I do not make predictions myself, but I have some recommendations for those of you who do. If you want to make sure that you're predictions are never "wrong," simply assign some probability and/or confidence level to your predictions above 0 but below 100%.
Here's how it works. Suppose you want to predict that the Cubs will win the World Series in 2010. Normally, we would say that anyone who would predict that must be an idiot. The prediction is bound to be wrong based, apparently, on the laws of physics. However, the predictor can protect himself or herself by using the following formula: "I am 97% confident that the Cubs have a 26% chance of winning the World Series in 2010." Then, no matter what happens, the prediction was not "wrong." If the laws of physics abate and the Cubs do win the World Series, then the predictor looks like a genius. But even if the Cubs lose, the predictor can say, "Well, there was always a 74% percent chance they would do so, and perhaps even more given that I was not completely confident in the probabilities I was assigning." In fact, the prediction is perfectly consistent with the Cubs extending Milton Bradley's contract for another 10 years and ending the season with a record of 0-162 (or worse).
Bottom line: Assigning probabilities to your predictions will ensure that you are never "wrong." You will be able to claim that all of your predictions were accurate, thus infuriating your friends and alienating family members.
Another alternative for predictions is to make them as normative assertions, such as the following:
The Cubs should win the World Series in 2010.
Iran should give up its nuclear program in 2010.
Dick Cheney should shut up in 2010.
Nancy Pelosi should get the hell off my television screen in 2010.
Again, no way anyone, with the aid of hindsight, could accuse you of making a "wrong" prediction, when Nancy Pelosi appears on television on January 1 to wish you, personally, a Happy New Year.