Mike Silva's New York Baseball Digest is a blog dedicated to coverage of the Mets and Yankees. The tone of the site is similar to sports talk radio. There is an open and prominent aversion to statistical analysis, except for the occasional pieces by Howard Megdal. Anything you might expect to hear on a Mike Francesca or Joe Beningo radio program is what you will find at NYBD. To Silva's credit, he responds to comments on his posts (both positive and negative) and seems to have interesting guests on his radio show, which I admittedly do not listen to. A few telling snippets from NYBD posts:
Sports use to be a sanctuary away from the tedium of school work. Bored with studying for your biology midterm – turn on the Knicks. Don’t care about the regents? Listen to WFAN. Not quite sure if that is the case today with the hostile takeover of SABR measurements. If there is one area of baseball where I will draw the line on numbers it’s with defense. You have to watch a player to know how he is defensively. Not all groundballs and zones are created equal. With that being said, if you’re into UZR, check out a friend of NYBD James K at Amazin Avenue. He breaks down the Mets team UZR over the years. Watch the hate mail I get on this one. Is SABR a research society or a political party?
The most amazing debate I have seen since entering the world of independent media is the one being waged by those who believe baseball is simply run by numbers. It’s the proverbial tail that wags the dog mentality that quite simply puts numbers out of context... After reading columns like the above you should feel confident that this is a man [Jerry Manuel] that has a very key ingredient: ability to understand the human side of the game. There are reasons why Moneyball has yet to win a championship. Its managers like Ozzie Guillen, Joe Torre, Terry Francona, and Charlie Manuel that have shown it takes more than knowing the numbers to guide a group to victory.
All season we have debated, much to the fans chagrin, Joba Chamberlain bullpen vs. starter. That is no longer the issue today because it’s clear that the organization is fully committed to developing Joba as a starter. If he were going to the pen, they had their opportunity, and the decision was Phil Hughes. Just like the organization did with Melky Cabrera and Phil Hughes last year, they should send Joba down to Scranton for seasoning.
[Note: Silva thinks that 23 year-old Joba Chamberlain, who has the best ERA and 2nd best tRA of anyone in the Yankees rotation, should be in the bullpen or sent down to AAA.]
Considering this, I was not surprised at what I read yesterday, in response to a piece I posted Wednesday, "Players To Avoid This Offseason":
As for using WAR, any stat that using a fictitious player as its benchmark automatically gets thrown in the trash by me. James you had me up until you cited WAR.
In another time, I would have been tempted to dial up the snark and "FJM" something like this. However, that's getting old and rather than mock I'll try to educate.
It is pretty clear Silva misunderstands the concept of WAR ("wins above replacement") and replacement level in general. I'll start with a brief, and hopefully easy to follow, explanation of the latter. A replacement level player's production can be easily replaced by readily available talent at minimal cost to a team. A team full of replacement players would be expected to win anywhere from 45-50 games a season, which varies based on league and some other factors. Current examples of "replacement players" are Willie Bloomquist, Cory Sullivan, Luis Ayala, Willy Taveras, and Nelson Figueroa. Basically half of the Mets roster fits this description. See this piece by Keith Woolner for more on the replacement level topic.
Eric noted in yesterday's Applesauce that Silva's misguided criticism of the "fictitious player" notion could also apply to other statistics as well:
What’s the benchmark for batting average: a nondescript .300 hitter. Sure, there are plenty of actual .300 hitters out there, but when you gauge the competence of a player by his batting average you aren’t comparing him to Emmet Heidrick; you’re comparing him to some fictitious .300 hitter.
I guess we won’t see Mike Silva refer to a player’s batting average anymore. Or RsBI for that matter, since the benchmark for that is a fictitious slugger who knocks in 100 runs.
The major difference between referring to Chipper Jones as a .300+ hitter, for example, and referring to Damion Easley as replacement level is that one concept is widely understood by baseball fans and one isn't. The "fictitious player" argument is not valid, and usually the last resort of the anti-saber crowd looking for any excuse to offer criticism while pretending to understand the idea.
