The resulting measure of similarity after a matching is a sum of Dayhoff entries, and hence it is 10 times the logarithm of this probability. For example, a matching between two amino acid sequences with a similiarity (or cost or score) of 238.8 means that the probability of both sequences coming from a common ancestor, as opposed to being a random alignment, is 1023.88 times more likely. Although crude, this gives a rule of thumb for estimating the quality of a matching.
This fact makes aligning with Dayhoff matrices a soundly based algorithm. This has been noted by many people including [2,9,12,13] yet such methodologies have been largely ignored.