Posted by Bill Baer on Wednesday, September 9, 2009 at 7:22 am

Click here for Part I of my interview with Chris Jaffe, author of Evaluating Baseball Managers: A Comprehensive History and Performance Analysis, 1876-2008.

Here is Part II.

. . .

What motivated you to write an entire book about managers?

First, I’ve always felt that the way people analyzed managers was a bit off-kilter.  When you listen to sports radio or read blogs, the overwhelming majority of talk about managers focuses on their in-game tactics.  That’s fine, but there is more to the job than that.  Ask baseball fans and they’ll readily admit that, and most will even concede, that the most important parts of the job aren’t necessarily the strategic elements.  Yet people’s conclusions about whether/not a manager is any good rely heavily and at times exclusively on the portion of the job we can easily evaluate.

I remember reading The Bill James Guide to Managers, and thinking it was great but finding there was a bit of a gap in it.  One of the main themes in that book is that the in-game decisions are only a portion of the manager’s job, and not necessarily the most important portion at that.  Yet on the occasions he tries to quantify managers, it either focuses narrowly on particular in-game decisions, or offers something a little too broad (most notably mid-way through he offers a system to judge managers based on wins and losses).

Don’t get me wrong, both those approaches were fine and informative, but each seemed a bit off the heart of where James was going in his analysis.  I thought the best way to analyze managers would be to look at if players improved or declined under their watch, but I had no way how to quantify that.

Fortunately, someone else did.   In summer 2005, at the annual SABR convention, I saw a presentation by Phil Birnbaum in which he introduced something he termed “the luck database,” which evaluated if teams under/overachieved in the course of a season.  Two items really caught my attention: he invented algorithms to determine if pitchers and hitters over/underachieved in a given year based on their performance in surrounding seasons.

It was one of the best presentations I’ve ever seen at a SABR convention, but what Phil termed “luck” was really under/overachievement.  Other factors could explain it, including manager.  According to his algorithms, the Earl Weaver Orioles were consistently one of the “luckiest” teams in baseball in the 1970s. Bobby Cox and Leo Mazzone have presided over some of the “luckiest” pitching staffs ever.  Birnbaum’s algorithms might quantify managerial improvement on players.

I ran a quick test on this to determine managerial impact.  I took the entire database and divided it into four categories: 1) games helmed by a manager who lasted at least 2000 games, 2) those where the skipper survived 1000-1999 games in his career, 3) 500-999 games, and 4) 499 on down.  I was making the assumption that some correlation exists between length of managerial career and managerial ability.  As assumptions go, that’s one that I feel pretty safe in making.  I know it’s not perfect and some duds last a long time and others don’t last as long as they should, but by and large it works.

I figures if the longer managers performed better, that means these algorithms likely are telling us something about managerial performance (which I believed anyway, but it’s nice to have proof).  Not only did the 2000s do best, but the 1000s did next best, then the 500s, and then the 499s in last.

And the shape of the differences really reinforced my notion that this indicated managerial skill.  If it was luck, you’d expect the last bunch to be last anyway (they were unlucky and it was a smaller sample size), but the 2000s should be the ones closest to average because over the course of 13+ seasons luck may not even out perfectly, but it ought to come close.  (And keep in mind, 2000 was just the minimum, some of these were up around 3,000 or 4,000 games or more).  Yet that didn’t happen.  The 2000s, with they huge individual sample sizes, came off as especially “lucky” on the whole.

Also, one thing I found interesting is that the gap between the 1000s and 500s was smallest.  If Phil’s database only indicated luck, the gap between the 1000s and 2000s should be smallest due to sample size.  Didn’t happen.  Instead, it was the two middle groups (neither the best nor worst managers, going by career length) that were closest together.

I should note, his database contained three other elements: runs scored versus how many runs scored they “should’ve allowed,” same thing for runs allowed, and pythagenpat deviation.  I ran them threw the 2000s/1000s/500s/499s test as well.  They results didn’t come out quite as heavily on behalf of managerial skill as Phil’s algorithms, but they weren’t far behind.  I ended up mentioning all five in the book and offering some explanations for why managers might have an impact on those areas, but ultimately the two algorithm-based elements were the ones I have the most faith in.  At any rate, once I had Phil’s database I knew I was on to something.

