As another commenter below astutely pointed out, with a mean time between recessions of 4.7 years you could blindly make this prediction every year and be right more often than not (with a 3 year window).

Um, how do you get 4.7 years as the mean time between recessions?

As someone elsewhere notes, since the 60s we've had a recession every 5 to 10 years. Were there two recessions a year apart that I missed?

EDIT: Ah, the other commenter supplied it as the mean length of an economic cycle, measured peak to peak or trough to trough.

But a trough in the economic cycle isn't necessarily a recession unless the trough is two quarters of negative GDP growth.

Perhaps, but a 50% accuracy rate is not all that impressive. While the sample size is low, it seems that historically, the yield curve as a predictor of recessions has been significantly more accurate than that.