February 20, 2008

Western Domination of the Eastern Conference.

The Record (W-L-T)
There always seems to be a few stories about how good the West is compared to the East in the few cross conference games we see each season. Since the new schedule of change , there have been fewer than 400 cross-conference game. Below I have the data I consider important about these games. The first table show the record for the western teams vs. the eastern conferences split into two groups home (H) and away (A). I've included the west vs. west figures for one season in the row below for comparison.

Some people have suggested the domination is a result of the eastern teams playing in a different time zones and traveling farther than normal, but as you can see from the table above the western teams do only 3% better at home and 9% on the road compared to the normal home-away advantage. Also worth noting, the home-away advantage is smaller for the inter conference games than the west vs. east games.

The statistics

The statistic that best correlates with winning is goal differential, but there are several ways to get to the same goal differential. Below is a table that presents what I consider the most important statistics in hockey for these west vs. east games. Like above, the table is split into home/away games.

[S = Shots, G = Goals, EG = Expected goals (based on shots on goals), SQN = shot quality neutral save percentage]

The most interesting portion of this table is the fact that the east is actually significantly better at getting short handed goals, but these goals seem to costing them power play goals against. It would appear that the east is encouraging offense at the cost of the powerplay goals against. It's hard to measure if this strategy is paying off though.

At home the west is getting 27% more expected goals against the east (15% more shots)

Some people have mentioned that goaltending is better in the west. This does not appear to be the case either as on average the east actually has better save percentage numbers.

None of the data can really clarify whether the west is better because of it's offense or it's defense, but it's worth noting the if the west played the east every game they'd have 9% more goals and if the east played the west every game they'd have 9% fewer goals! So if they split the season evenly the east would be down by about 5% and the west up by about 5%.

Note: The data is a couple weeks old.


Anonymous said...

Interesting post. I love that you use Expected Goals a lot in your posts and on your website. I think it's the way of the future.

Having said that, expected goals is so heavily dependent on the shot quality model, that it's crucial to keep trying to perfect it imo.

It would be cool if you did a blog post on exactly how your shot quality model works.

I see certain teams that significantly under- or over-perform their Expected Goals. (Red Wings, Rangers, Blue Jackets, etc) It makes me wonder what it's due to.

It could be that the difference isn't as significant as I think. Or it could be just plain luck/variance. Or it could be something wrong with the shot quality model. Or it could be mistakes in data collection. (As for the latter, I do know about the Rangers' unreliable home shot records, for example.)

One big data collection mistake in shot quality recording that I've noticed is when a team scores on a deflection by the other team. For example, Lidstrom shoots the puck from 60 feet and it deflects off of Pronger (who's standing 5 feet in front of the net) and into the goal. The shot gets recorded as "GOAL - Lidstron - slap shot - 60ft", which is HIGHLY incorrect.

I'm not sure how significantly that affects the numbers over large sample sizes, but to my selective memory I sure seem to see a lot of goals get scored that way (and recorded incorrectly).

Anonymous said...

Oh also, to add to my last comment, there's one other thing I forgot to mention...

Another possible factor in a team under- or over-performing its expected goals is goaltenders faced.

I'm not sure how significant that is though.

But in the Atlantic Division (the division with the best goaltending), four out of the five teams have underperformed their ExG so far this season. (Interestingly, Philly has overperformed it. Not sure why.)

JavaGeek said...

Over/Under performing: Standard variation (one standard deviation) is equal to the square root of shots faced multiplied by 0.09. This is the natural variation one should see in the data. Expected goals should not equal actual goals all the time. Detroit has taken 2129 shots, so their standard deviation is sqrt(2129*0.09) = 13.8. They’re actually a +37 off expected right now, so that’s about 2.68 standard deviations (2.68*13.8 = 37) away from expected. There’s a 0.4% chance of that happening. If it were just luck what is happening to Detroit should happen once every 8 years (to one of the 30 teams). It’s more likely that there’s a mistake in the data, but it is quite possible a good chunk of the problem is simply luck. Either way it takes a lot to figure out what the true problem is.

Rangers and Red Wings are on the negative end (fewer goals then expected)
While the Blues, Lightening, Stars and Wild are on the positive end (more goals then expected).

