The Kershaw decision was decidedly not an analytics decision. Kershaw is the WORST ERA in the playoffs guy in history for pitchers with 10+ innings, and his second half sucked. I mean you've got 5+ better guys on that team to piggy back with, from a statistics standpoint.
As Moose points out, "analytics" wouldn't be based on a broad statistic like career playoff ERA.
But these days we really don't know what's going into making some of these decisions.
That Roberts telegraphed the move before the game suggests it was a group decision.
But there's more than just one decision in the "Kershaw decision".
I mean, they bring him in with 2 on and 2 out in the 7th, and he gets Adam Eaton to strike out.
That decision worked.
But the decision to leave him in to pitch the 8th.
The decision to leave him in to face Soto after the HR.
Maybe it was just Roberts staying with his guy. Was listening to LA Times writer Arash Markazi on with Jim Rome this morning and that's what he suggested, but I don't think anybody outside the organization knows.