Tuesday, December 14, 2010

Contrast & Compare

After tackling the great WAR divide for positional players, it is only logical we examine the way value is determined for pitchers. Not nearly as uniform or tidy as batter Wins, as we will soon see.

The biggest difference between Fangraphs WAR and Rally WAR is FIP. Fangraphs attempts to isolate the factors controlled by the pitcher1 (strikeouts, walks, and home runs) and work from there. They don't make "corrections" for defense because the figures they incorporate come pre-stripped of defensive meddling.

Rally WAR (pioneered by Sean Smith, a very smart guy who now works for an unnamed big league team) starts with runs allowed and works backwards. rWAR makes adjustment to the replacement level runs based on the quality of defense (Total Zone derived) and ballpark (derived using magic and answering the bridge troll's riddles three.)

While FIP and its brethren are good and runs against seems uncouth and cruel, it turns out run allowed per 9 innings pitched has the greatest correlation for year-to-year performance. So I can see value in each of these methods.

This schism within the nerd herd threatens to split the fraternal brotherhood like nothing since Princess Lea in the Gold Bikini versus the Canadian broad on BSG. The great love of over-complication called out by the simplistic. WHO WILL TRIUMPH???

As it turns out, us. The functionally nerdy benefit from the spastic yelps of socially-bereft savants lurking in dark corners of the computer supply store. Let's see how it shakes down. First is Baseball Reference's version with Fangraphs below. The rWAR sits on both for comparison purposes.

Brandon Morrow146.126769201161.6
Brett Cecil172.2288710900.9222.2
Brian Tallet77.15604500.8-15-1.4
Casey Janssen68.20293800.490.7
Dana Eveland44.29352701-8-0.8
David Purcey340161900.430.2
Jason Frasor63.20303601.160.6
Jesse Carlson13.207800.810.1
Kevin Gregg590243202.181.3
Kyle Drabek1739100110.1
Marc Rzepczynski 63.21237390120.2
Ricky Romero210329813001.1323.4
Scott Downs61.10193301.5141.8
Shaun Marcum195.1318412101373.8
Shawn Camp72.10264001.2141.5
Shawn Hill20.2481201.140.4

Brandon Morrow146.
Brett Cecil172.
Brian Tallet27.3-13.250-1.5-1.4
Casey Janssen0.03.468.70.40.7
Dana Eveland44.
David Purcey0.02.4340.20.2
Jason Frasor0.08.663.70.90.6
Jesse Carlson0.0-2.613.7-0.30.1
Kevin Gregg0.07.8590.81.3
Kyle Drabek170.
Marc Rzepczynski62.70.710.50.2
Ricky Romero2100.00.043.4
Scott Downs0.011.861.31.21.8
Shaun Marcum195.
Shawn Camp0.02.672.30.31.5
Shawn Hill20.

Weep for Brandon Morrow. Deprived of his gaudy components, Morrow is left to rot as a below-average starter. You know that isn't true, and I know that isn't true, but runs allowed shows no mercy.

Notice the lack of defensive adjustment afforded any Jays pitchers. Total Zone ranks the Jays as pretty much neutral, which is hard to argue unless you try. I can't imagine what sort of ballpark factor is included but if it doesn't cut the Jays plethora of lefties a break then consider me highly dubious.

Relievers who pitch loads high-leverage innings benefit greatly under the Rally system. The adjustment for leverage index/saving grace of runs allowed pumps up the value of both Shawn Camp and Kevin Gregg. I prefer to think of relievers as interchangeable, with the replacement bar set very high as other guys on the same staff can often step into bigger or higher-leveraged roles without issue.

One thing both systems — and all of humanity, I trust — can agree on: Brian Tallet is terrible. That Baseball Reference and Fangraphs can join in chorus proclaiming Brian Tallet is fundamentally awful gives me hope through this dark and gloomy holiday season. Hallelujah!

1 - This is contentious assumption number 1 of many. But we soldier on.

Thanks again to Fangraphs and Baseball Reference for the data. And Modern Life is War for actually writing a song with a decent payoff.


  1. I've never heard of this "Sean Smith" you speak of, and quite frankly, he sounds made up.

  2. The Canadian broad from BSG made a few all-star games in her time but Princess Leia in the gold bikini is a first ballot Hall of Famer.

  3. When it comes to position players, I like Baseball Reference's version of WAR more. IMO, it is calculated from a more reliable defensive stat, total zone, than UZR. The difference in offensive stats that play a role in their calculation of WAR is pretty insignificant.

    When it comes to pitchers, it's more interchangeable. I really like FIP and how that affects the pitcher's WAR, though BR's WAR is also pretty good if you want to consider the quality of defense and the ballpark the pitcher played in. However, if I were looking at how well a pitcher played regardless of the quality of defense, I'd more be in favor of Fangraph's WAR.

    They are both great sources of information regarding baseball statistics.


Send forth the witticisms from on high