Relievers (especially closers) are the worst example to use for this, because by definition they will have the smallest sample sizes and are therefore the most volatile and susceptible to variance. For a guy like Fernandez, his HR spiked a bit (opponent-dependent), his OAV went up by about .02 (could be luck and/or could be defense-dependent), OBP went up by about .03 or .04 which can be luck (opponent variance, umpire variance), slugging went up .100 points which is definitely bad luck and/or opponent-dependent and/or because you have bad player settings such as pitch counts too high and/or call bullpen too low. The difference between max-pitch 15 + call bullpen 1 and max-pitch 10 + call bullpen 3 can mean the difference between his 3rd at bat being a 3-run HR versus a situation where simmy calls in a RHP to get an inning-ending double-play. Bullpens are incredibly nuanced