Saturday, May 29, 2004

Kauffman Stadium, Not Wall St.,
Shows the True Nature of Markets  

The religious fervor for market-anchored analysis has slipped a little since the beginning of the 2000 stock market hiccup and the 2001 recession, but while society as a whole has eased off on this particular faith quite a bit, in managerial circles and in organizations where finance dominates managerial cognates, lazy adherence to markets is too common. Baseball illustrates both where market faith works but also where it causes the knee-jerk marketeers to implode with as much mess as a Wilbur Wood-sized jellyfish on Jupiter.

Market-based thinking -- assuming mass-markets "know" something any individual organism doesn't while also assuming an individual can get ahead of the market by inside information or quicker reaction to trends -- can be very useful in all forms of strategic and operation decision-making. But not in Las Vegas, it appears, and not when people are trying to gauge ballpark effects.


It is possible as a manager to follow the money to some essential wisdom. I used market-based thinking, for example, when I founded the InfoWorld Test Center. I was installed to do a myriad of things but one of them was to establish a testing system that would kick the ax of the better-funded market leader, PC Magazine. PC Mag's Lab was run by very smart, very educated, very ignorant of real world computer use guys (I mean guys...I don't think they had ever had a single female working there when I started the Test Center). They measured computer execution speeds in 1/100ths of seconds, sometimes even 1/1000ths of seconds -- and not a set of operations chained together, but a big process, like repaginating a 100-page word processing document, and they would intone portentously that word processing program "A" was "faster" at 9.17 seconds at this task program "B" at 9.24.

They had human beings with stopwatches trying their darndest to time these things. They failed every time -- it was a foolish model, substituting insignificant digits for thought-out analysis. The reality is, the market proved that that "Men With Stopwatches" model could only measure accurately to the nearest 1/5th of a second. And I knew this because of markets: The highest-value money application for "Men With Stopwatches" was thoroughbred horseracing, where very bit of information that a bettor or bookie can harvest is something they'll apply to the billion-dollar endeavor. But in horseracing, human-stopwatch times had a 1/5th second increment...even with millions riding on a single race, and both bettors and the house looking for edges, the market knew human accuracy operating a stopwatch was not 1/1000th of a second or even 1/100th of a second or even 1/10th of a second, but 1/5th of a second (0.2 seconds).

Clearly, the PC Mag guys knew nothing about horse-racing. Clearly they knew nothing about end-user computing either, because perhaps one in five hundred end users would ever care about .07 seconds lost in re-paginating a 100-page document in their Wordstar 2000. They had not only gotten the wrong answer, they'd asked the wrong questions.

In the InfoWorld Test Center case, those of us who worked in the lab got together and immediately started a research project to use automatic timing devices (those PC clocks measured in 1/18ths of seconds), aiming to remove the human error factor, and we started reporting no closer than 15ths of seconds (the human perception interval).

In that case, the market based approach -- assuming the largest application of money based on human operation of stopwatches -- worked. But it's not automatically useful.


Baseball Primer cited an article in the Las Vegas Sun's regular sports gambling feature about park factors (the way a stadium affects offensive events), which I've discussed in this weblog before, most recently here. Of course, sports bettors and bookies are most interested in park factors because the way a ballpark promotes or deflates offense affects the "over/under" line, a bet based on how many runs a pair of teams will score combined. The bookie picks a sum and bettors gamble on whether the actual teams' combined scores will be more or less than the over/under number.

Because gambling is such a big industry in the U.S., I'm suggesting the biggest market application for park factor information is Las Vegas sports bookies. On the surface, it appears the bookies are using a classic business intelligence approach, gathering data to shape strategy. As story author Jeff Haney notes

Whenever a new baseball stadium opens -- this year there are two of them, Petco and Philadelphia's Citizens Bank Park -- bettors scramble to determine whether it favors pitchers or hitters, overs or unders.

While Padres fans -- and even some San Diego players -- have focused on the daunting 411-foot power alley in right-center field at Petco, from a betting perspective the new park is close to the league average one-quarter of the way through the season.

In 22 games at Petco, 11 have gone over the posted total; 10 have gone under; and there was one push. And those totals have not been freakishly low. The average over/under at Petco so far is 8.09 runs. In four of the past six games there, the total has been set at 8 1/2 runs, a typical number for the National League.

"It looks like 8 or 8 1/2 will be a pretty good figure for this ballpark," White said.

Petco's perception as a pitchers' park could stem from early-season slumps that plagued several of the team's high-profile hitters, White said.

"Some of the big Padres bats are left-handed," and those players have had to adjust to Petco's wide-open spaces in right field, White said. Left fielder Ryan Klesko, for instance, has only one home run this season, and right fielder Brian Giles managed only five hits in his first 45 at-bats.

Games at Citizens Bank Park, meanwhile, are 13-5 in favor of the over, with a couple of pushes.

