There really isn't any stats I'll outright complain about but the least for me are ones that end up being total crapshoots in hockey pools, especially head to head. Like a lot of people pick out +/-, but if you draft carefully it's a stat you can intentionally win at. Also while it often gets misused it does have it's value within a certain context, especially when you have a line that consistently plays together (ex: look at Carolina's team player stats for last season). So anyways my main culprits are usually GWG, SO, and to a lesser extent SHP.
GWG is pretty obvious, it's almost completely random and more often than not entirely meaningless in the context of a game. I mean if the game is tied 2-2 and Stamkos scores a clutch goal with a few minutes left in the third, that's noteworthy. But if TBay is winning 6-1 and the other team scores 2 late goals in the third, the fact that Salo happened to score the 4th goal earlier is completely meaningless. If anything this stat would have more meaning if a number of key goals from a few very specific scenario's were all compiled into a single 'Clutch Goal' stat.
For Shootouts? It's again pretty random, and while a noteworthy accomplishment for a goalie is it really much different than a skater scoring a hat trick? If Stamkos and Ovechkin both score 60 goals on a season, you wouldn't award Ovy the Hart because he scored 8 hattricks and Stamkos only had 3. Yet for goaltenders with roughly equal starts, wins, GAA and SV%, SO's can stand out as a factor.
And of course like GWG they're almost impossible to predict in a hockey pool