From here via Greg Mankiw, the economist Jeremy Siegel is quoted as saying:
If post-World War II patterns hold for the future, he calculated last week, prospects for stock investments are excellent: there would be a 96.6 percent probability of a positive return for the next 5 years, going up to 100 percent for 10- and 20-year periods. Average real returns would be stellar — about 11 percent annually in holding periods from 1 to 20 years.
A 100% probability, eh? An impressive, stick-your-neck-out prediction! But wait; what was that predicate? It was, “if post-World War II patterns hold“. And what are the post-WWII patterns in question? Why, only that over any given twenty-year period, stocks went up. So what he’s saying is that, if stocks continue to go up over twenty-year periods, you can always get a positive return from stocks over twenty years. No duh! If A then A, very good! But we’d quite like to know the probability of A, thanks. In other words, can we please have a probability for the variable of actual interest, namely “will post-WWII patterns continue to hold”?