With 47 games left for Boston, let's say they go 28-19 (.596 clip). That would put Boston at 109 wins.
Now, Boston's worst 47 game stretch so far this season was from May 1 through June 20, where they went 28-19. So this is a reasonable low bar to set.
It isn't really, though. Just because the Sox haven't had a real, extended slump yet, that doesn't mean they can't.
Any team, no matter how good, can have a bad month. And in the Sox' case, that's most likely to come in September, when 15 of their 26 games are against likely playoff teams (ATL, HOU, NYY, and CLE), 9 of the 15 on the road. It wouldn't be at all shocking even for a team this good to have something like a 10-16 record for a month like that. If the Yankees are performing at roughly the opposite of that pace at the same time (easy to do if the Sox are losing, since 6 of those games are head-to-head), that's most of the lead right there. Which the Sox could survive,
if they take care of business for the rest of August, when their schedule is almost as easy as the Yankees'. But they have to do that. If they take the foot off the gas and walk into September with the lead down to 5 or 6 games, things could get serious very quickly, and before you know it the game threads will be swarming with chicken and beer references.
If past performance against other teams means anything though, the Yanks are just 6-6 vs TB and 6-6 vs BAL. So, 7 more against each of them.
Why would past performance mean anything in this case though? It seems far more likely that TB and BAL just randomly outperformed their true talent vs. the Yankees over a small sample of games than that there is something quirky about their rosters vis-a-vis the Yankees that makes those two pairs of teams, in spite of their records vs. everybody else, truly an even match for each other.
In short: the Yankees are a better team than the Rays and a much better team than the Orioles. We should expect them to win more games than they lose against those two teams, regardless of what's happened before (but given the sample size, we shouldn't be surprised if they don't).