The Algorithm Method | Media | Chicago Reader

News & Politics » Media

The Algorithm Method

Hot Type's coveted Golden BAT award goes to a computer program.

by

Sign up for our newsletters Subscribe

comment

Consider Jay Mariotti: he's long been the blithe spirit of the local press box, yet when he competes for the Golden BAT a curse is upon him. You'll recall that Mariotti's 2000 pennant picks earned him first place in this column's assessment of horsehide foresight. But in 2000 the U.S. Supreme Court decreed that there was not only no shame but a presidency to be had for finishing second, and in deference to that shift in the zeitgeist, when the time came to declare a winner the next spring I renamed the coveted honor the Dimpled Chad BAT and gifted it to Teddy Greenstein, who'd trailed Mariotti by so little that what the hell.

Three years later Mariotti and Greenstein finished in a dead heat. Thinking it only just, I tilted the scales in Mariotti's favor. But the gesture left him cold—unless he's certain he's earned hosannas he'd rather not hear them, thank you, and in every conversation we've had since he's threatened to sue me for something or other (though come to think of it, that was also true before).

At any rate, the results of this year's Golden BAT competition are in, and Mariotti has again been denied the unalloyed triumph that would cap his career. He and the defending champ, the Tribune's Paul Sullivan, tied for first among local baseball writers, but it is not merely that Sullivan nipped him in the tie-breaker. Both, alas, have been shown up by an algorithm.

For the benefit of children in their parents' laps, hearing of this storied spring ritual for the first time, some history is in order. The BAT was introduced in 1981 by Hot Type's custodian at the time, Neil Tesser, in a spirit, it must be admitted, of derision. It was Tesser's theory that when it came to measuring baseball talent the so-called solons of the press box were no better qualified than the average drunken Cubs fan pissing against a lamppost on Fremont Street. Time has not disproved this thesis. But it has smoothed the edges of the BAT—that's short for Baseball Acumen Test—and the once dubious honor has evolved into the most distinguished token of achievement most sportswriters can hope to achieve.

Stripped of its subtleties, the BAT simply requires that I review the list of the previous year's division winners and wild card teams and calculate which local scribe predicted the most of them. Tie-breakers take into account pennant winners and World Series champions and, every so often, a presidential election victor.

So let's review. In 2007 these teams won division championships: the Boston Red Sox, the Cleveland Indians, the Los Angeles Angels, the Philadelphia Phillies, the Arizona Diamondbacks, and, of course, our Cubbies. The New York Yankees and the Colorado Rockies were the wild card teams. The Red Sox defeated the Rockies in the Series.

It was an unexceptional year for the locals. Rick Telander and Chris De Luca of the Sun-Times and Dave van Dyck of the Tribune foresaw the Cubs' NL Central Division championship, while the Sun-Times's Toni Ginnetti had them as the NL wild card, but it was a love-is-blind performance by Telander and van Dyke, who also picked the White Sox to lead the AL Central. De Luca stumbled in other races, while Ginnetti was the only one of the 14 competitors from the two papers to leave the Red Sox out of the playoffs altogether. (On the other hand, she did see the Indians prevailing in the AL Central, and no one else even had them in the postseason.)

There was a clear, unflattering pattern to the competition. The scribes favored the usual suspects, and in the AL, where squads like Boston, New York, and Los Angeles performed as expected, they racked up points. But in the NL, upstarts like the Cubs, the Diamondbacks, and the Rockies came to the fore, and nobody saw that happening. Only Sullivan, who called the Phillies and the Diamondbacks correctly, named more than one NL playoff team. Sullivan was also right about the Angels winning the AL West; and he (and just about everyone else) predicted the Yankees and Red Sox would both reach the playoffs, though he was wrong about who'd win the AL East and who'd be the wild card.

Sullivan named five playoff teams, Mariotti only four—the Phillies, the Red Sox, the Angels, and the Yankees—but Mariotti nailed his exactly. (It's only half a point if you name a team that made the playoffs but are wrong about how it got there.) The tie-breaker was the playoffs. And while Sullivan had Boston overcoming their wild card berth to beat the Phillies in the Series, Mariotti went off the rails and chose the Angels to defeat the Dodgers.

Anyway, none of this matters—thanks to the new sheriff in town. Let me begin the introduction by quoting something van Dyck wrote in an end-of-2007 retrospective. "Most reader response I got—In spring training fans were irate about a story from Baseball Prospectus that projected the White Sox to finish 72-90. 'That's a good sign for us because usually they're wrong about everything regarding our dealings,' GM Ken Williams said. The White Sox final record: 72-90."

Bear in mind that even chess genius Garry Kasparov can't beat computers anymore. In this case lesser humans are up against Baseball Prospectus's secret weapon, PECOTA, which stands for Player Empirical Comparison and Optimization Test Algorithm. It's an acronym that may one day rival even BAT for the affections of America's sporting public. Baseball Prospectus didn't just rank the major league teams as the 2007 season began—its computer crunched the numbers and predicted how many games they'd win and lose. And it did a very good job.

