NYMEX CL, or the oil price that is in the yahoo box in the right. We have a new poll a) when 60 days have expired without hitting a boundary from the previous poll, or b) when we hit a 10% boundary. We did not hit 154 in NYMEX CLQ/U08 at any point, but we did go through 127 today briefly, therefore we hit a 10% boundary from the previous poll.

Regular intervals would be just as, if not more arbitrary; at least with these conditions, we have conditions for expiry. If oil goes through 114 tomorrow, then we will have another poll tomorrow night. If nothing happens in the next 60 days, then we have another poll then.

A Poll is a "sampling"exercise, and I agree with Kiashu that it would preferable to sample on a regular basis ..... and otherwise to sorta pay tribute to the sampling theorem in organizing the interval not only to be regular(ized), but also to be sufficiently fine to display the underlying process you want to elucidate ....
so as to avoid alias distortion.

I can assure you that there are few polling agencies out there who consistently use "time" as the basis of putting multiple polls into the field, even if they end up being non-random or event driven. If anything, the method I am employing here is more valid than what you propose because it attempts to capture the motivation/momentum after large moves in the market if they occur. Otherwise, it is exactly the design you propose with an interval, mine just happens to be 60 days instead of 30.