Yes, there is definetely something fishy with 3-2 in the stats. As an example, I chose to play 3-2 in the cup semifinal this year
(8119581). From my anaysis this would leave the game as a 50/50 tossup, and just as expected it was a very close game (I lost by 1 point), however accoring to the ratings I should have lost big time.
Looking back to the ratings from this season I have achieved my best perimeter defense, repsectable 2 times, when I've played man-to-man, while with 3-2 and 1-3-1, and more or less the same lineup, I have reached mediocre at best. However, the results seem to be "correct" even if ratings are missleading.