The father of PECOTA is Chicago's Nate Silver. "The idea behind the system," he explains by e-mail, "is really just to use baseball history to inform our projections in the form of comparable players. So for Mark Buehrle, we might look at other durable left-handed pitchers like Jim Kaat, and for Alfonso Soriano, we might look at speed/power outfielders who struck out a lot, like Joe Carter. By seeing how the comparable pitchers performed, we can have a way to predict future performance that is both intuitive and accurate."

And it's constantly being tweaked. "This year," Silver says, "we started looking at things like platoon splits in more detail, and first/second half splits—if a guy performs better after the All-Star break, that's a good sign for the next season. But there's no such thing as a 'perfect' forecasting system."

Even so, it seems likely that Baseball Prospectus, with its computer and mountains of data, will come closer to one than sportswriters, who've been known to scribble their picks on coasters ten minutes before they're due. Silver says the PECOTA projections represent a "solid block of six weeks or so of work."

Silver describes himself as a former underemployed consultant who "worked with a lot of spreadsheets and statistical models." He continues, "So I started working on PECOTA—which looked just like one of the models we might put together for our clients, meaning my bosses wouldn't give me any trouble when they walked by." After months of tinkering, Silver went public with PECOTA in 2003.

A year ago van Dyck interviewed Silver about the historic 72-90 White Sox prediction. Van Dyck then commented, "The scary part is that the computer can be accurate much of the time. It projected five of the six [2006] division winners last spring and predicted the Detroit Tigers would finish with a better record than the defending champion White Sox."

En route to his first Golden BAT, Sullivan also had five of the 2006 division winners. But this time around PECOTA has left him in the dust. It picked four of the 2007 division champs, including all three in the treacherous National League, and also predicted that both the Yankees and the Red Sox would reach the playoffs—it merely had their order of finish wrong. So Silver's system named six of the eight playoff teams, and it had a seventh, Cleveland, finishing a game out of first. PECOTA called the NL Central for the Cubs with an 86-76 record. They won it at 85-77.

PECOTA's only conspicuous stumble was over the Colorado Rockies. PECOTA projected 79 wins for the Rockies, but they squeezed into the playoffs with 90 and went on to win the NL pennant. Silver's given the matter some thought, and he says PECOTA underestimated the importance of the defense the Rockies added last year, such as rookie shortstop Troy Tulowitzki. "Plus," he added, "it was also one of those cases of everything sort of coming together. There seem to be one or two of those every year, like the White Sox in '05, and they're very hard to predict ahead of time."

I don't know whether to allow Baseball Prospectus into future BAT competitions. But here's what has me worried. What's to stop some desperate scrivener who seeks Golden BAT glory from simply copying the projections of Baseball Prospectus? Is there any way to prevent such an assault on the integrity of the Golden BAT?

No, Silver said.

One last piece of business—the Wiffle BAT, given each year to whoever winds up last. Perhaps there's a HAL out there, a willful computer that played a hunch and picked Pittsburgh to knock off Kansas City in the Series. Could be, but I don't know about it. So this year the consolation prize is again a reward for being only too human, and it goes to the Tribune's Phil Rogers, who correctly named New York and Boston (in the wrong order) as playoff teams, and no one else.

Being There

There are reporters who play the telephone the way Heifetz played the Stradivarius, but sometimes you just have to be there. Stacy St. Clair, a Tribune reporter on duty Easter morning, was sent over to Holy Name Cathedral because assignment editor Jeff Carlson thought there might be a story in the fact that mass was being held in the auditorium while the sanctuary was being renovated. Would there be enough room for everyone showing up for the best-attended service of the year? Would St. Clair spot bewildered once-a-year worshippers rattling the locked cathedral doors?

St. Clair took some notes outside and then slipped into the auditorium for Cardinal George's homily. That's where she found her story. "Often we hear—we even say it—that love is blind," George began. "Someone might look at a friend's fiancee—" And he said no more, for that's when the people calling themselves Catholic Schoolgirls Against the War stood up near the rear of the auditorium and began shouting. More than a million Iraqis have been killed during the war, one of them cried; another hollered that on January 7 the cardinal had met George W. Bush for lunch; and then they began spraying fake blood and somebody screamed. Within seconds the protesters were escorted out the back door, chanting "Even the pope calls for peace" as they went. WMAQ TV would call it a "bizarre and shocking demonstration."

Because St. Clair was unreachable when word of this outburst got to the newsroom, Carlson sent over a second reporter, Erika Slife. When I read the Monday morning story the Tribune carried under their double byline, I assumed they'd re-covered it by phone until I realized that even a Heifetz would have had trouble accumulating so much precise detail and intimate comment that way. So the Tribune was at the scene! Why? Had it been tipped off?

Carlson understood the point of my call even before I got to it, and he insisted the Tribune had not. This was no small matter, for if the paper had known what was coming and had done nothing but make sure it was there to see it, it would almost certainly have exceeded the public's capacity to appreciate dispassionate witness. Possibly Catholic Schoolgirls Against the War assumed the media would be out in force for Easter mass at the cathedral, but if so it assumed too much. Yes, the TV stations were on hand when the mass began; but by the time Cardinal George began speaking most had shot their footage and packed up. Only WMAQ and CLTV captured the fireworks.v

For more on the media, see Michael Miner's blog, News Bites, at chicagoreader.com.

Add a comment