With apologies to Charles Dickens. I'm sorry this is so long, but he got paid by the word, so his stuff is pretty long, too...
Terry Francona is a great manager. If you quibble with this statement, you're crazy or trolling. He's calm and funny in a city with voracious media, and has done a fantastic job controlling very disparate personalities while still keeping the respect of his clubhouse. While he sometimes manages with the "long view" in the regular season to the short-term detriment of the team (sticking with veterans when they're not performing, putting out "Sunday lineups," leaving starters in longer than seems rational), he's often proven right in the long-term, and we've seen that when the games really count in October he goes for the throat and rarely makes a poor tactical decision.
That said, there were two tactical choices in the past two games that I strongly disagree with, and in my opinion Francona is lucky that the Sox ended up winning both games. Note that this doesn't change my opinion of Francona very much at all -- I still think he is the greatest manager I've ever witnessed and I want him in charge of the Sox as long as he wishes / is physically able, which I hope is a long time. But these decisions were poor in many ways, and I'd like to point them out and discuss them, and I really hope he doesn't make similar choices again.
In the Sunday night game against the Yanks, the Sox were down by one in the bottom of the ninth against the greatest closer of all time when Scutaro got a leadoff double. Jacoby Ellsbury then bunted him to third. Rivera caught the bunt on one bounce, looked at third, but then saw no one covering. He then was still able to turn and throw to first to just get Jacoby. The bunt was "successful," and Pedroia was able to get Scutaro home on a sacrifice fly, but it wasn't a good decision to bunt.
Intuitively, the bunt seems like probably a good idea. You just need the one run and Rivera is one of the stingiest pitchers in the game, so if you move the runner over you give yourself a shot to tie the game on a fly ball, or even a grounder. But Fangraphs has it as -.037 WPA, which means that given average pitching and hitting, the Sox were 3.7% less likely to win the game after that play than they were before.
There are a couple problems with this number, though. Obviously, Rivera is far from an average pitcher, but Ellsbury's also far from an average hitter. Something also to consider is that just scoring the one run isn't going to bring your win expectancy all the way up -- it's going to bring it to around 50% since you still have extra innings to deal with.
So here's another way to look at it: As I see it, the choice is either let all three guys (Ellsbury, Pedroia, Gonzalez) swing away and hope one can get at least a (non-infield) single to knock Scutaro in, or have Ellsbury sacrifice, and then hope Pedroia can get either a hit or a fly ball (as the infield would be in, so a grounder probably doesn't do it), and if he fails, then you have Gonzalez with two out. So let's analyze these two choices.
In choice 1, we can look at batting averages of the players involved as all we care about is their chance of getting a non-infield hit (a walk would be ok, but probably somewhat detrimental since it increases the chance of a double play, though obviously not as bad as an out). Rivera allows just a .211 BA against for his career, which is 80% of the league average during that time. Entering the game, Ellsbury was hitting .321, Pedroia .310 (though hotter recently), and Gonzalez of course is leading the league with a .352 BA. Let's apply a "penalty" to those numbers, since they'd be facing Rivera, and take 80% of each of them. For his career lefties hit Rivera slightly worse, but he's not a LOOGY, and these are just guesstimates anyway. Also, Ellsbury has 18 infield hits this year, Gonzalez 13 and Pedroia 22 so I'm going to exclude those since they aren't going to score Scutaro.
So doing the math we get expected non-infield hit BAs off Rivera of .233 for Jacoby, .222 for Dustin, and .265 for Gonzalez. The chance of all three failing at this is (1-.233)*(1-.222)*(1-.265), or 43.9%, with a small chance of getting a walk or infield hit and bringing Kevin Youkilis to the plate with a runner on third. Or to put it a more positive way, they'd have a 56.1% chance to get the run home.
But that's not all. Many outs aren't complete failures in this situation, as a "productive out" would essentially act as a sacrifice hit. Amazingly the Red Sox have been able to move the runner from second to third (or better) 51.2% of the time even when you remove sacrifice bunts entirely (whether by hit or "productive out"). B-Ref doesn't break these numbers down into hits v. productive outs, but I think you can assume that if Ellsbury swings away, he has not only a good shot at a hit, but also a shot at getting a grounder to the right side or a fly ball that could move Scutaro to third, giving you the same result as a sac bunt.
The math on choice 2 (just going for the sacrifice) is a little simpler, but there are some other things to consider. First, the sacrifice bunt could go wrong. Most people (and managers) would probably be surprised to learn that sacrifice bunts are only successful about 70% of the time (this year it's 68% in the AL). So by choosing to sacrifice, you're not automatically going to end up advancing the runner(s). As we saw with Ellsbury's bunt, Rivera got to it quickly and had the Yankees put the "wheel" play on (with the shortstop covering third) Scutaro probably would have been out at third. Ellsbury's not a very good bunter, so moving Scutaro over is far from a sure thing. To do this math, we need to split it up into two scenarios.
First, we'll assume Ellsbury is successful bunting (which he was). That means you're just hoping for any kind of hit from Pedroia or a sac fly, or if he fails then you want a hit from Gonzalez. B-Ref has splits for Pedroia with a runner on third and less than 2 out, and they're pretty good - in 154 such PA, he has 48 hits and 29 sac flies, so exactly 50% of the time he's done one or the other. We'll call it 40% on account of Mo. So added to Gonzalez's chance of getting a hit above, we get (1-.4)*(1-.265), or 44.1%, again with a chance of either batter walking. This 55.9% is about as much chance of scoring as we got in the first scenario.
But, there's about a 30% chance that Ellsbury will fail in his sacrifice, likely leading to a situation with one out and a runner on first. Boston's chance of scoring isn't zero in such a situation, but it's not good -- teams only win the game 20% of the time in such a situation, and that's against average pitching.
Adding in the chance that Ellsbury would fail in his sac attempt to choice 2, it was a very poor decision to ask him to do so.
This is a very longwinded way of saying that the win expectancy is right. It's meant for average pitching and hitting, but the hitters due up were much better than average, as was the pitching, so they probably come close to canceling each other out.
Last night, the game was tied in the seventh when both Scutaro and Ellsbury singled off Dumatrait. This brought Carl Crawford to the plate against the lefthander, who went to a 3-0 count before throwing a strike. THEN Crawford laid down a sac bunt, which moved the runners over. This ended up not helping the Sox, as Gonzalez was walked and then Pedroia hit into a double play with the bases loaded. But the decision to bunt was even more of a bad choice in this instance.
Fangraphs has the bunt as only lowering the Sox win expectancy by .003, or 0.3%, which is very slight. But it doesn't take the count into consideration. Crawford has a 1.131 OPS for his career after a 3-1 count, which isn't surprising. At that point, the pitcher is one ball away from a walk, so even the free-swinging Crawford has an OBP of .602 after a 3-1 count. Even this year, when he's been struggling, he has a .455 OBP after such a count. So, you're giving up a 50-60% chance of getting on base and moving the runners over without giving up an out, and taking the bat out of one of your hotter hitter's hands (and one who could certainly use a confidence boost), for the 68% chance that you will execute a successful sacrifice and move the runners over?
This post is already way longer than I meant it to be, so I won't go into the math of this decision. But even if the bunt made sense according to win expectancy, and even if the worst hitter on the team was up, it wouldn't make sense to bunt after a 3-1 count, because you're pretty likely to just get the runners to move up on a walk anyway -- and almost as likely as the 68% success rate on sac bunts.