Wouldn't their prediction necessarily affect the outcome? Even more so if they intend to market it to investors. If they predict startup X is going to fail, and

Wouldn't their prediction necessarily affect the outcome? Even more so if they intend to market it to investors. If they predict startup X is going to fail, and this prevents people from investing in Startup X, then it's more likely that Startup X will fail. On the other hand if they predict that Startup Y will succeed, and Startup Y gets a lot of press coverage and interested investors because of that, then they helped it happen. It's kind of self-fulfilling. In order for them to prove that their algorithm is really accurate, they'd have to make the prediction and keep it secret, then see later if it came true or not. Of course, that doesn't really make for a marketable service...