The problem with simply deciding Hembree and Kelly have been too ineffective lately (which they have) and therefore should be dropped from the post-season roster still eaves the question - for who? and are all these sample sizes just too small to be representative?
It's safe to assume Kimbrel, Barnes, Brasier and ERod are going to the post-season bullpen. Wright probably is, too.
That leaves a maximum of three spots from Kelly, Hembree, Workman, Velasquez, Poyner, Johnson, Scott, Thornburg, Cuevas and Pomeranz. Four if they decide not to take ERod. Right off the bat, I would say Thornburg and Pomeranz have struggled to stay healthy and can be left off from further consideration, leaving 8 pitchers for 3 or 4 spots.
Of these pitchers, only Poyner and Cuevas have thrown more than 7IP this month.
Should the Sox leave Kelly and Hembree off the post-season roster because Cuevas looked better in the 11 IP he has thrown, half of which came in one outing?
A mere 7IP ago, Kelly had an ungodly August with a 1.42 ERA/1.90 FIP. Is that no longer relevant because of those seven innings since? (Overall, Kelly has had better second half than Barnes has had. Is Barnes on he bubble?)
I really don't think the Sox should play "hot hand" with these small sample sizes and hope it continues for another month. At some point, they are going to need to bring some of the pitchers that got them where they are in the first place...