We have previously reported that rating appears to be either stagnant or deflating on the 2v2 ladder. With patch 9.0.5, Blizzard introduced a change to the 2v2 bracket which will supposedly cause rating inflation in the bracket over time. The patch notes were somewhat light on details, so here we present an analysis that shows what's changed with 2v2 rating.
The way we analyse this is to look at matchup data (REFlex) gathered from the last week. We're looking at matches played in the week leading up to 9.0.5 and matches played in the week after.
The following plot shows what the rating change distribution looks like for games where the player won. We exclude extreme rating change values from this analysis.
And here is the distribution for rating lost in 2v2.
It's not obvious what exactly has changed from just looking at these plots. However, there does appear to be a bigger difference for rating gained before and after the patch compared to with rating lost. Small differences here are to be expected: Blizzard announced that these are small changes that should gradually increase the 2v2 rating. If we plot the mean of rating gained following a win for the two patches, we get a very clear result.
The average rating gained has increased by approximately 0.3 rating per match following the 9.0.5 patch. Note that we have limited data for this analysis (just over 300 matches, yielding around 1200 observations), so the 0.3 figure has a fair bit of uncertainty attached to it (i.e. it could be 0.2 or 0.4), but there is enough data to say that this is a significant effect: you now gain more rating on average than you did previously.
If we now look at the average rating lost following a loss, we observe the following result:
There is no difference at all. In other words, what has happened following 9.0.5 is that people now gain around 0.2-0.4 more rating per match, and they lose the same amount of rating as they did before. Over time, this should lead to rating inflation in the 2v2 bracket.