It's interesting the two agencies you mentioned have opted for such divergent approaches.

Ratings are symbolic of default probability or estimated loss. The idea of keeping them consistent for both bonds and derivatives has been the mode of rating agencies since they began rating things like CDOs. It now seems they've all come around to the realisation this is fundamentally misleading, if the rating itself is more likely to change in the latter case.

But the new approaches raise questions of their own. Does Moody's really have the ability to assess the likelihood its analysts will change a rating once given, and incorporate that into its assumptions? In raising the bar for top ratings -- by assuming higher correlation, for instance -- would Fitch be fully recognising the higher ratings volatility inherent in derivatives? Or do we just have to wait for another so-called 'six-sigma' event to merit the bar's upping once again?

Trying to grade the probability of movements in asset prices has given all the major rating agencies similar problems. Rating 'market risk', like derivatives, is somewhat of a departure from their original core competencies. Clearly, they'll do all they can to restore credibility to their rating systems -- because if they can't, the logical alternative is to admit defeat and return to old fashioned (and less hansomely-paid) credit-rating.