This takes us to WAR. The goal of the stat is to determine how many "wins" a player contributes above or below that of the theoretical replacement player. A win is equal to 10 runs, as explained here, so the calculation of WAR is actually driven by how many runs a player contributes compared to the replacement level player, whose WAR is 0. Factoring offense, defense, and adjustment for position and playing time, a run value is derived. Without going too calculation crazy, the most important takeaways here are the adjustment for position and the importance of defense. Certain positions are tougher to play than others. Generally, the tougher positions do not feature as many great hitters as the easier positions. The defensive spectrum, from easiest position to play to hardest is as follows: DH-1B-LF-RF-3B-CF-2B-SS-C. As such, players on the right side of the spectrum get more credit for playing more demanding positions. Additionally, high-level defensive metrics like UZR and +/- aim to assign a run value to a player's defense. When it is said that Endy Chavez is a +10 defender, this means he saves 10 more runs with his glove than the average player at his position. These are the factors that make WAR quite useful for comparing values of players who play different positions.
A good illustrative example is a comparison of Hanley Ramirez and Ryan Braun this season. They have posted the same OPS (.961) in a comparable number of plate appearances. However, Ramirez is a 5.1 WAR player while Braun is at 4.0 WAR (according to Fangraphs). The main difference here is that Ramirez plays a premium position where offense is scarce, while Braun plays an easy position where offense-heavy players are the norm. Also, Ramirez is more capable defensively than Braun. Despite nearly identical offensive production to Braun, Ramirez is a full win more valuable because of the position he plays. Accounting for position and defense is the main reason why players like Joe Mauer, Ramirez, and Chase Utley are underrated, while guys like Jason Bay, Carlos Lee, and Justin Morneau are overrated. Pitcher WAR is obviously calculated differently, and is done using FIP or tRA while accounting for innings pitched.
One will often see WAR abbreviated to "wins", as in "David Wright is a 7 win player." A 2 win position player is about average; 4 wins and we're looking at an All-Star; 6 wins will have a player in contention for MVP and Cy Young Awards; 7-8 wins is a Hall of Fame season; any more than that is an all-time great season (we have seen Albert Pujols, Barry Bonds and Alex Rodriguez have such elite seasons recently). The same scale generally applies for starting pitchers, but is much different for relief pitchers. In his outstanding 2005 season, Mariano Rivera was a ~3.2 win pitcher. Frankie Rodriguez broke the single season saves record in 2008 but was just a ~1.8 win pitcher. Relievers do not provide nearly as much value as position players or starting pitchers, no matter how dominant.
Refer to the Fangraphs Glossary for a better WAR tutorial by Dave Cameron (scroll down the page to see the links). The concepts may not be easy to grasp at first, and my explanations above are a bit clunky, but when understood they will enhance one's knowledge of the game. Just ask Nate Silver.
In February, I used Sky Kalkman's WAR spreadsheet to "post-dict" the 2008 Mets season. I input the components of WAR to determine how many wins the Mets would expect to have based on the metric. Initially, I came up with 87 wins, but I have since updated the spreadsheet to fully account for baserunning (linked here) and the number is bumped up to 88 wins. How many wins did the 2008 Mets actually have? 89. How many were predicted based on Pythagorean Record? 89. The "WAR method" came within 1 win of these marks, which I think is pretty impressive and a rudimentary way of displaying the value of WAR as a metric. Additionally, a quick look at Sean Smith's all-time WAR leaderboard at Baseball Projection shows us that the best players in baseball history, per general consensus, are also the best per WAR.
No stat is infallible and there are plenty of legitimate criticisms of WAR (defensive metrics aren't perfect; simply stating a win value does not tell us a player's strengths and weaknesses). Regardless, the goal is to sift through the noise and determine which stats are the best for player evaluation. WAR is certainly one example. This is not some fringe science, practiced only by those residing in their mothers' basements. It's going on in front offices around baseball, be it Tampa Bay, Boston, Oakland, Seattle, Texas, etc. I respect that people choose to enjoy the game however they like. Sabermetrics is not for everyone, I understand that. However, to "trash" a stat without understanding what it's about, especially when the stat is in use in Major League Baseball everyday, is flat out ignorant. Let's move away from the idiotic WFAN model of baseball discussion. Let's try to improve the discourse, not bring it down.