They also get 55% on success for A-rated companies, which is also pretty good. It's not quite 60%, but it's very close, and considering how much above 60% they get on predicting failures, I think that shows a fairly high accuracy of prediction.
Far far less than half of companies succeed. So identifying successful companies by random draw is not equivalent to a coin toss. In the case of this study, its 10%.
55% on A-rated companies is way better than the "coin-toss" model of randomly assigning a rating to each company, because there are more than two ratings.
