Unconventional Wisdom - West Coast Bias

A few weeks back when the Pittsburgh Penguins were going through a rough patch and people were panicking and ready to write off the whole season we looked at how important possession was to playoff success. At the time I was looking at the list and noticing what appeared to be some trends in the data with a fairly prominent split in the East vs West numbers. So I wanted to split them up and run the numbers to see if there is indeed a difference.

Averages

The first thing we want to do is compare the average range of the different metrics to see if there is a difference on the whole between East and West teams. We have 6 years worth of data which gives us 90 different teams in each conference.

Goals For Percent in the West the mean in .502 with an average range of 1 standard deviation between .457 and .547. In the East the mean is .498 with an average range between.450 and .546. So the East and West are very close, with an ever so slight edge to the West in mean and the high end of average. However, the East appears to have lower lows. Perhaps this is indicative of the top teams in the East and West being comparable, but the fringe teams being more competitive in the West.

Fenwick For Percent in the West the mean is .501 with an average range of 1 standard deviation between .466 and .536. In the East the mean is .499 with an average range between .470 and .527. Again we have a very close mean with an ever so slight edge to the West, but the East has a much tighter spread since the West sees higher high and lower lows. As has been suggested, the West coast focus is on driving the flow of play.

Corsi For Percent in the West the mean is .502 with an average range of 1 standard deviation between .465 and .538. In the East the mean is .498 with an average range between .470 and .527. Once again we see the same tight spread in the East and an even larger margin favoring the West in possession, higher highs and lower lows than we even saw in Fenwick.

Correlation - GF% vs FF% and CF%

So now that we see there is a slight difference in the averages between the East and the West, let's take a look to see if there is any difference in how important dominating possession is to dominating the scoreboard.

GF% to FF% R²

West - .4396

East - .1942

GF% to CF% R²

West - .4622

East - .1844

That is incredibly suggestive. In the West there is a fairly suggestive moderate correlation between puck possession and goal scoring, with the teams that control the flow of play more likely to be the ones that also dominate the scoreboard. In the East, however, there is just a weak correlation. What is more, it is less than half of what it is in the West.

It is also interesting to see that, while it is a fairly small difference, Corsi is more important in the West while Fenwick is more important in the East. This could be a product of play style in the East, the stereotypical rough and tumble gritty defense relies on blocking shots, so getting out possessed is less of an issue if you aren't even letting the puck reach the net. In the West, however, it is more about keeping the puck away from the opponent and winning the possession battle.

Correlation - Points Percent vs GF%, FF%, and CF%

Looking at P% gives us a proxy for which teams make the playoffs, since the higher the number of Points earned the more likely one is to qualify for the playoffs. It also allows us to see what works in the regular season.

P% to GF% R²

West - .7180

East - .5987

P% to FF% R²

West - .4589

East - .2077

P% to CF% R²

West - .4660

East - .1932

It is interesting to see that GF% is even more important in the West. However, as we saw above there is a much higher correlation between possession and scoring in the West, so that could have something to do with the differences. In the West there is a strong correlation while in the East there is a moderate correlation, although it is on the high end of moderate. That doesn't really tell us anything we didn't know though, since the teams that outscore their opponents are the ones that tend to win more games.

Possession though we see a definitive East-West divide. The West has a moderate correlation between winning and possession, and once again an ever so slight edge to Corsi over Fenwick. However in the East possession is far less important, a weak correlation and once again less than half of what it is in the West, and once again an ever so slight edge in Fenwick over Corsi. This is quite suggestive, leading us to believe that there is indeed a difference not only in the way the game is played in the East and West but which strategies lead to more success.

Correlation - Playoff Success vs P%, GF%, FF%, and CF%

Our first look at using the regular season numbers as a predictive value, as all the other correlations were just telling us what works during the same season. So how do the various metrics stack up when looking to see how teams perform in the post-season? So we are taking a look at how deep teams make it, with a 0 for teams that failed to make the playoffs and a 5 for those that won the Cup.

