In the bottom of the 6th inning of last night's game, Jose Bautista came to the plate to face Cesar Ramos with 2 out, Melky Cabrera on 1st and the Jays up 7-3. Despite there not being a base open, Ramos proceeded to intentionally walk Bautista to face Adam Lind instead, forcing Cabrera to second base and into scoring position. The motivation was clear - get the lefty/lefty matchup with Lind - but issuing IBBs with bases not open is quite unconventional, and provoked a fair bit of discussion in last night's thread as to whether it was the right move for Joe Maddon to have ordered or not.
Of course, we know what happened: as a result, it backfired, since Lind successfully reached; though as a process it was successful - Ramos got ahead and induced very weak contact on the infield. In general, I dislike the majority of IBBs issued, as the value of the extra base runner generally outweighs the benefit sought. But in this case, I thought at first instance it was a good idea, though my conviction was weakened on digging a little deeper. So what follows is a look at the costs and benefits to attempt to somewhat definitively answer the question.
The cost of the IBB
Per Fangraphs' box score, the IBB to Bautista added 0.21 runs to the Jays' run expectation (RE24). That's quite high, almost the average of the other 4 walks in the game (0.23 runs) which is not a particularly promising margin for a tactic that is supposed to be beneficial.
And that goes to the unconventional nature of this IBB, since they're most commonly issued with first base open. For example, using Tango's 1993-2010 run expectancy tables, the IBB was worth 0.23 runs (the current run environment is lower, hence why Fangraphs' gives the lower run value above), but if the base had been open it would only have added about 0.1 to 0.15 runs.
So if you start from a premise that IBBs are generally bad ideas, and the IBB in question has something like twice the average cost at 0.21 runs, exceptional circumstances will be required to justify. Not impossible, but unlikely.
Cesar Ramos' Splits
Giving Bautista the free pass allowed Ramos to secure the platoon advantage against Lind, rather than Bautista having the platoon advantage over him. The value of that depends in part on the pitcher in question; Ramos' career splits are summarized below:
BB% | K% | LD% | GB% | FB% | HR/FB | wOBA | |
LHB | 8% | 21% | 19% | 53% | 28% | 6.9% | 0.281 |
RHB | 10% | 17% | 21% | 39% | 40% | 8.4% | 0.316 |
A quick note: when I calculate BB%, it's actually (BB - IBB + HBP) since IBBs reflect managerial decisions and HBP have such similar characteristics to non-IBBs that it's not worth treating them separately.
Across the board, Ramos is better against his fellow lefties, with the biggest difference in the batted ball profile and his ability to induce ground balls from lefties at a well above average rate compared to below average against righties. It's definitely better for him to face left-handed batters, but with a .316 career wOBA, he's not a trainwreck against RHB. All in all, depending what one evaluates, he's about 10-20% better at run suppression against lefties.
Lind v. Bautista
This is the big factor. Lind, of course, has a long record of failure against lefties whereas Bautista is quite good against them, as is clear from their career production summarized below:
PA | AVG | OBP | SLG | wRC/PA | |
Lind | 871 | 0.216 | 0.260 | 0.337 | 0.07 |
Bautista | 1174 | 0.270 | 0.377 | 0.522 | 0.17 |
Difference | 0.10 |
Lind gives up 54 points of batting average, 117 points of OBP and 185 points of slugging percentage to Bautista. But the last column is most of interest, which is wRC/PA, or runs created per plate appearance. On average, just under 0.11 runs are scored per late appearance, so Lind's production of 0.07 runs/PA is well below, and Bautista's mark of 0.17 is well above average. The difference between them, 0.10 runs/PA represents the difference between Bautista and Lind. And it's only about half of the 0.21 run value that the IBB added for the Jays, which would clearly not be worth doing.
On the other hand, more recently, the difference has been even larger. The same data for 2014 only:
PA | AVG | OBP | SLG | wRC/PA | |
Lind | 16 | 0.071 | 0.188 | 0.071 | - 0.03 |
Bautista | 59 | 0.420 | 0.508 | 0.800 | 0.30 |
Difference | 0.32 |
Bautista has utterly destroyed lefthanded pitching so far, whereas Lind has been even more futile than normal. If one were going strictly off this year, the difference in production between the two is about 0.33 runs/PA, which is well above the the 0.21 runs added by walking Bautista. Therefore, the IBB would make perfect sense.
Of course, the reality is somewhere in between. While the career data is a much more robust sample, most of the older data is of little predictive value. From a run perspective then, whether the IBB made sense or not really comes down to how far one regresses the most recent data against the old.
To estimate the current true talent of Lind and Bautista, I calculated the same data from 2011-2014, weighted by both PA and by a recency factor (1-4 for 2011-2014):
AVG | OBP | SLG | wRC/PA | |
Lind | 0.201 | 0.245 | 0.308 | 0.06 |
Bautista | 0.308 | 0.415 | 0.611 | 0.20 |
Difference | 0.14 |
Overall, it ended up much closer to the career data, with a run creation gap estimated at 0.14 runs/PA. This is only about two-thirds of the value given up with the IBB. And in reality, since there's no regression to the mean factored into the 0.14 runs/PA estimate, the true predicted difference between Lind and Bautista is smaller, making it even more of a bad general strategy. In general, even going from a masher like Bautista to a very poor hitter is likely not worth the added run value from an IBB. But that's not the end of the story....
Wins, not Runs
Looking at runs is useful to assess a general strategy, but baseball's ultimate currency is wins, not runs. Down 4 runs in the 6th, Maddon was trying to give his team the best chance of getting out of the inning without allowing further damage, even if it came at the expense of increasing the chance of giving up a larger number of runs. At that point, allowing 2 runs was basically as bad 4 or 6 runs from the perspective of winning the game.
This is where the big difference in Bautista and Lind's OBP figure in. Using the estimates in the last table, Bautista can be expected to reach base just over 40% of the time against Ramos - or conversely, ends the inning without further scoring 60% of the time. After walking Bautista though, Lind can be expected to end the inning about 75% of the time. And this difference really matters.
I used this win expectancy tool to calculate the expected difference in Win Probability Added (WPA) according to whether Bautista is intentionally walked or allowed to bat, based on the possible outcomes and the likelihood of those happening. I used the same weighting scheme to estimate those probabilities (WPA calculation from the Jays' perspective):
Pitch to Bautista | IBB Bautista, Pitch to Lind | |||||
Outcome | Likelihood | WPA | Outcome | Likelihood | WPA | |
BB | 14.9% | 0.5% | BB | 5.5% | 0.7% | |
1B | 13.6% | 0.8% | 1B | 13.8% | 1.4% | |
2B/3B | 3.8% | 1.8% | 2B/3B | 3.0% | 2.7% | |
HR | 7.1% | 2.7% | HR | 2.3% | 2.8% | |
Out | 60.7% | -0.7% | Out | 75.5% | -1.2% | |
Average: | 0.0% | Average: | -0.5% |
On average, pitching to Bautista gives the Jays an estimated 0.02% higher chance of winning in that situation, so essentially zero. Walking him and pitching to Lind means that when there's a positive outcome for the Jays, their WPA rises more due to the extra runner - but this is dominated by the higher chance of making an out. On average, I estimate it reduces the chance of winning by about 0.5%.
That's not a huge effect by any means, but certainly an advantage every manager should be looking for. Again, there's no regression to the mean in the forecast, so the true change is probably smaller - probably closer to 0.25% than 0.5%. But it's still positive, and still could have helped Tampa win. It appears there was certainly solid method to Maddon's unconventional perceived madness.