It seems that given the large number of base items that can drop, it would be expected in a random situation that one or more of those items would be seen to drop more often than others during any given period of time and that this more common item would be a different base item in the next period of time. That (as Fabian and Grape have mentioned) is the nature of randomness.
From anecdotal data from the MFO, sometimes the period of time for any given common item is a few hours, sometimes a 'day' and in some cases it was a trend seen across the entire period of the MFO (so the period of time is at least 1 week). This supports the idea that it is a random effect, because if it was a fixed pattern then everyone should be experiencing the same period of time. (Although I guess it could be related to some variable other than time, such as # of sessions or # of saves, etc...)
Also a 'day' as the time period is rather problematic, firstly because it requires some interaction with your computer's clock (although I believe it has been shown that in the past D2 did interact with the clock in one specific circumstance), secondly because a day is really very arbitrary. Is it midnight to midnight, or from when you wake up to when you go to sleep? If it is midnight to midnight, then it should be easy to test. Start running a couple of hours before midnight and identify your common item. Keep running after midnight and you should now have a different item as your common item. Of course you would need to keep frequency data on all item drops both before and after midnight to prove that it wasn't observer bias and run the test over quite a few 'days' to prove the effect.
To be honest, it smells a lot like aromatherapy.
