When I'm wrong, I must admit that I'm wrong. Some time ago I published a post pointing out potential vulnerability of VIX settlement price to manipulation. As was recently pointed by a reader (and I contacted CBOE to confirm the details) I was incorrect in my understanding of settlement calculation of VIX index.
The reader pointed out that "bidding 0.05 for a 1 lot makes the option very unlikely to be included in the strip. Indeed, if you have a significant size of seller at 0.05, your bid will be hit and filled while no other bid would be left behind which would make the option a "0 bid" option, hence non included in the official strip.
You would be left like many traders before you wondering how they got filled on their bid thru the official rotation while this strike isn't included in the official strip."
CBOE documents state that the settlement price is calculated from all opening trades, and opening quotes for strikes where trades were not available. My initial understanding was that that zero bid filter applied only to regular everyday VIX calculation that is calculated from quotes, but not to settlement calculation that is based on trades. However the correct understanding that zero bid filter applies to settlement calculation as well. So e.g. in the case of some really low put strike, if you submit a nickel bid at the pre-open, it will get filled, but if the resulting opening quote has zero bid that strike will be ignored in the VIX settlement calculation.
My sincere apologies to all the readers, and thanks to WB for pointing out the error.