One of the best-kept secrets in Cleveland sports is not who the Cavs will take with the first pick in the NBA draft, or who the Browns’ starting quarterback will be come fall. Rather, it is a computer program you've probably never heard of. But for years, the Indians have used a proprietary statistical analysis program called DiamondView. This program aids decision-making for the Cleveland Indians' front office, and it is a reason for numerous highly publicized endings to players' careers with the Tribe, including those of Jim Thome, CC Sabathia, Victor Martinez and Cliff Lee, to name a few.
Before Moneyball, there was DiamondView. As analytics have been adopted across Major League Baseball, DiamondView has stayed exclusive and evolved within the confines of Progressive Field, allowing the Indians to stay competitive as a small-market team during an era of extreme salary inflation.
One of just a few articles about the program was written in the Cleveland Plain Dealer in 2003. When it came to negotiating a contract extension with Jim Thome, the Indians felt uncomfortable offering the slugger more than five years. Thome’s new salary also would have exceeded 12.5 percent of the team’s payroll, a threshold the Tribe brass felt uncomfortable exceeding. General manager Chris Antonetti explained that Barry Bonds was the only hitter in the 22 prior seasons without a major decline in production after the age of 35, and that Thome would play the final two years of a five-year contract at ages 36 and 37.
The Indians were willing to risk three years of Thome playing at age 35 and beyond, but not four. For that reason, the Tribe's hero of the '90s jumped ship and left for Philadelphia, where he got the extra seasons under a contract he sought. In those six seasons of his new deal, three with the Phillies and three with the White Sox after a trade, Thome hit just .265 and missed an average of 32 games per season, numbers down from his Cleveland peaks. The extra seasons the Phillies gave to Thome became a clear overpay for a .245 hitter who could no longer field a position regularly. In his final six Cleveland seasons, Thome posted 29.1 wins above replacement, compared with just 17.2 WAR during his six-year contract. Though still productive into his late-30s, Thome was nowhere near the player he was at his peak, and his contract would have proved problematic for a smaller-market team.
Instead of paying Thome what would have amounted to $63 million over five years, the Indians spent that money on bonuses for draft picks, along with rising salaries for players such as Sabathia, Jake Westbrook and Matt Lawton. When those players became too expensive for the Indians to afford and too old for DiamondView to believe in their continued success, the cycle started again, with Sabathia, Victor Martinez, and Cliff Lee traded for prospects ahead of their impending free-agent paydays.
The creation of DiamondView was revolutionary because of the program's ability to decipher large amounts of data very quickly, updating its statistical database on a daily basis while also being able to point out trends and other important information. Though this does not sound extremely impressive today with the numerous sports databases at our disposal, DiamondView was a state-of-the-art analysis program upon its inception in spring 2000. In 2003, Antonetti was giddy about the team’s creation of the statistic OPS-plus (not to be confused with OPS+), an early metric used to better quantify and predict offensive performance.
Former Tribe outfielder Matt Lawton was thought of highly by DiamondView, leading the Indians to sign him for four years and $27 million in late 2001. Lawton disappointed in his first two seasons beside the shores of Lake Erie, battling through injuries and inconsistency, but he was an All-Star in 2004 once his health problems subsided. The performance roller coaster that Lawton rode on during his seasons in Cleveland was used as a way to refine and improve DiamondView. This mistake became a teaching point for the database, as it incorporated potential warning signs for future decisions.
Another interesting finding from the early days of DiamondView was how it tracked and forecasted the development of power hitting, and how it appears later during players’ careers. That has been exemplified this season through the power surge of Michael Brantley. The Tribe outfielder has hit nine home runs in 48 games this season after hitting just 10 in 151 games in 2013. His doubles rate is also up, leading to a triple-digit improvement in slugging percentage. Blue Jays first baseman Edwin Encarnacion is another key example of the later-career power surge, as he did not post a slugging percentage above .500 until his age-29 season.
The age of 29 is important to note, as this is the age that the Tribe front office has pinpointed as a player’s performance peak. Generally, players begin to decline once they turn 29, with a slower rate of decline making a player more valuable. During winter 2012, I had the opportunity to travel to the corner of Carnegie and Ontario and visit with multiple members of the Indians' front office. Normally tight-lipped in all aspects, the front-office members who spoke with my game theory class, including Antonetti and team president Mark Shapiro, talked a lot about DiamondView and their reliance on it, even sharing some of the secrets the program has revealed. They elaborated on the importance of DiamondView and divulged some fascinating factoids, as well. This “29 and decline” rule was one of the points mentioned by the analysts and Antonetti during their lectures.