I didn’t realize I had enough for a book until I came up with a database on my own I term the Tendencies Database.  The concept is simple: take a stat – we’ll use sacrifice hits for example – and adjust it for context (which would be sacrifices divided by the times there’s a runner to sacrifice).  Then see how a team ranks, and adjust for league size (which is a little more complicated than it sounds, but I don’t want to bore people with the math).  Then average out for a manager’s entire career to get an idea where he stands.  This works with any stat: sacrifice hits, relief pitchers used, likelihood to have the platoon advantage – anything.  The score tells you something about the talent on hand, but I only use the Tendencies Database to look at those on the extreme end of the leaderboards, and it’s mighty hard to be the all-time king of a particular tendency unless the manager really prioritized that.

Phil’s database gave me a jumping off point and way to get the big picture of managerial performance.  The Tendencies Database allowed me to look at the details and compare across eras.  Once I had both, I knew I had a book.

Frankly, neither are perfect, and I acknowledge their problems in the book.  But imperfection isn’t a synonym for useless.  As long as you’re aware of their problems and limitations (which I tried to do throughout), you can use them pretty effectively.  With these things propelling my research, I analyzed every manager who lasted at least ten seasons as a team’s primary skipper, plus a dozen of the most important men with shorter careers.

How long has the whole process taken you? Would you advise any wannabe authors out there to make the leap?

I figured I might have a book in during the summer of 2007, and submitted a proposal to McFarland (my publisher) just before the annual SABR convention.  I signed a contract that fall to finish and deliver a manuscript by April 1, 2009.  I worked on it throughout the next 18 months.

However, a lot of work was done before I ever thought it could be a book.  I saw Phil’s presentation in the summer of 2005 and began applying it to managers later that year.  I wrote some articles for Baseball Think Factory on managers in the summer of 2006.  (They’re now offline, a condition McFarland insisted on for the book contract).

Also, an entirely different project I’d done on the leveraging of starting pitchers (or, on how managers used to intentionally spot starters against particular rival clubs) came in handy.  I started dabbling with that research in 2004 and did the bulk of that research by 2007.

As for advice: write a book only if you are really passionate and interested about what you’re working on.  Do it because you think you have something to say about it.  The whole process takes an enormous amount of time, work, and energy and it just plain won’t be worth it otherwise.

Were you surprised by a lot of what you found in your research?

Hugely.  I don’t dare go into detail – I’d end up retyping the entire book right here.

Do writers and fans generally have a good idea of who is and who is not an effective manager? If not, what are some of the key factors that are often left out? Do you think a “player’s manager” is any more or less effective than a dictatorial manager?

I don’t think any particular strategy is necessarily less effective than the other – it’s more about how one implements them.  Any strategy can go badly and all popular ones have also gone well many times.

As for fans’ opinions, they can be informative, but I tend to be a bit skeptical.  Again, I think people underrate managers on the whole.  Go online and it seems like fans from ten teams think their manager is the biggest dullard in the world, fans from another ten teams will merely think their guy is lousy, a few don’t mind their manager, and denizens from the remaining handful of squads will shockingly have nice things to say.  That can’t possibly be the case.

That’s what happens when people generally rate managers based on in-game decisions.  Usually you only notice them when they go wrong.  It ain’t like football where the guy can draw up a new, wondrous play.

In bar discussions, you’ll hear people describe the perfect woman as someone with — let’s say — Natalie Portman’s face, Scarlett Johanssen’s body, and (for me) Janeane Garafalo’s intelligence. If you could build a perfect manager, from whom would you draw and why?

Well, I wouldn’t take Don Zimmer’s body, that’s for sure.

One oddity in answering this question is that many managers’ strengths and weaknesses are related.  The same intensity that allowed Billy Martin to improve teams on arrival also caused him to wear out his welcome so quickly.  I’d love to let Bill McKechnie design my team’s defense, but that also means he’ll be in charge of its offense.  Trying to isolate particular strengths is a bit weird.

I’m tempted to say as a joke I’d go with Joe McCarthy as the perfect manager.  Actually, I’m not sure that would be a joke.

