I have come across on multiple market blogs (here is one example on MarketSci) that the VIX displays seasonality by weekday. Basically, it does what one expects: best average returns on Monday and worst average return on Friday (although the effect is shrinking apparently). This makes sense because everyone saves up their bad news for the weekend and often during the summer people are gone on Friday,so it is very quiet. However, the VIX is not directly tradable. I wondered if the same effect applies to VIX futures - specifically the VXX ETN which tracks the daily movement of the front and second month VIX future contracts. Here is the results of a very brief analysis of averaging the daily return of the VXX (close to close) by weekday over the life of the ETN (1/30/10 - 04/05/12).
Note that we actually see the opposite effect. The best day is on Thursday close to Friday close, and the worst day is Friday close to Monday close. There are a couple caveats with this. First, VXX has only been trading since the start of 2009, the top of the largest VIX spike on record. We really need more data from future contracts to be more certain of this effect (and it also explains partly why we see a lot of negative returns here). Second, we really should break down the analysis by bear and bull market because I can imagine that we would the original seasonality during bear market periods.
That being said, it is clear that shorting VXX on Friday appeared to be good on average during the last two years. My speculation is that during this period people were expecting bad news to be released so they bought up VIX futures into the close on Friday as a hedge, when that didn't pan out (or wasn't as bad as expected) people unwound their positions.