I oversimplified it, but I think you read it wrong a little too. Paper would be the one that is 50% favorable, 25% poor, 25% tossup to start. Think you meant paper in the first sentence, and rock in the second.
However, when you run it out, the rocks start to thin out a bit from beating each other and losing to paper. So, after the first 2-3 rounds, paper becomes a larger percentage of the field and then scissors is wrecking them. So, if you were a good player that can just out-skill his opponents for the first few rounds, and dodge your counter, other things hate out your counter deck for you and Rock becomes the best deck again. I was envisioning this as a 12 round tournament, and applying some of the things I've learned internally, but I didn't spell it out well.
Yes, in the very first round, paper has the best overall matchups in that example. It beats half the field, is 50/50 with 25% and only loses to the other 25%. But, the decks that have the best chance of having a X-3 record, or better, after 12 rounds are actually rock and scissors. Also, I was thinking that you'd play a rock (shaman) deck that can potentially gain up to 5% in the mirror, which again helps your chances dramatically.
Again, this is ignoring a lot of other variables, as no deck is 100% against any other deck. If you can find a deck that is favorable against shaman while still being close to 50% against everything else, then that would become the ideal deck to play, for sure. I was just saying that in the hypothetical situation of a deck beating shaman but losing to everything else, it would be a very bad idea to use that to try and go 12-x because while shaman is popular, it would need to be a ridiculously huge amount of the meta for that to work.