Inspired by this article I decided to see if there was any intersting analysis that could be done.
Looking at the past two normal (non-winner's league) proleague rounds, the number of 3-0 wins is 12 (18%) for round 1 and 16 (24%) for round 2. I combined them for a sample of 21% (28/132). Doing a basic hypothesis test, under the assumption that .21 is an accurate predictor for the future, the z-value ends up being around 3.4. This makes the probability of getting this result by chance somewhere near a tenth of a percent. Using that data, it seems that we can say that the new maps have influenced the result and the coaches who won 3:0 are right in their gloating.
Thats probably not a good conclusion though. Taking the wind out of that argument's sails is Samsung's performance: they won 3-0 against Woongjin Stars and then lost 0-3 to Hite Sparkyz. Its especially damning because statistically Stars and Sparkyz should have roughly equal strength vs Samsung as a whole. Feeling frisky, I ran the numbers again after having throwing out one and then both of those matches, and the probability hits just above 1% in the second case. Still a strong number, but the sample size becomes so small its hardly trustworthy data (in a case where the "good" sample size is still way too small).
The more interesting thing I found doing this is the distribution of match lengths.
In rounds 1 and 2 of the current season, just over half the matches went to the ace match. This means that the skill difference between teams looks like its pretty small (even though final standings might not reflect it), but more importantly, shows just how necessary having a good ace player is. Oz and Samsung take many of matches to ace (45% for both of them between rounds 1 and 2) and were able to stay above the pack for the duration of that period by having ace players in the 60-70% winrate.
For more evidence as to just how close the teams are in wins, a distribution of who holds the most embarassing losses
Unsurprisingly, ACE is the biggest contributer to the number of matches that only last 3 games. With a single team having 30% of the losses that end before set 4, it seems that the skill gap between teams 1-11 is even smaller than the match length chart indicates.
Ultimately, there aren't any strong conclusions that can be made, as the sample size is small and I'm using methods you might find in a 9th grade stats class, but some of the numbers are fascinating. I think there still might be a way to try and test the idea that some teams are likely to do better on a new map pool, but I haven't decided on any good metrics of measurement yet. The maps are the start of this season weren't new so the data set to work with will stay impossibly small for the forseable future.
I am not ready to throw out the idea that skill gaps change as the metagame developes, but I can't say it DOES happen with a ton of confidence. Tenatively, I want to say that there is probably a (larger than normal) difference between teams right now, and this week we are likely to see a slightly higher percentage of short matches with the numbers normalizing the week after.
An interesting aside:
+ Show Spoiler +
I played around with some probability distributions, and if we assume each player is exactly 50% against each other player, then the chance of getting 7 3-0s in 10 matches (first week) is .000038. If we assume 65% (which I think is near what we would find if we could quantify it) then it is still .00029, both of which are considerably less than a tenth of a percent. There is absolutely nothing that can be said from getting those numbers of course, the assumptions are wide and the methods used are rough at best.