This is why using a small sample size distorts perception. Sometimes a run is testing weakness/strength of the other team, attacking from a different angle or using a different look. Sometimes the result might be -2 yds, but it showed you what you needed to know about a defense. Sometimes you are running into a strength of a defense so the next play action works as intended. If Ezekiel Elliot goes out of a game injured after 4 carries and he has 2 total yds, do you immediately say "man I'm glad that bum is out of the game." Or do you believe that had he been given a few more carries he was going to find a rhythm and start hurting the other team? My point was simply that in this example 1 carry represented too high of a percentage of the total output to be casually thrown away. If I made a statement of " sure but if you take off 20% of his best plays his stats don't look good at all", I would assume people would look at me a little bit funny. That's what the 1 carry in a game is when you only get 5 runs.
And why throw away any carries anyway (beyond 20 carries)? In almost any statistic, if you want an accurate picture of a player, consider everything. If you want to suit an agenda, start picking out the carries that go against said agenda. I agree with you. There are far too many of those "If you take away..." posts on boards.
so if a QB has 40% completion % 4 ints and 2 tds and a bunch of yards but the team wins... bc that is exactly what cutler is... so bc your reasoning proves ajayi is good and cutler is good. but everyone hates cutler....