I have been thinking a lot about calendars lately and why they make me uncomfortable. At a basic level, you are betting that the underlying's price will not move, or not move much.
Ever since 1998 when LTCM crashed, the frequency of large, unpredictable moves in the market has increased. This is even more pronounced in the post 2008 world (think May 2010 flash crash).
What I like about straddles/strangles, is that we are playing a solid strategy of getting paid off big if IV spikes as we hope, and we don't really lose too much if the spike doesn't really occur.
The straddles also have a HUGE added bonus. If the stock moves big, either way, we also win. As a very recent example, the first JOY straddle opened a few days ago provided about an 8% return the very next day as the stock jumped to right around $57.50.
Now, back to the calendars, if an underlying has a big move, for ANY reason, you lose. While I cannot prove it, I would be surprised if the positive vega of the calendar can outweigh the large move from a crash. And with all of the program/high frequency trading going on, and the fact that nobody really knows how they interact with one and another, the risk of these types of crashes is higher.
For this reason, I get nervous when I see a calendar spread. You can do all of the correct research in the world, and then some unexpected problem comes up, and boom, you lose.