Did you see
this post comparing the ratio between $20 and $200 winners in that thread? Based on my sample size of 118 winners, the $200 prize should have drawn 12 times more than it did.
I saw it and I don't think that numbers that you counted mean something. There is no way how we can know real number of $20 and $200 winners. Users who post in that thread proof of winning get extra $5 bonus. Users who won $20 are more incentivised to post because $5 is significant bonus to their winning. I think that big number of $200 winners don't bother to sign up on forum to upload proof just to get five bucks. And when people win bigger amount of money, not everyone want show it in public and attract attentiin to themselves. Offcourse, it's just my assumptions without knowing real number of $20 or $200 winners. There was also complaints about their lottery, that there was only few users that posted proofs that they won. But if I would win over 2 BTC one day, I'm not sure that I would want to post about it in public to get few bucks as a bonus.
I don't claim to understand how their particular hash generations work but it seems to me like you're right in that you can't use forum posts as a statistically valid way of counting winners. It is only potentially indicative of something and doesn't constitute as proof of anything.