Just for fun and the sake of learning I am researching on the vast literature on economic prediction. This topic is the one that has always enthusiast me due to the magic behind (you get to see the future!). However, nothing more apart from reality: you CAN'T see the future!!!. Rather, sometimes you'll interfere with it (i.e. the self-fulfilled prophecy in financial markets), maybe you'll be able to scratch the surface (i.e. if you are smart enough to give a good confidence interval), but the only 100% accurate prediction is the ex-post one (OK, that's cheating).
At the moment I'm in a stage when I've already chosen my branch (black-box statistical models, with no underlying economic structure), and the strategy (brute force). Hey!, although I'm an economist, I'm quite practical as well; if it works, it works for me. However, if any results come to light, I'll start with the economics science baggage and will try to go one step ahead.
As a final thought, a good prediction process will need to be resillient. What I mean is against the unexpected events, or shocks, it should be able to reframe and recalculate the predictions made. Anyway, that's more a comment on the scientist, rather than on the model itself.
Let's see what we get.-