A single set of CATs results may not be reliable.
Even NFER never claimed they could be trusted 100%.
There was a discussion about this here:http://www.elevenplusexams.co.uk/forum/11plus/viewtopic.php?f=16&t=17357&p=207889&hilit=CATs#p207889
Ideally I think one would want to see a series of CAT tests over time.
It's interesting that the Bucks 11+ consists of two VR tests (set by GL Assessment/NFER) which are meant to be of equal standard, and they accept the higher of the two scores, presumably to allow for the reliability factor. 121 is the qualifying mark. It's not unusual for appeal panels to see scores of 110/120.