Since we don't keep track of where the selected range is in the deck, anecdotally I would say you should pick something from the bottom (furthest from the contestant) since they seem to like to hide the largest amounts down there.
I combed through YouTube, and from the 101 Card Game playings since its refurbishment, I found 97 of those games (I couldn't find 2/9/2015, 10/7/2015, 11/10/2015, 12/7/2015). I was able to identify 164 of the cards used (thanks Drew for proving the game is fair on many occasions!), and was able to this chart, which compares the cards from furthest away from the player to closest.
Avg. Value | $3,750 | $2,714 | $2,000 | $2,200 | $1,889 | $2,783 | $2,875 |
# of reveals | 12 | 21 | 31 | 25 | 36 | 23 | 16 |
# of picks | 3 | 16 | 25 | 18 | 23 | 8 | 4 | |
|
|
$1,000 | 2 | 3 | 11 | 7 | 19 | 5 | 3 | |
$2,000 | 1 | 6 | 15 | 10 | 4 | 5 | 2 | |
$3,000 | 2 | 9 | 2 | 6 | 12 | 8 | 8 | |
$5,000 | 7 | 3 | 3 | 2 | 1 | 5 | 3 | |
Yeah, out of what we have seen, it seems like the edges, specifically the edge furthest from the player, yield the highest chances of being a big hit. However, I'll just throw out that the leftmost card across those 97 playings was only picked a solid 3 times, and those yielded 2 $3,000s and 1 $1,000, so unless something happened in those four aforementioned playings, technically no one has picked up the $5,000 from hitting the leftmost card. Even then, I believe there is solid evidence to believe that the cash deck isn't shuffled.
I look forward to seeing the Check-Out data. I remember running the numbers a while back across three seasons and noting that a range of $19.31 to $19.45 would have won 11/16 times, which was significantly higher than how the contestants did.