 318 ** 0.129
CAL *** 2528 ** 318 ** 0.126
BOS *** 2401 ** 301 ** 0.125
WAS *** 2274 ** 285 ** 0.125
NJD *** 2357 ** 293 ** 0.124
NYR *** 2472 ** 307 ** 0.124
MON *** 2238 ** 277 ** 0.124
PIT *** 2433 ** 297 ** 0.122
MIN *** 2454 ** 296 ** 0.121
STL *** 2347 ** 281 ** 0.120
DET *** 2173 ** 260 ** 0.120
HAR *** 2404 ** 287 ** 0.119
VAN *** 2376 ** 282 ** 0.119
TOR *** 2410 ** 286 ** 0.119
BUF *** 2426 ** 280 ** 0.115
WIN *** 2453 ** 279 ** 0.114
CHI *** 2561 ** 290 ** 0.113
NYI *** 2480 ** 279 ** 0.113
QUE *** 2385 ** 267 ** 0.112
.

Other than the Oilers' 15.8%, every other team is bunched between 11.2 to 12.9%. Every team in the league is +0.7% to -1.0% of the league average, except the Oilers who are +3.6%. The second place team in shooting percentage is Much closer to the bottom than to the top.

Now that's what I call an outlier.

The Oilers finished 16th in the league in shots on goal, yet scored 54 more goals than any other team. They were outshot by their opposition, yet their goal differential of +91 and their GF:GA ratio of 1.324 were the best in the NHL. For them it was a game of percentages far more than of outshooting.

I found some other interesting data from the chosen season on team PDO #, but this comment is already overlong. Maybe later, or maybe I'll post something on Oil Droppings.

Sunny Mehta said...

Fantastic stuff, Bruce. Both of those charts are awesome.

It's incredible how much of an outlier the Oilers were in terms of S%.

Makes you wonder if:

1) They got dominated territorially but used their shooting skill to score a lot when they did have attack zone time.

2) They actually did dominate opposing teams territorially, but it's just not captured by their shot ratio. (I've hinted before at the fact that, since we don't have Time On Attack as a stat, perhaps instead of using corsi, we could separate Shots Directed For and Shots Directed Against and use something like SDF/60 and SDA/60. I wonder if the Oilers had low shots directed at their own net but just weren't shoot happy themselves, indicating that they did have territorial dominance, but were just more selective with the shots they took.)

3) Their sick S% led to them having big leads often, thereby also influencing shot numbers. (I.e. - wonder what the shot numbers looked like w/game tied.)

Great stuff.

Bruce said...

Fantastic stuff, Bruce. Both of those charts are awesome.

Thanks, Sunny.

It's incredible how much of an outlier the Oilers were in terms of S%.

They were something else. It was my privilege to be an every-game season ticket holder throughout the dynasty years, which perhaps explains my appreciation of shot quality both for and against (arguably, and very loosely, described by PDO #) where the common wisdom among the Staterati seems to favour shot quantity (Corsi +/-).

The modern Wings are a great example of a team that dominates through outshooting. But historically there's more than one way to dominate, and those Oilers prove it. With sustain: seven consecutive seasons with a team Sh% of 15.5% or greater, averaging 15.92% for that stretch against a rest-of-league mean of 12.46%. Meaning that for 7 years running, an Oiler shot was about 28% more likely to find twine than a shot by the rest of the league.

Makes you wonder if:
1) They got dominated territorially but used their shooting skill to score a lot when they did have attack zone time
.

No, I would definitely not say they were ever dominated territorially. They did however, score off the rush more than any team I've ever seen. With their success on the counter attack, they didn't mind inviting their opponents into Oilers territory occasionally. :)

2) They actually did dominate opposing teams territorially, but it's just not captured by their shot ratio. (I've hinted before at the fact that, since we don't have Time On Attack as a stat, perhaps instead of using corsi, we could separate Shots Directed For and Shots Directed Against and use something like SDF/60 and SDA/60.)

I agree with this approach. It tells you more, in the same manner that plus/minus stats shown as +50/-30 or +85/-65 convey more info than +20.

I wonder if the Oilers had low shots directed at their own net ...

The Oilers were an average defensive team. In 1986-87, they finished 10th in both GA and SOGA, and 11th in Sv%. Doesn't get much more middle-of-the-pack than that in a 21-team loop.

... but just weren't shoot happy themselves, indicating that they did have territorial dominance, but were just more selective with the shots they took).

Certainly they were selective with their shots. They scored an astonishing number of goals into wide-open nets. A lot of this was the tendency to make the extra pass, the kind of stuff that drives modern fans nuts. Except these guys could make it work. Rather than, say, take a 25-cent chance directly on goal with maybe a second 25% chance to score on the rebound, they would make the 50% pass to set up the 80% shot. Or the 80% pass to set up the 50% shot. Not only was the success rate higher, the cost of failure was less likely to be registered as an opposition save.
3) Their sick S% led to them having big leads often, thereby also influencing shot numbers.

Bang on. The Oilers were absolute killers as front-runners. If the other guys were forced to open it up, look out. Thus the blowout factor was pretty high.

(I.e. - wonder what the shot numbers looked like w/game tied.)
.

The Contrarian Goaltender has done some great work in this area generally, with these two articles referring specifically to the 1980s Oilers. The comments section of the latter contains multiple observations to how the Oilers and Gretzky played the percentages.