02/16/10 12:00 AM EST
Before games, prognosticators clash
Sabermetricians weigh in on 2010 season, spurring debate
By Doug Miller / MLB.com
Those projections for the 2010 season are brought to you, respectively, by the increasingly familiar yet still difficult-to-understand statistical projection systems known as PECOTA, Marcel and CHONE.
These, along with others called CAIRO, ZiPS, FANS and maybe a few more that don't have strange anagrammed names, are soon to be the talk of the preseason.
What they all have in common is the ability to provide at least a general barometer for the current paths Major League clubs appear to be on -- and to raise the blood pressure of their hardball-obsessed followers.
And boy, are they fun to talk about.
"Baseball fans, and people in general, love to argue," said Dave Cameron, who analyzes stats for the popular site FanGraphs. "You can write absolutely anything, negative or positive, and people will manage to somehow accuse you of being biased about their teams."
So far, Baseball Prospectus' PECOTA has published projected 2010 Major League standings, along with CHONE, the product of Maryland-based seamhead -- and Angels fan -- Sean Smith, and CAIRO, which is available on the baseball blog Revenge of the RLYW.
Meanwhile, ZiPS, created by sabermetrician Dan Szymborski of the Baseball Think Factory site, and Marcel, the brainchild of Seattle Mariners stats consultant Tom Tango, have released projected individual player results.
LOOKING INTO THE CRYSTAL BALL
|AL EAST||BOS (94-68)||NYY (98-64)||NYY (99-63)|
|AL CENTRAL||CWS (80-82)||MIN (86-76)||CWS (87-75)|
|AL WEST||TEX (87-75)||TEX (85-77)||TEX (82-80)|
|NL EAST||PHI (89-73)||ATL (89-73)||PHI (90-72)|
|NL CENTRAL||STL (88-74)||STL (92-70)||STL (91-71)|
|NL WEST||COL (88-74)||LA (84-78)||LA (91-71)|
in NL West
in AL West
|FLA finishes 72-90|
|RISING STAR||J. Soria|
(.303/25 HR/84 RBI)
(24 HR/81 RBI)
All told, the findings are provocative, to say the least.
Let's start with PECOTA, which was invented by Baseball Prospectus co-founder Nate Silver, who devised the projection formula and named it for Bill Pecota, the affable and ultimately forgettable infielder from the 1980s and '90s.
It stands for the Player Empirical Comparison and Optimization Test Algorithm and it uses old-school and new-age hitting and pitching stats and defensive metrics, adjusts for stadium variances and injuries, and compares current players to players with similar stats in baseball history to compile "player cards." When all of the cards are stacked in their proper order on 25-man rosters, the overall MLB standings are computed.
"A lot of what we do at Baseball Prospectus is try and debunk or improve upon conventional wisdom," Silver told MLB.com last year. "You see things about clutch hitting or batting average vs. on-base percentage, and we're trying to build a better mousetrap. Discussion in papers or on radio is not as well-informed as to what really goes into winning baseball games."
Silver's system pulled off a big-time stunner when it nailed the Tampa Bay Rays as projected American League East champions prior to the 2008 season.
Then Silver stunned Beltway insiders later that year when his statistics-driven political projection site, FiveThirtyEight.com, gained national attention by predicting Barack Obama's victory over John McCain in the presidential election by 6.1 percentage points (Obama won by 7.0). Silver has devoted most of his time to that site -- and to a book deal.
That leaves PECOTA in the hands of Prospectus co-founder Clay Davenport, who has ironed out a few bugs and most surprisingly projects the Texas Rangers as the runaway AL West winners, while the three-time defending division champion Angels will finish below .500 at 80-82.
The Red Sox are on top of the AL East at 94-68, two games ahead of the Yankees, and the Rays are in third place at 91-71. Other than those three teams, no other club in the Majors is predicted to crack 90 wins.
Last year, PECOTA nailed the Yankees as World Series winners but also picked the Cleveland Indians, New York Mets and Arizona Diamondbacks to win their divisions. None of those three had winning records.
"The biggest problem with projections is that, on a player level, even the most accurate systems are somewhere around 75-percent accurate, which obviously leaves a tremendous margin for error," said Cory Schwartz, director of stats for MLB.com. "They're useful and provocative for evaluating the relative strengths and weaknesses of players and teams, but there's simply no way to account for the unexpected changes that happen during the course of the season."
Said Davenport: "Doing these projections, we simply can't predict injuries and we can't predict trades that might happen in the middle of the season. But we're driving the conversation until the season starts."
Down in Maryland, Smith is doing the same thing with his just-released CHONE rankings, but he's personally having a tough time with it.
"For the first time since 2006, I had to pick someone besides the Angels to win the AL West," says Smith, who in part named CHONE after his favorite player, former Angels spark plug Chone Figgins, and later gave it the name Comprehensive Holistic Objective Numerical Estimations.
Last year, CHONE also picked the Yankees as the best team in baseball and correctly tabbed the Phillies, Dodgers and Angels to win their divisions. But the system also picked the Indians and Cubs. This year, Smith's Halos just don't make the cut.
"By the numbers, they just come out to be average across the board," said Smith, who also has the Rangers winning the division at 86-76, with the Angels at .500 and five games back.
"For them to win it, they're going to have to get the Ervin Santana of 2008 and the Scott Kazmir of 2007. Also, Kendry Morales is going to have to prove that last year wasn't a fluke and Howie Kendrick is going to have to play well for a full season. And it wouldn't hurt if Brandon Wood did well at third base."
CHONE, like PECOTA, has only three teams in all of the Major Leagues with more than 90 wins -- the Yankees, Red Sox and Cardinals. Smith explained that you'll often see lower-than-expected numbers overall in the projection game, which often doesn't forecast extremes.
"What you have in an actual season is the true talent of the teams and what they actually do," Smith said.
"Some teams get a little lucky, some teams don't, some teams have career years all over the roster, some have injuries. So most estimates are on the conservative side because the observed differences are often much more severe than what really happens."
Tango's Marcel projections, available at TangoTiger.net, might be the most humorously named, taking its moniker from the pet monkey on the TV show "Friends."
He considers it a system that any "monkey" can run because it merely takes into account past performance by players over the three most recent years, weighing the latest seasons more heavily, regressing the performances to the league average based on plate appearances or innings pitched and applying an age adjustment.
There are team projections and there are player projections. The latter are common tools used by fantasy players.
"Our Fantasy Preview projections are a combination of objectivity and subjectivity," said Gregg Klayman, MLB.com's vice president of content developmment. "We factor in all of the usual elements -- recent stats, ballpark effects, surrounding lineup, age, historical trends, etc -- but also tweak our numbers based on gut feel."
CAIRO, meanwhile, has one reasonably surprising result, with the Chicago White Sox coming out on top of the AL Central and the Cincinnati Reds winning 84 games and looking like serious Wild Card contenders in the National League.
Written clearly on the Revenge of the RLYW blog, under the links for the 2010 CAIRO standings and pitching statistics, is a warning that might offer a healthy approach to all of these projections: "... these are EXTREMELY PRELIMINARY and SHOULD NOT BE TAKEN TOO SERIOUSLY."
As Cameron says, the sooner passionate fans can understand that, the better.
"There's a big difference between predictions and projections," he says. "People need to understand that projections are more about understanding the path a team is on, not exactly where they'll finish in 2010. These are not predictions.
"So I would advise people to just have fun with it. Definitely don't lose sleep over it."
Doug Miller is a national writer for MLB.com. This story was not subject to the approval of Major League Baseball or its clubs.