Another thought is when you buy a strip you're buying one call and buying two putts at the money so first of all the vix is based on the entire series so you're buying a future on future implieds of the entire series... You would have a changing risk with respect to strike risk that wouldn't show up in the future. Hence the differential... Just wildly guessing
TLDR: there is no static replication of a vol swap*. You need to re-read the VIX white paper. Well your expiries are mis-matched by 30 days, as the vix SOQ is calculated on options expiring 30 days later, but that is a minor point, Even if you chose the correct expiry for your strip, and ignoring minor idiosyncrasies of VIX construction, you still would not be replicating VIX p/l because VIX is a square root contract the payoff of which is not a linear function of var. You've got a var hedge but a vol* p/l. * the VIX is not exactly the fair price of a vol swap, more the price of a var swap expressed in vol terms (as vol = sqrt(var))
LOL Even disregarding the technical details (i.e. convexity adjustment, strike inclusion etc, read Kevin's post), the option strip would include the period leading to VIX expiration, while VIX futures is from futures expiration to options expiration. So you would have a bunch of gamma/theta to deal with.