Scorers at different arenas look at shots differently a 20’ shot in Vancouver is considered to be a 16’ shot in New York. I have tried to make adjustments based on the past two seasons, but it is difficult, because these factors can change without any notice.

In terms of deflections: unless there is evidence to show a team deliberately uses deflections to score it will just appear as a random variation in the data, but it should increase the variation in the data.

Yes goaltenders should have a small effect, but it would produce the results you see.

Anonymous said...

Interesting post.

The West has been the superior conference for quite some time -- the West has had the better interconference record every year since 1999-00, and the difference has been large every year except in 2003-04, where the West was 119-118-(x number of ties) aagainst the East. The East was the better conference from 1993-94 until 1998-99, save for 1997-98, a year when the West was slightly better, and 1995, a year which featured no interconference games.

I'm not exactly sure what causes the disparity in strength in conferences, although the fact that the East was better for the first 6 years with the West being better for the last years leads me to believe that the process is cyclical to some degree.

Having said that, it's not as if the difference in strength is shrinking. The gap was very large from 2000-2003, but had all but disappeared by the 2003-04 season. However, it was fairly large in the first two post-lockout seasons and appears to be even larger this year. The fact that the gap has increased in the post-lockout ear is not consistent with the existence of a cyclical process.

T.C. said...

Yet, since 1999-2000, the East holds a slight edge with four Stanley Cup champions while the West has won three times.

It reminds me of MLB and the NFL. One league of conference dominates the regular season but in post season play it tends to equalize. Anyone remember the St. Louis Cardinals a couple of years back? 83 lousy wins and they knocked out the Detroit Tigers.

More recently, who would have thought an NFL team would have won the Super Bowl? Yet, that's what happened with the New York Giants beating the New England Patriots.

That one annoyed me!

T.C. said...

I meant to say "one league OR conference." Sorry.

JavaGeek said...

RE: the commentator

Playoff domination by west

T.C. said...


Yes, what you posted is true but hey sports are so fluid right? I use a similar table for Champions League soccer. Soccer employs a two-leg system. The aggregate score of both determines who wins. However, it is possible to win (thanks to the concept of away goals) by splitting the series.

Just another stat in the on-going debate as to which league dominates soccer.

Let me further explain to use the soccer analogy which you're sort of on the same path with.

The Italian soccer league has won more total trophies (for selected five tournaments) and reached more finals than any other league. With Spain and England trailing. However, Spanish clubs have a better head to head record against its English and Italian counterparts.

Furthermore, soccer calculates a coefficient factor (which includes all games not just finals) to determine successes and the strength of leagues. It's not perfect but in this way, Spain leads with Italy second and England third.

Of course, these are just stats. It does not consider the subjective nature of each league.

That's why we watch!

A hard realist will counter by saying that all that matters is who hoists the Cup at the end of the year.

JavaGeek said...

And if you go back 8 years instead of 5, Spain has won more championships and made it to the final more often (5 vs. 4). Go back 9 and Italy and England tie for second behind Spain. And one more year back and Spain defeats Italy in the final!

Also, try flipping a coin five times. Now what side occurred more often? should I now conclude that that side is more likely to occur than other side?

T.C. said...

I'm assuming you are considering just Champions Leagues. Since '98 La Liga placed six teams in the final to Serie A's 5 with Spain winning four to Italy's 2- including that Real/Juve final you allude to. Although, Juve eliminated Real on two other occasions in the knock out phase.

No doubt I used an "alltime" or overall aspect for what is acknowledged major tourneys: Champions, UEFA, Super Cup, World Club and the defunct Cup Winners'. If I would break them down in 5 or 10 year patterns the comment post would be too long! That's why I mentioned the coefficients. The breakdowns since 1950 are interesting indeed. These things tend to work in cycles but there is no doubt who the top three leagues are - strictly speaking regarding tournament play. Domestically it's another ball of wax.

The first decade in the 21st century belongs to Spain. Since 2000 Spain has had more teams in the finals in the major tournaments than England and Italy. However, I'm beginning to see a slight shift. But that's for another time.

Enough of me. This is a socc - erm hockey blog.

As for the coin, I plead ignorance. ;<)