The New Inefficiency

There are two ways to win a lot of baseball games on a tight budget: drafting very intelligently, and properly valuing major league and major league-ready talent. I suppose that luck does the trick as well, but those first two are the only ones a general manager has any control over.

Let’s look at the latter, which can be boiled down to two words: spending wisely. Billy Beane got famous for buying on-base percentage at the bargain bin, and he’s since made a name for himself tracking down the most undervalued defensive players. Both approaches have worked very well for the A’s, but it takes only one other bidder to drive up the price of a free agent—the sort of player who might’ve ended up in Oakland in 1998 (or even in 2005) may not land there today.

So what is the “new” inefficiency? When Beane acquired both Milton Bradley and Frank Thomas, some suggested that he was identifying “bad clubhouse guys” as an opportunity for bargain. It’s a credible theory: If your clubhouse can handle a disruptive influence, why not get the player other teams are shying away from? To take just one example, the Giants have gotten a lot of production from a guy like that. But I suspect that Bradley is wearing green and gold for a different reason entirely.


In Finance 101, you learn that whatever you’re valuing—a stock, a collectible, even a can of soup—has some risk component, and that risk component itself has a value. In general terms, if you buy a “safe” stock, you pay more than you would for an otherwise equal stock with much greater variations in price. That makes sense: Part of the value of a stock is its potential sale price, and if the price jerks up and down, you might not get as high a sale price on the day you choose to sell it.

Likewise, of course, ballplayers. It’s tougher to identify risky and non-risky players who are otherwise equal, but here are a few 2007 free-agent signings who might be considered risky:

Here are few you’d probably consider much less risky, though in some cases, not as good:

Your categorizations might differ from mine, which is fine. What matters here is that we think about players in terms of the likelihood that they’ll do what we expect them to do.

With baseball players, risk comes in many forms. With relievers, the issue is often the variation in their year-to-year performance. Joe Borowski made the “risky” list not just because he’s had some injury problems, but because he’s been mediocre or worse about as much as he’s provided quality late-inning relief. With position players (and starting pitchers, to some extent), it’s more commonly injuries that stand in the way of expected outcomes becoming actual outcomes.

The underlying theme here is that, for any player, there’s a wide range of possible performances. As a starting point, you might look at the various percentile projections on his PECOTA card, but you have to use your imagination, too: That 90% projection means that nearly 10% of players will exceed that. And while PECOTA does forecast playing time, it doesn’t suggest outcomes like Luis Ayala’s 2006 stat line.

Handling Risk

So, if you have to choose between Jeff Suppan for $42 million for four years and a hypothetical fragile Jeff Suppan (who averages 15 starts per year with the same rate stats) for $10 million for two years, which do you choose? Such a calculation is harder than it sounds.

It’s tempting to take a median projection for those two players in terms of Win Shares or SNLVAR or Runs Above Replacement and measure how much you’re paying per win. That’s a good start, but it ignores two key components of operating a baseball team: Replacements aren’t guaranteed to perform at replacement level, and roster spots (whether you’re discussing 25-man rosters or 40-man rosters) are finite.

The first issue is the more important one. Whichever replacement-based stat you use, there are plenty of players who accumulate negative totals. Some of those are decent players having bad years; others are fill-ins who didn’t do the job. When you put, say, Carl Pavano in your starting rotation to open the season, you’re virtually guaranteeing that you’ll have to go to starter No. 6 at some point. That guy, whether a prospect like Philip Hughes or a Triple-A lifer, carries plenty of risk in his own right.

In other words, if you bring in a fragile starter to be your No. 5, calculating your expected outcome isn’t as easy as forecasting his number of starts and runs above replacement and then assigning zero to the rest; you have to figure out the risk profile of your No. 6, your No. 7, your No. 8, and perhaps beyond. That usually isn’t starkly evident in early May, but Yankees fans know exactly what I’m talking about.

