Sunday, January 4, 2009

Rating the Managers by Intentional Walks

The intentional walk is one of the more controversial baseball plays in sabermetric circles, and for good reason. As you can see from this run expectancy table at Baseball Prospectus derived from the 2008 run environment, putting a runner on increases a team's run expectancy in every base-out situation possible.

In other words, regardless of whether there are no outs and bases empty on or two outs and runners on second and third or whatever other situation you can imagine, issuing a walk always increases the average number of runs a team scored by the end of the inning in 2008. For instance, no-out, bases-empty walk raises run expectancy by 0.379 runs, while the least damaging IBB (two outs, runner on second) raises run expectancy by "only" 0.129.

Of course, a key word there is "average." Those run expectancies are calculated looking at every batter all season, and intentional walks are used in very specific, presumably strategic situations when it benefits the pitching team, like with a star hitter at the plate and a much less dangerous hitter on deck or to set up a force out or double-play.

In Chapter 10 of The Book: Playing the Percentages in Baseball, the authors, Tom Tango, Mitchel Lichtman, and Andy Dolphin, examine this exact set of factors in detail (you can read the intro to the chapter here). The key outcome of their analysis is a table that lists all the different base-out-score situations in which an intentional walk may be advantageous for the pitching team and how much better the hitter must be than those coming after him to make it worth it to issue the walk. For instance, the table shows that with two outs in the bottom of the ninth, runners on second and third, and a tie score, an intentional walk may be issued if the hitter's wOBA is at least 23% better than the hitter in the on-deck circle.

My goal with this post is to identify which managers are the best (and worst) in their use of the intentional walk. Ideally, I would have factored in the base-out-score situation as well as the relative ability of the hitters at the plate and the on deck circle. However, the dataset I have from Baseball-Reference.com only gives us base-out-score situation for the 1310 IBBs issued in 2008. To look at hitters' wOBAs would require a whole lot more work that I don't have time for.

Still, as a starting point we can get a decently good idea of which managers are the worst in their use of the intentional walk by identifying the intentional walks that were issued in situations that are never advantageous for the pitching team.
This way, we get a list of "maybe smart" IBBs and a list of "definitely not smart" IBBs. My assumption is that the managers who issue the most "definitely not smart" IBBs are probably not very smart in their use of IBBs overall. And indeed, with a whopping 233 walks issued "definitely not smart" situations in 2008, we have a pretty decent sample size to go on. (You can check out the database with the "maybe smart" and "not smart" IBBs noted on Google docs here.)

Here's the list of managers in 2008, ranked in order from most to fewest "definitely not smart" IBBs:

Bobby Cox ATL 21
Jim Leyland DET 17
Charlie Manuel PHI 14
Cecil Cooper HOU 13
Dave Trembley BAL 10
Ozzie Guillen CHW 10
Clint Hurdle COL 10
Fredi Gonzalez FLA 10
Bob Geren OAK 10
Lou Piniella CHC 9
Joe Torre LAD 9
Ron Gardenhire MIN 9
Willie Randolph/Jerry Manuel NYM 8
John McLaren SEA 8
Bob Melvin ARI 7
Eric Wedge CLE 7
Mike Scioscia LAA 7
Ned Yost/Dale Svuem MIL 6
Bud Black SDP 6
Bruce Bochy SFG 6
Ron Washington TEX 5
Manny Acta WSN 5
Terry Francona BOS 4
Dusty Baker CIN 4
Joe Girardi NYY 4
Tony La Russa STL 4
Joe Maddon TBR 4
Trey Hillman KCR 2
John Russell PIT 2
John Gibbons TOR 2

A few observations here:
  • Manny stacks up pretty well. That's good, especially since he ranks so low on base-stealing.
  • If you assume that the average intentional walk costs the pitching team something like 0.185 runs (in a 5 runs-per-game scoring environment, which is a bit higher than what we have now, the run value of an IBB is 0.198), Bobby Cox's 21 "definitely not smart" IBBs cost the Braves 3.885 runs, a little less than half a win. Considering that the Braves issued 80 IBBs total, it seems likely that Cox is costing his team as much as a full win per year with his aggressive intentional walks.
  • You might be surprised that an AL manager (Jim Leyland) ranks so high on this list, given that pitchers don't hit in the AL and conventional wisdom would indicate that intentional walks to get to the pitcher are often smart. But really since managers will pinch-hit for the pitcher late in close games when intentional walks are most often an effective strategy, the actual win value of walking the pitcher is quite limited.
  • You can see the number of total IBBs by team here. Interestingly, Manny is in the middle of the pack (14th of 30) on this list with 44 total IBBs. That means that Manny is using it and using it pretty well, while a manager like Trey Hillman has basically just eschewed the IBB altogether, issuing just 12 of them all year.

8 comments:

KCDC said...

You need to explain the types of IBB's that are "never advantageous."

Steven said...

Well, I'd love to, but it took Tango/Dolphin/Lichtman 34 pages to do that.

Let me work on it and see if I can't write up some basic principles so people can follow what we're talking about.

Benzbe said...

I love SABR stas, but in this case, I think you end up trying to put a objective numeric measure on something that is entirely too subjective and with really little ROI. If it takes 34 pages to explain what constitutes a poor IBB, then you are probably off the reservation in terms of real measurment. There are just too many variables to consider to create decent model. Isn't really enough to say, that good managers probably should consider questioning the traditional IBB situation? Bobby Cox is fabulous manager, IMHO, so any statistical measure that shows that he is the worst lacks some credibility.

Just sayin'

slammalammadingdong said...

Benzbe: regardless of your disclaimer, what you have posted it a backlash typical of people who don't like Sabermetrics. Just because something is complicated doesn't make it wrong, and you shouldn't assume flaws in someones methodology unless you have, i don't know, *read* their methodology.

just sayin.

Benzbe said...

The amount of poor IBB choices doesn't appear to have any negative long term affect on the winning percentage of the managers. Perhaps something else they are doing gets back those 1.4 games they lose each season by going against the statistical best choice. Since the managers strategies are indivisible (there not going to change how they manage), the list is pretty irrelevant. The methodology to come up with the stat isn't necessarily in question. It's application to create a list of managers rating by this statistic doesn't prove much of anything.

Love the blog, BTW.

Steven said...

Well, I think the answer to your observation is that IBBs are an incredibly small part of a manager's job. It's always been my contention that 70% or more of a manager's value is in his ability to motiovate the players. Then teaching skills, then in-game strategy, of which IBBs is just one part.

As you can see from the post, the most extreme oulier of bad IBB usage is still worth less than half a loss total. This kind of strategy is important because optimizing walks, pitcher usage, lineups, and all these small things can add up to a win or two and get you into the playoffs. But it's not going to turn a bad team into a good one or vice versa.

rickeyruns said...

Hey what's the split between Jerry and Willie on the Mets 8 bad ibbs?

Steven said...

6 Jerrys and 2 Willies.