Round to P% R²

West - .5200

East - .3939

Round to GF% R²

West - .4110

East - .1950

Round to FF% R²

West - .4460

East - .0937

Round to CF% R²

West - .4352

East - .0635

Once again we see quite a clear split between the East and West. The West has a moderate correlation in all 4 metrics, with possession now being more important than simply being better able to outscore your opponents. Interestingly though, P% is even higher. If I had to wager a guess, I would say that it is because being able to win in the regular season means you not only excel at whichever of these other metrics is more important but also all the other aspects of the game, some of which are quite unquantifiable.

In the East there is a moderate correlation with P%, although much lower than it is in the West. Again, a team that wins more often in the regular season tends to possess the qualities of a successful team that are unquantifiable, so it stands to reason that some of those little things are indeed important when it comes to making your way through the early rounds. GF% is a weak correlation, less than half of what it is in the West. However, it is vastly higher than the possession metrics, which are so low as to be completely irrelevant to teams in the East.

Looking Deeper - Round by Round

Those results shouldn't be overly shocking if you look back over the past playoffs with an eye towards who plays in which conference. A poor possession team isn't going to win the Cup, but in the East where possession is less important they can still make a deep playoff run with solid defense and goaltending or a deadly accurate offense, preferably both.

In the East 6 teams with poor possession have made the playoffs, 4 of which made the 2nd round, 3 that made it to the 3rd round, and just 1 that made it to the Cup finals. Over in the West, however, only 3 teams with poor possession have made the playoffs and only 1 of them made it into the 2nd round. Sure, that's not a huge difference, but it does go to show that when you are squaring off against other East teams in the early rounds you can get by even if you don't win the possession battle. But even more compelling is looking at the difference of good possession teams.

In the West 16 good possession teams made the playoffs, and none failed to make the playoffs. 11 of those made it to the 2nd round, 8 to the 3rd round, 5 to the Cup finals, and 4 won the Cup. Over in the East 16 good possession teams made the playoffs, and 1 failed to make the playoffs. 9 of those made it to the 2nd round, 3 to the 3rd round, just 1 to the Cup finals, and none won the Cup.

So in the West you want to be a good possession team above all else. But in the East, while you don't want to be a poor possession team, you can have just as much success being an average possession team that excels in other facets of the game. But that is just a general observation, let's see what the numbers say.

1st Round to P% R²

West - .2546

East - .0088

1st Round to GF% R²

West - .1305

East - .0277

1st Round to FF% R²

West - .2904

East - -.0084

1st Round to CF% R²

West - .2710

East - -.0277

That is quite an interesting difference, a huge gap between the East and the West when we are looking at the 48 teams from each Conference that have made the playoffs over the past 6 seasons. There is a weak correlation between any of the metrics and playoff victory, but in the West it is at least on the higher end of weak. And once again we see that possession is more important that goal scoring in the West, in fact it is nearly 3 times as high, although again a weak correlation is not overly suggestive.

It is interesting though, as is the fact that in the West P% is quite important as well, but as noted above that could be due to the fact that teams that finish high in Points tend to have the qualities that are conducive to victory. 3 of the past 6 President's Trophy winners were from the West and went on to make the Cup finals, two of whom ended up winning the Cup.

In the East, however, the correlation is so low as to be practically non-existent. There is a much larger emphasis on goals over possession though, and we see that strangely enough there is a negative correlation with possession in the East. But as I said, the correlation is so low as to be completely meaningless, so while it is interesting it is not suggestive. It would seem that everything is random, upsets occur with some frequency, and anything can happen.

2nd Round to P% R²

West - .3751

East - -.0023

2nd Round to GF% R²

West - .2189

East - .0070

2nd Round to FF% R²

West - .3640

East - -.0442

2nd Round to CF% R²

West - .3751

East - -.0623

Now we are down to 24 teams from each conference that have reached the 2nd round over the past 6 seasons. In the West we now see a moderate correlation with everything but GF%, although it is certainly on the low end of moderate, but at least a moderate correlation is somewhat suggestive. We again see that possession is more important than scoring, and Corsi moves ahead to be more important than Fenwick. However, GF% did become markedly more important than it was in the 1st round, now being less than half that of possession. Although oddly enough P% is equally as important. So it would seem that when trying to determine who makes it into the 2nd round its best to lean towards the high possession teams.