The other issue is the size of the roster. The Blue Jays are a good example here. They opened spring training with a number of high-risk guys: newcomers Victor Zambrano, John Thomson and Tomo Ohka along with A.J. Burnett, Josh Towers, and Gustavo Chacin. If you stack up enough players with injury histories and variations in their performance history, you can get to 162 starts. But you can’t keep a dozen starters on your active roster, and unless a lot of them have options (or are completely uninteresting to the other 29 teams), it’s tough to keep that many in your organization, too.

Shannon Stewart, Travis Buck, and Chris Snelling

It’s a challenge to piece together 162 games at one position from more than two or three players, but it can be done. If you have six major league outfielders with occasional injury issues, you’ll usually have three or four of them ready to play, and you’ll rarely have a roster crunch because one or two will be on the disabled list. To some extent, this is the way most teams handle their bullpens every year.

Consider the A’s situation going into this past offseason. The alignment of Nick Swisher, Mark Kotsay, and Milton Bradley with Bobby Kielty in reserve offered a nice possible opening day roster, but came with more than its share of question marks. Those questions were part of the bargain: if Kotsay and Bradley were 162-game warriors like, say, Miguel Tejada, they’d be more expensive, and the A’s probably wouldn’t still have them around.

A Hardball Times Update
Goodbye for now.

To address the problem, Beane signed… Shannon Stewart? There weren’t many riskier choices, but, of course, the other side of that coin is Stewart’s low pricetag. The A’s also knew that they had a couple of outfielders in the minors in Travis Buck and Danny Putnam who they’d trust with occasional big-league playing time. They certainly weren’t going to hand those guys starting jobs, but there was no reason to sign yet another vet to provide the same insurance.

Back to financial terms for a moment, we can look at the A’s outfield in terms of Modern Portfolio Theory. To grossly oversimplify, the idea behind MPT is that you can lessen risk by putting your eggs in a lot of different baskets—specifically, baskets that aren’t all affected by the same external forces. Buying stock in both Google and Yahoo doesn’t lessen your risk, but buying an index fund does.

We don’t have to dive very far into MPT to apply it to baseball because, unless Will Carroll’s database is hiding some very strange information, injuries don’t occur to multiple players based on the same external variables. Sure, a cold month might aggravate more injuries than usual, but the health of Stewart’s leg and Kotsay’s back are independent of each other.

For the most part, then, it would be unlikely that three or four of your six top outfielders would go down at the same time. When they do, a gutsy (or reckless) GM can go all in and take someone else’s risk, as Beane did with Chris Snelling. The more injury-prone players you acquire, the more likely it is that they’ll fill out your starting lineup in any given game. You’re still subject to a potential roster crunch, but we can only discuss so many variables at the same time.

The End of an Inefficiently Long Article

A couple dozen paragraphs ago, I promised you the new inefficiency. You’ve surely figured it out already. In short, it’s the acceptance of risk. That could mean, as in Cleveland’s case, starting the year with a bunch of platoons with the understanding that some halves of those platoons won’t be available for a month here and there. For Toronto, it means fully expecting to use eight or nine starters to get through the year. For Oakland, it means accepting that you may have to improvise to put three outfielders and a designated hitter in the lineup every night.

Ultimately, the amount of risk a team accepts doesn’t matter a whit in the standings at the end of the year. What matters, of course, are wins, and every team is going to get its wins in a different way. I wouldn’t be surprised if Oakland center fielders (or Toronto No. 4 starters, or Cleveland right fielders) outperform a quarter of their peers on other teams, when taken in the aggregate. That might mean more work for the GM: Certainly it’s more stressful to get 162 games of center fielding out of five players than from Dave Roberts. But if Kotsay+Bradley+Swisher+Putnam+Kielty+Langerhans+Snelling produces more than Roberts does, who did the better job running a baseball team?

Putting these thoughts into a workable model to appoximate value is probably where a few teams are ahead of a game. They are certainly ahead of me. To do that, you’d need a much more granular projection system and an ability to work with probability theory that would leave most of us in the dust.

But if one team figures out a way to get far more production out of seven players and $6 million than other teams get out of two players and $10 million, you can bet that the market value of baseball-crazed actuaries is about to go up.

Comments are closed.