Looking back at the first major trade of a talented Indians starting pitcher during the DiamondView era, Bartolo Colon follows the data strongly. He was 29 when the Indians shipped him to Montreal, with DiamondView proving its worth on both ends. Colon became a free agent after the 2003 season and signed a four-year contract with the Angels. That contract saw Colon post a 4.66 ERA in 96 appearances, including a 5.90 ERA in the final two years of the deal. And I don't think I need to go into detail with the Tribe’s acquisition of All-Stars Cliff Lee, Grady Sizemore and Brandon Phillips in that trade.
Antonetti focused on the trade of former American League Cy Young award winner CC Sabathia in summer 2008, mentioning how top prospect Matt LaPorta was rated highly by both traditional scouting and DiamondView’s analytics. They believed that he had a high probability of becoming a middle-of-the-lineup power hitter with All-Star potential. The scenario played itself out with LaPorta ended up as a bust, most recently being granted his release from a Mexican League team earlier this season. During the meeting, Antonetti mentioned that, statistically speaking, if the career of Matt LaPorta had played out 100 times, he would have been a successful player on 75 or 80 occasions. He also made it clear that there are no do-overs in the real world and that the team had to accept the outcomes handed to them.
On multiple occasions, the front office mentioned that, though computers can give the team an advantage, intangibles may also affect a player’s career path. DiamondView has a database of off-field information on players, but this information cannot be quantified as simply as on-field statistics. Antonetti and Shapiro also stressed that DiamondView is just one tool of many when it comes to player evaluation and analysis. The pair stressed that other important aspects of team building include character, work ethic and team chemistry. It explains the emphasis on good-character player signings, namely Nick Swisher and Jason Giambi.
In my research of DiamondView, it becomes clear that sample size is the biggest factor in the program’s accuracy. The Indians have struggled to draft successful players since the program’s inception in 2000, most likely due to an inability to contextualize the few statistics college players have at a level much lower than any in minor league baseball. With high school sample sizes even smaller than college, the Indians have to rely more heavily on typical scouting. Even just one or two years of minor league statistics lead to a better rate of success for the computer program, as shown through plenty of heists of minor leaguers in recent years. Once a player is part of a professional team's infrastructure, his intangibles can also be better understood. At the macro level, as DiamondView continues to mature and the database absorbs more information, its overall accuracy improves because it can learn from its mistakes.
But has DiamondView been ultimately successful? The short answer is somewhat. Since DiamondView’s creation in spring 2000, the Indians have a record of 1149-1171 (as of May 26, 2014). However, among smaller-payroll franchises, the Indians have the fourth-best record in the time frame, behind only the Oakland Athletics, Chicago White Sox and Minnesota Twins. This means that the brainchild of the Shapiro/Antonetti regime has not been able to fully nullify the disparities that differing payrolls create, but it has allowed the Indians to be competitive among franchises in similar situations.
I am only scratching the surface on DiamondView, what it does and how it helps the Indians succeed on the field. I have never seen the program in action, relying on primary and secondary sources to learn about it. Moreover, Shapiro and Antonetti are two of the best in the business at staying secretive. As a small-market franchise, the odds are stacked against them, with almost no room for error. DiamondView has granted the Indians improved results in the 15 seasons since the program first came to fruition.
DiamondView is a known but not completely understood commodity within baseball, with some franchises even resorting to the creation of DiamondView knockoffs. Former Indians scouting director Josh Byrnes even offered top prospect Carlos Quentin in a trade for a copy of the program. The Indians said no, as mentioned in "The Yankee Years" by Joe Torre and Tom Verducci. Both authors praise the Indians for becoming the first team to adopt a giant searchable database of statistics and information. Antonetti and Shapiro do not get enough credit for being ahead of the curve and adapting to the challenges small-market teams face.
DiamondView was a revolutionary database when it was created, even predating Moneyball. While the secrets exposed in Michael Lewis’ book have been adopted across Major League Baseball, DiamondView’s findings have remained private. The ability for the program to adapt to the evolving nature of baseball’s sabermetric era has kept the franchise consistently competitive among constantly lessening odds for small-market teams.