I'm reminded of LTCM's death spiral, the story of which is best told in Nicholas Dunbar's book. In both cases, a firm holding large positions got hit by something like a speculative attack where every counterparty in the world had an incentive to bet against it.
No credit rating agency could have possibly comprehended with the complexity of information, and the pace of events, that unfolded in the endgame. Bear Stearns was a publicly listed firm, so the stock price generated a daily estimate of the default probability (as did the CDS). Both these market-based indicators of distress worked very well. A curious little difference between Bear Stearns and LTCM: Neither of these two realtime indicators existed for LTCM.