In April it was common to see totals of 7 1/2 in games at Citizens Bank Park. More recently they have been in the range of 9 to 10 runs, and White expects those totals to continue rising as a hot, sticky summer begins to smother South Philadelphia.

"We'll see some more 10 1/2's as soon as the weather gets better there," White said.

Sensible enough. But to be efficient, bookies need to find an effective sum to set as the middle so they can balance their risk. If they make a habit of setting games set in a park too high or too low, they make it too easy for a smart bettor to bet the easy side and the bookie won't have enough margin from the favored side to make a profit or simply even break even.

And it's clear that most of Las Vegas is flushing its resources away faster than a retiree pouring quarters into a one-armed bandit. The article cites some venues where over/unders are not balanced but out of whack and costing bookies big money:

Snapshot: Baseball over/unders

Stadiums where "overs" have ruled in 2004 ...

1. Citizens Bank Park,.13 overs to 5 unders

2. Oriole Park at Camden Yards, .15 overs to 7 unders

3. Metrodome, .13 overs to 7 unders

4. SBC Park, .13 overs to 8 unders

... and leading "under" parks so far:

1. Kauffman Stadium, .5 overs to 12 unders

2. SkyDome, .8 overs to 14 unders

3. Wrigley Field, .8 overs to 13 unders

4. Turner Field, .7 overs to 11 unders

The snapshot numbers indicate two common business intelligence problems: The Citizens' Bank Park Problem and the Kauffman Stadium Problem.

The Kauffman Stadium Problem I sometimes call the "What is Past is Prologue" problem, a lazy belief that whatever .happened in the past is likely to keep on going, and that trends will continue in the same direction. Kauffman Stadium has been a blisteringly offensive park for a couple of years. So bookies seeing games scheduled in that park have overestimated the number of combined runs for games there 12 times out of 17 games. The Royals fiddled around with the park's configuration in the off-season (reported on widely) to reduce the extreme nature of the offensive boost, so the environment of the park has changed. Bookies' set points didn't though. Facing the unknown, the bookies found it easier to do nothing than try to steer their center point to a new place.

This happens all over baseball (Steve Nelson at Mariners' Wheelhouse weblog wrote about how this effect undermines management in building a squad -- back in late February, he suggested the Mariners would be disappointing this year because of the Ms' front office's inability to get beyond their Kauffman Stadium Problem, a prediction that, so far, looks hauntingly true). It happens outside baseball as well. I used to work as a journalist at a weekly newspaper, and we'd get free "analyst" reports. The analysts hoped we would quote them, making them better-known, convincing potential buyers of analysis that it was worth a ton of money to buy their reading of sheeps' entrails. And, of course, editors liked the reporters to use this slop because it looked important, and it was cheaper and easier than having to do independent research. But more often than not, analyst numbers are laughable. I read one study that took a three-year trend line of early sales of home computers, and just made a straight line out of it. By the year 2003, there should have been 3.4 home computers being used per person in the U.S. Anyone but an analyst would have figured out the infeasibility of that, but it was eay to just assume things would stay the same, and no one in the chain of command bothered to look at the numbers for sense, because the line looked so...cute.

The Citizens' Bank Park Problem is a little different. It's solidifying a solution based on a little scattering of early data. Very early on, that park was favoring pitchers. Whether it was the weather or the wind or whatever, scores were lower than they have been more recently in the season. And the bookies have been setting the line too low.

Las Vegas author and handicapper Andy Iskoe, who keeps close tabs on baseball totals, said he likes to wait until later in the season before reaching a verdict on new stadiums.

"There's always a rush to be the first to make a conclusion on what kind of biases a stadium may have," Iskoe said. "Probably, statistically, you have to wait several seasons to make a determination.

"But no one wants to wait several seasons. No one wants to wait even several weeks. ... But when you're looking at only one-quarter of a team's home schedule, that's an awfully small database."

While the Phillies have allowed an average of 4.65 runs per game at home and 3.91 runs on the road, they have scored 5.32 runs per game on the road, compared with just 5.15 at home.


The market screams to pay attention, but the data is to scant to read. As Iskoe notes, in the rush to "know" bookies and bettors are driven by the market to ineffective decisions. Bettors probably shouldn't be trying to guess over/under numbers for new parks most of the time. And when you combine this Citizens' Park Problem -- deciding on scanty data -- with the Kauffman Stadium Problem -- finding it hard to change an opinion once made -- you get instant organizational suicide: Premature matriculation followed by unyielding adherence to the distortion.

In our culture, the religious fervor for market-based faith drives people to market-based analyses sometimes without a lot of close, skeptical questioning. Like a great hitter, sometimes the market is safe, and sometimes it whiffs. But I've noticed in my practice that frequently, the more money there is riding on a decision, the more temptation there is to rush to judgement with scanty data. It happens in baseball, in Las Vegas, and it may be happening in your organization.

This page is powered by Blogger. Isn't yours?

free website counter