The ninth-inning last night was the decisive one, and the difference between the top and bottom of the frame basically decided the game. Both teams got their lead-off man aboard, one with a hit, one with a walk, and both then tried to steal second-base. The Diamondbacks failed, with Justin Upton being caught; the Giants succeeded, and Cody Ross then won the game for them with a single down the line.
As noted in the comments, Arizona's failure reduced their stolen-base percentage to 69%, with 27 successes against 12 times caught. That's below league-average and is also quite possibly at a level where it is having a negative impact on their offense - the break-even point, depending on your source, is somewhere between between two-thirds and 80%. But it does depend significantly on the game situation.That's because a stolen-base can have a very significant effect on your win probability, if carried out at the appropriate time. And the late innings of a game where the score is tied is possibly the best time to be doing it, because it improves your chances of scoring that one run, and that one run is all that matters. To look into the specifics of last night, I'm using the Hardball Times Win Probability Enquirer, which lets you plug in two situations - typically, one play apart - and tells you the chances of your team winning the game, before and after. [I used a run environment of 4.0, towards the low end, because of both bullpens being good]
In this case, we start with the situation being the top of the ninth, a tied game, with a man on first and no outs. At that point, the Diamondbacks' chances of victory were 58.14%. After Upton was caught, the bases were empty with one out, and our win probability was down to 44.49% - it cost us 13.65%. But what if he'd succeeded? With a runner now on second and no outs, our win probability goes up to 67.35%, improved by 9.21%. To find the break-even point, the SB% where trying to take the base becomes a good thing, we use those two changes: it's 13.65% / (13.65% + 9.21%).
That number works out at only 59.71%, demonstrating that given the circumstances last night, it was more likely to be worth trying last night. 13 of the 16 NL teams, including Arizona, have a better SB% than that, and coming in, Upton was four of six, a 67% success-rate. As Mark Grace noted during the broadcast, the failure was not in the attempt, it was J-Up getting a poor jump off first-base which led to him getting nailed. As an aside, for the Giants' stolen-base in the bottom of the ninth, the break-even point was lower still: they needed just a 56.06% chance to make it worth a shot.
It would take an analysis like that of all our stolen-base chances to figure out whether or not it has been a boon or a burden to our overall win probability. I'll maybe take a look at that later in the week. Maybe...