Getting at what you’re asking for though, I’d take John McGraw’s focus on getting on base, Charles Comiskey’s philosophy for preventing teams from scoring runs, Dick Williams’ willingness and ability to get the most from his unproven players (check it out, if you’re curious – his track record was amazing), Casey Stengel’s ability to handle the press, Billy Martin’s intensity, Walter Alston’s unflappability, and a generous spackling of Earl Weaver to fill in the rest.

Then again, maybe I’d just stick with Joe McCarthy.

I know he’s a pitching coach, but I’d love to get your take on St. Louis Cardinals’ pitching coach Dave Duncan. I hear — and see — that he works wonders. Do you buy it?

He’s done a great job.  Where does Dave Duncan end and Tony LaRussa begin?  I definitely give Duncan more credit for what’s happened to pitchers than his boss, but pitching coaches aren’t magic bullets and can’t be easily divorced from their surroundings (a statement that holds true for managers as well).  If you don’t believe me, ask your friendly neighborhood Orioles fan what he thinks of Leo Mazzone.

Rather than rehash what I say in the book, how about I end this with a little mini-excerpt?  Here’s what I say about the performance of starting pitchers under LaRussa/Duncan in my book:

St. Louis experienced a terrific stretch in the mid-2000s built around a core of Albert Pujols, Jim Edmonds, Scott Rolen, and Chris Carpenter.  By 2008, only Pujols remained (well, St. Louis still had Carpenter but he was too badly injured to be of any value).  The team especially lacked dependable starting pitchers.  They featured mediocre journeyman Kyle Lohse, and reclamation project Joel Pineiro backed up by two converted relievers – Braden Looper and Todd Wellemeyer.  Their most reliable hurler was Adam Wainwright, a 26-year-old with only 32 major league starts in his career.  That was a prayer, not a stable starting rotation.  LaRussa made it work, and the Cards ended the year with an unexpectedly strong 86-76 record despite playing in the NL’s toughest division.

LaRussa has made a career of getting more than one would expect from his starters.  He had numerous quality staffs despite rarely having elite starting pitchers.  Chris Carpenter had a great stretch, but it was brief before injuries felled him.  Besides, though Carpenter had been promising, he had never established himself before joining St. Louis.  Tom Seaver is the only established great pitcher LaRussa has ever had, but he was at the end of his career when he came to LaRussa’s White Sox.  LaRussa is more likely to get good production from veteran pitchers who never wowed anyone before.  The prototypical LaRussa success story was Dave Stewart.  A struggling reliever before LaRussa got a hold of him, Stewart posted four consecutive twenty-win seasons for the A’s.  LaRussa also oversaw revivals from Darryl Kile, Woody Williams, Kent Bottenfield, Mike Moore, Floyd Bannister, Garrett Stephenson, Todd Stottlemyre, Jason Marquis, Jeff Suppan, and Bob Welch.

However, LaRussa has not had much success with young pitchers.  The White Sox featured a flock of young arms emerge under him, almost all of who had disappointing careers. While drug addiction took their toll on Cy Young winner LaMarr Hoyt, and Britt Burns’s career foundered due to a degenerative hip, Richard Dotson, who went 22-7 in 1983 at age 24, blew out his arm.  Ross Baumgarten earned some Rookie of the Year votes in 1979, but won only seven more games in his career.  Super-prospect Todd Van Poppel was a disaster in Oakland.  Bud Smith came up with the Cards in 2001 and despite throwing a no-hitter, was out of baseball by his 23rd birthday.  Matt Morris was runner-up in the Rookie of the Year voting in 1997, and survived an arm injury to win 22 games in 2001, but then faded out.  Rick Ankiel suffered an epic mental meltdown in the 2000 playoffs, and his pitching career never recovered. 

And as I said just before the excerpt: that’s just starters.  Oakland’s all-world bullpen from 1988-90 consisted almost entirely of castoffs and unknowns.

Yeah, I think pretty highly of Dave Duncan’s pitching coach career.

. . .

Once again, I heartily thank Chris for his well-thought-out and detailed responses to my questions. I think it’s great that he chose to investigate baseball managers as they tend to be enigmas. Due to the many unknowns, we are heavily biased when we reach our conclusions about managers. At last, we have some good old-fashioned facts as a starting point.

Be sure to get your hands on a copy of Evaluating Baseball Managers.