Maybe not quite that persistent, but I've definitely tried to sell investors on startups that didn't end up doing very well. Even by the end of the 3 months we are far from perfect at predicting who the big winners will be.

The idea with this kind of thing, as far as I can tell, is to be as nuts (i.e. risk-friendly) as you can possibly get away with. If you're never wrong, that's probably an indicator that you could get away with being a lot more nuts.

Moreover, if you do pump a startup and you are wrong, then to explain the situation is often to piss all over the work of said startup.