Management by Baseball

Baseball is a feast of metrics, and with all the new forms of measurement sabermetricians are applying to the game, it's no surprise that some of these numbers are models you can apply in your own non-baseball organization.

Here's one of my favorites: Pythagorean Won-Loss Records.

While the actual won-loss standings establish who ranks where at the end of the season, during the season, chance, luck and the Yankees can deflect the actual field accomplishments of a team to give them a record better (or worse) than how they are playing. That shouldn't surprise you. Even with a long 162-game season, some teams are going to lose or win more than their fair share of closely contested games. That's the way it should be.

Back in the early 1980s, Bill James, the most well-known sabermetrician of that era found that in general you could predict how many games a team would win and lose over the season by knowing how many runs they scored and how many they surrendered. By running these through a simple (though requiring a calculator) formula, you could guess within two games for most teams how many wins a squad would have by the end of the season. The method is called Pythagorean Won-Loss Records, a.k.a. Expected Won-Loss Records.

But who cares, some might ask? The only thing that matters in getting to the playoffs is how many games you did win, now how many you should have won.

True, but the slick thing about the Pythagorean metric is it has strong predictive abilities. Because nature and the tug of probabilities tends to draw most things towards the mean average, luck tends to even out over a long season. So it turns out a team that 30 games or so into the season is way underperforming or way over-performing the wins they could be expected to have based on their Runs Scored (RS) and Runs Allowed (RA) is very likely to move in the direction of their Pythagorean. And that they will win games in the future part of the season at the rate more closely to their Pythagorean won-lost record than their actual-to-date record.

Alan Schwarz wrote a piece the Contra Costa (Calif.) Times published July 4th about this very metric. He explains the formula and some of the teams that appear ready to slide and those that appear ready to make a strong move, based on their Pythagorean numbers at the All-Star break. It's a well done newspaper article, though the on-line version would benefit from a Standings table that illustrated every team's actual and Pythagorean won-loss records.

TUESDAY EVENING ADD Studes, the Stats Stud from Baseball Graphs, pointed out that they have an unusual graphic representation of RS and RA, with Pythagorean notations. I looked and found it very informative -- it's less numeric but more visual. Take a look at this if you're interested.

Baseball Prospectus runs a daily chart that does this, though before you go there, it does too many other things too, leaving no tern unstoned in an effort to rationalize in one place everything from The Unified Field Theory to why the majority of people in an ice cream parlor that sells over twenty flavors will still choose vanilla to Tony Batista's batting stance to why people still use Microsoft's Internet Explorer when a thousand deranged orcs spend half their waking hours obsessively writing code that turns it into a death trap for its owners.

The part of the Baseball Prospectus chart to look at includes the first seven columns (feel free to ignore the rest unless you like clever numbers, which the rest are). Those seven columns are: Team W L RS RA W1 L1.

The RS and RA I explained already, and the W1 and L1 are the wins and losses the Pythagorean calculation projects the team should expect based on their RS and RA. Here's Baseball Prospectus' standings table. The short look: Teams one should expect to see improved performances from in the second half: The Red Sox, Blue Jays, Tigers, Mariners & Pirates. Teams one should expect to see erode a little in the second half: The Yankees, Reds, Giants and Twins. Of course trades and violent strategy shifts in reaction to the real standings can change the probabilities for teams in the second half, though Schwarz' article suggests some teams do observe Pythagorean metrics to tune their second half behavior.

The Reds' (director of baseball operations Brad) Kullman has the ultimate challenge: His club, hanging tough at 42-37, has been outscored by 30 runs and is probably due for a fall. "It's tough to say, 'Let's trade for the stretch drive and mortgage the future,' when realistically that might not be the best move," he said.

A week later, the Reds' Pythagorean won-loss measure was 40-48 according to the Prospectus table. And some franchises (well, the Twins) seem to go against their Pythagorean tendency a lot of years. There's no good theory, even a glib one, that explains the Twins' seeming ability to do a Nijinsky and defy the laws of gravity that seem to bind other teams to this dark-hearted orb.

In spite of its many exceptions, Bill James Pythagorean projection method is a fine general metric. I love it for several reasons, all of them illustrative of things that are valuable to non-baseball organizations.

BEYOND BASEBALL

1: The Numbers Behind Success

Non-baseball organizations need to track their progress with certain numbers and, like baseball's wins and losses, these really do matter. But frequently they don't match the quality of the group's efforts. How many times have you run a project or new initiative you know was really good and well-executed but the bottom line hasn't shown equal-quality results in the short term? Too often. Sometimes your competitor does something half-axed and blows away the market. The bottom line result doesn't always reflect the quality of the decisions and effort that went into the efforts.

Quality, clarity of vision, persistence, ability to change plans adaptively on the fly tends to win over the long term, but guarantees nothing over the short term. If management knows what the constituents of successful performance are (in baseball, runs scored relative to runs allowed), they can keep track of these key components and ratios as well as the obvious surface indications of success and failure. They can use these measures to take a longer view and apply their resources more intelligently.

Tools like balanced scorecard are good starts in this direction, but that's usually something analysts apply broadly across an organization. If you think about it, you can devise tools like that to a department or workgroup with quicker results.

2: Natural Ratios

Lots of managers believe in "S" curves of product adoption. Some things actually do work out that way, though most don't. An incredible number of people believe in Bell Curve distributions even though very few things in nature actually fit a bell curve pattern, including most of the classic examples we were taught in school like "average height of adult males".

The Pythagorean ratios are actually very common in nature, not to mention algebra. It's almost like something out of The Eight or the Da Vinci Code.

You can try to discover Pythagorean ratios in your own organization's measures and use them to gauge performance. Of course, the work of Pythagoras is not alone in clarifying natural tendencies -- you might discover your own if you pull yourself away from the exlicit and implicit assumptions of standard measures.

3. Breaking the Stranglehold of the Past

By breaking measures into two parts, the actual and the expected based on performance, you're analysing things the right way instead of the traditional way othat turns it on its head. It's easier just to measure final results, but ultimately far less informative to shaping the way you are executing than measuring the constituents of success. Existentialist manager Alvin "Swamp Fox" Dark lived by this system when he managed the K.C. Rotals and Oakland A's, making sure he didn't beat himself to death for every decision that didn't work out because you have to remember not all right decisions work out, not do all wrong ones cause you to fail.

Are you doing the right things in the right proportions? How do you know? Because results and efforts are always out of synch to some degree, results can pull management to violently in some short-term direction based on a factor that was outside the effect the organization can have. This isn't an argument, btw, for inaction in the face of change, it is though an argument for not allowing panic to undermine the things you do well or allowing complacency over good results after mediocre efforts lull your organization into a sense of invulnerability.

Baseball teams have big decisions to make this time of year. Is the team good enough to compete for the rest of the year, or is it going to become a black hole from which no light escapes?

The won-loss record is indicative, but the Pythagorean metric really tells a fuller story. It can for you, too.