An excellent point! 20 games is not a sufficient sample size to draw a valid conclusion. Just like an individual who thinks he’s getting unfavored matchups does not have enough of a sample size to conclude that the matching system is rigged against him.
I agree. These are very unexpected results. Can you post that data in a google doc? I’d love to take a look. And I know there are definitely some stat guys here that would also like to review it. If the results are as you describe them and your methodology is sound (you’d probably need to provide a little more detail on that), then this is a very interesting result which would indicate that the mulligan is not completely random. I’m sure VS and HSReplay have a lot of data on mulligans and I’d be curious if they’ve done a similar analysis.
This is a little more vague, especially for a card like Brawl. If it’s one minion on each side, you’d expect a 50-50 result. But in many cases, it can be difficult to determine if it’s a “favorable” outcome. If your opponent has an 8/8 and you have an 8/8 and a 1/1 and the 1/1 survives, is that a “favorable” outcome? I think you’d need to provide quite a bit more context with this data set. But go ahead and put that data up as well; I’d still be curious to check it out.
Ok. But it would seem counter productive to spend all that effort creating a fake spreadsheet of data points and all the effort to craft a detailed methodology for testing, unless you know your claim is false and you have some stake in perpetuating that falsehood. But then you’re getting into the territory of that crazy Mr. Pillow dude and his wacky nonsense about the voting machines.