How much bias the magical pie shows toward recent performances over those from past years is a question I don't know the answer to. A good system would definitely weight recent performances higher while trying not to be tooooo swayed by brief patches of really good form. Wells batting numbers might suggest the numbers are not too heavily swayed in favour of recent performances, because 38 and 40 batting averages in last two seasons are not great compared with other lower middle order players around (averaged 60 back in 09/10 FC). Wells is ranked 69th in the MVP points thing (FC) and ironically 1 and 2 are two batsman/medium-pacers (Franklin and Ellis). Yet these two were passed over (Franklin perhaps because of past failures) and Wells picked.
Same with Bates - agree last few seasons has been pretty good in limited overs so if considering that whole period then deserves to be there. But then we have Latham who's been selected from a very short time frame.
Another factor - dunno if they go into this much detail - Wells may (I haven't looked and can't remember) have scored runs when his team were in a bad position, it was a match where not many runs were scored in total and the opposition bowlers were rated highly (by the same pie system). Therefore the system may give his runs a higher value than someone who scores runs in the opposite situation ie. when it's easy. Similar to your point about Bates taking higher-order wickets being more highly valued by le pie.
Anyway that's all speculation. My last bit of speculation is that although only 5% is meant to be based on intuition, I'm pretty sure they rate people on being a good team man, fitness and fielding, all of which are probably just ratings based on gut feel themselves. If you're the coach/selector, this would be very easily manipulated to get the answers you want