Novae explode without warning, and even recurrent novae are pretty unpredictable. (For instance, we've been waiting way too long for T Pyxidis to erupt again
But here's a prediction about a recurrent nova that should go off in a specific year: 2009. I've calculated that the recurrent nova U Scorpii, north of Antares and east of the head of Scorpius, should explode any month now. My "crystal ball" is based on old archival photographs and data from amateur astronomers. This brings a golden opportunity for amateurs and professionals to catch the early hours of a nova eruption and to prepare in advance for an intensive observing campaign.
"Recurrent novae" are those that have had more than one eruption observed within a century or so. Classical novae too are believed to repeat, but on timescales of perhaps 1,000 to 100,000 years; we just haven't been watching long enough to see them do it yet.
U Sco is one of the most famous recurrent novae. In quiescence it usually hovers around magnitude 17.6, but in outburst it briefly shoots up by about 8 or 9 magnitudes. In fact, U Sco is the all-time fastest nova known. Its brightening from minimum to maximum takes only about 5 hours, and its decline to two magnitudes below maximum takes about 38 hours. U Sco's last three eruptions, in 1979, 1987, and 1999, were all discovered by amateur astronomers.