How can it be a pattern if it doesn't happen frequently? It only happened 3 times out of the sample you chose. You guys are seeing things that aren't there. Look at every single game for the season, and tell me what percent of the time it has happened and calculate the percentage. It didn't happen at all in April. It happened once in May in a win. So out of 72 games it has happened 4 times. That is not pattern.
A more likely explanation is that Farrell, smart manager that he is, has determined after two months who his most effective relievers are and uses them to enhance the chance of winning close games. Not a pattern, but learning from the performances two months into the season. Would you rather have him send out others just to break the pattern that you perceive that really isn't there?
Four times out of 72 games is 6%. Not a pattern.
One other thing: what is wrong with the "pattern"? The Red Sox have a winning record when the so-called "pattern" occurs.