In the East though we still see that it is completely wide open, sometimes the better team loses, a concept we are quite familiar with here in Pittsburgh. GF% becomes even less important than it was in the 1st round, while the possession metrics becomes even further negative. They are still low enough to be basically meaningless noise, but it is still interesting to see that possession has a negative impact of playoff success for teams in the East.

3rd Round to P% R²

West - .2429

East - -.0099

3rd Round to GF% R²

West - .4329

East - .0878

3rd Round to FF% R²

West - .3495

East - .0112

3rd Round to CF% R²

West - .3005

East - .0256

At this point we are down to 12 teams from each conference that have reached the Conference finals over the past 6 seasons. These are the truly elite of the elite, if you are good enough to make it this far you are one of the best teams in the league. In the West we see that by the time you are one of the final four teams vying for the Conference championships the teams are much more evenly matched, so the team that is better able to outscore their opponent gets a slight edge. GF% and the possession metrics are a moderate correlation, although both possession metrics are on the low end of moderate, while P% is the high end of a weak correlation. Possession is still important in the West, so a poor possession team isn't going to win, but it is now much less important than it was in the 2nd round. And P% is now nearly half that of GF%.

In the East everything is still a weak correlation and the numbers are low enough to be meaningless. P% is a negative correlation again, but possession moved back up to being a positive correlation. Although it is nearly 1/4 that of GF%. Again the results are interesting, but not at all suggestive. The East is just a free for all, and there must be some other factors involved that are better able to capture why some teams win while others lose.

4th Round to P% R²

West - .0001

East - .0157

4th Round to GF% R²

West - .0227

East - .2848

4th Round to FF% R²

West - .0807

East - -.0050

4th Round to CF% R²

West - .0221

East - -.0022

We are now down to just 6 teams from each Conference, so there are some serious sample size concerns here, which could explain why the West now has no relevant data available. The correlations are weak in every single metric, and low enough to be utterly meaningless. While not suggestive, it is still interesting to see that Fenwick is nearly 4 times as high as GF% and Corsi, so there is still evidence that suggests possession is important in the West.

In the East though we still have utterly meaningless weak correlations in P% and possession, but now things flop and P% is a positive correlation while possession again becomes a negative. But again, the numbers are so low as to be utterly pointless. The GF%, however, while it is still a weak correlation is now on the high end on weak. So it is somewhat suggestive that outscoring your opponent is more important than outpossessing your opponent in the East. But when we get into the Cup finals we actually have teams from both Conferences again, so let's go back to the numbers we found last time when combining the data.

4th Round to P% R² - .0524

4th Round to GF% R² - .2182

4th Round to FF% R² - .1019

4th Round to CF% R² - .0770

So once we get down to the Cup finals we see that there is a weak correlation with everything again. The P% and even CF% are low enough to really be meaningless, but we do see somewhat suggestive numbers nonetheless. The GF% numbers are twice as high as the Fenwick numbers, suggesting that once we get to the Cup simply being a good possession team is not enough to win. Again, being a poor possession team is not good enough, but an average possession team has just as much of a chance as a high possession team. The other factors play too big of a role, unfortunate bounces, hot goaltending, solid D.

I imagine that the focus on Corsi as the preferred metric arose back in the early days of its existence, since the data available came from the time when Detroit, Chicago, and San Jose were the dominant teams in the league and had plenty of regular season and playoff success. But as we see possession is much more important to teams in the West than it is to teams in the East, and the West coast bias remains prominent because 4 of the past 6 Cup winners came from the West. However, it is important to keep context in mind when looking at the numbers rather than just blindly assuming a team will lose because they do not dominate the flow of the game.