I used the same code from the State of the Game thread, and found two interesting things.
- Shorter sets do make it easier for someone with an overall character disadvantage. In particular, comparing someone playing the 19XX meta with someone playing the 20XX meta, using the historical win rates and optimal counterpicks, you could expect the 19XX player to win 44.3% of Bo7 sets, and 46.6% of Bo3 sets. Intuitively, I think that’s because when you’re playing at a disadvantage, the shorter sets give less time for the games to trend towards the overall average, and instead the natural variance of the game tends to dominate.
- If you know that you’re going to be down a game, you’re better off with a longer set. In the same 19XX vs 20XX match, if 19XX is counterpicking in a Bo7, having lost the first match, then they have a 30.3% chance overall, but if they’re counterpicking in a Bo3, then they only have a 22.1% chance.
- In the hypothetical set posed by @CKR, where both players are playing 20XX, but @CKR has picked Zane to his opponents Troq, he’s got a 45.6% of winning a Bo7, but only a 41.6% of winning a Bo3.
The fun thing about all of this, then, is that it just sets us nicely back up into neutral and we can continue to debate the relative merits of set lengths, with extra arguments on both sides. 