rWAR: What Is It Good For?

[Bob Breum]

I use OOTP’s WAR to evaluate my team’s performance. I have written several team news articles demonstrating how I use it to plan my offseason moves. As I am the GM in charge of position players, I rarely need to give any thought to the pitching side of the house, other than to marvel at how Erik manages to construct a competitive pitching staff out of duct tape, bailing wire, chewing gum, and the loose change that we find between the cushions of the office Lazy Boys. All of his pitchers earning above the league minimum combined are paid less than either of my top two salaried position players.

As you may know, there are several competing versions of WAR out there. The Sabermetric crowd uses fWAR to refer to the FanGraphs version, and rWAR for the Baseball Reference version. Within OOTP, WAR uses the FanGraphs methodology, i.e., fWAR. If you want rWAR, it is also available within OOTP, you just have to look a little harder to find it. It’s listed for individual pitchers under Expanded Pitching Stats 1. For league leaders, you’ll find it under BBA -> Statistics -> Pitching Leaders. For teams, look under BBA -> Statistics -> Team Statistics -> Sortable Stats -> Pitching Stats, then use Customize to add rWAR to Pitching Stats.

To paraphrase the Dos Equis guy, I don’t always give much thought to pitchers’ performance, but when I do I prefer rWAR. fWAR for pitchers uses FIP (fielding independent pitching), which is calculated based solely on strikeouts, walks, home runs, and HBP, as it is designed to be independent of team defense. Its weakness is that it does not take into account the nature of contact for balls put in play, i.e, groundball percentage, hard hit percentage, and other factors that affect BABIP (batting average on balls in play). rWAR, on the other hand, is derived from runners allowed per nine. OOTP’s glossary says that rWAR affords “a better measure of a pitcher's skill at limiting balls in play and stranding runners, which do not factor into the … calculation [of fWAR].” Both fWAR and rWAR are counting stats, i.e., they are proportional to playing time. Twice as much playing time roughly translates into twice as much fWAR/rWAR, whether it be positive or negative.

We do not know exactly how faithfully that OOTP has implemented these two versions of WAR. Markus is on the record as stating that WAR in OOTP is based on his version of FanGraphs fWAR, but implies that it is not an exact replica. Likewise, we can assume that OOTP’s rWAR is close but not identical to the actual Baseball Reference model. I think we can stipulate that they are representative of the two for our purposes.

Baseball Reference publishes a comparison chart. Here are some advantages to their rWAR calculations for pitchers vis-à-vis fWAR:

• rWAR adjusts for team defense; fWAR does not.

• rWAR adjusts for park factors for parks that the pitcher actually pitches in; fWAR does not.

• rWAR adjusts for team offense of teams pitched against; fWAR does not.

• rWAR adjusts replacement level according to the quality of the league; fWAR does not.

These differences are valuable in many ways.

• rWAR gives you a realistic metric to assess the performance of your BBA pitching staff relative to the league at large, removing variables such as your home park, your team defense, and the presence of particularly strong or weak offensive teams in your division.

• rWAR gives better insight into your minor league pitching, as minor league parks are frequently not the same as your BBA home field, and the quality of the hitting is obviously different.

• rWAR allows you to evaluate free agent pitchers and/or potential trade candidates independent of their particular park/team defense/league circumstances.

It is important to keep in mind that WAR/rWAR is a counting stat. When comparing starting pitchers, you must consider how many games that they have started, i.e., derive rWAR per start for a more direct comparison. This rewards starters for going deeper into games. For relief pitchers, you’ll want to use rWAR per inning pitched to normalize the data.

Here are the top twelve starting pitchers ranked by rWAR/start (minimum 10 GS)

Name	GS	ERA	FIP	fWAR	rWAR	rWAR/GS
Phil Cole	19	2.40	3.89	2.7	4.7	0.25
Hector Silva	23	2.80	3.05	4.8	5.6	0.24
Robert Hanson	22	2.50	3.04	3.7	5.0	0.23
Barney Lindsay	22	2.84	3.45	3.4	4.8	0.22
Greg Shaw	23	3.01	3.90	3.0	5.0	0.22
Juan Marroquín	21	3.22	3.81	3.0	4.3	0.20
Tommy Akers	15	3.03	3.73	2.0	2.8	0.19
Vassilis Zorba	22	3.16	3.20	3.9	4.1	0.19
Andrés Gonzáles	16	3.70	3.58	2.7	2.8	0.18
Arnold Cantrell	23	3.22	3.95	3.0	4.0	0.17
Yoshii Takuda	23	3.22	3.52	3.3	4.0	0.17
Kinzo Iwamoto	23	2.96	4.05	2.2	3.9	0.17

Here are the top twelve relief pitchers ranked by rWAR/IP (minimum 40 innings)

Name	IP	ERA	FIP	fWAR	rWAR	rWAR/IP
Alejandro Chávez	53.2	2.18	3.42	1.3	2.7	0.051
Max Birstall	45.2	2.17	3.17	1.2	2.2	0.049
Xuan Ngo	50	2.16	3.38	1.1	2.4	0.048
Tom Miller	57.2	1.25	3.13	1.1	2.6	0.045
Lindsay Barber	55.1	2.28	2.96	1.9	2.5	0.045
Francisco Salgado	52.1	2.75	3.62	1.2	2.3	0.044
Todd Stone	56.1	2.40	2.67	2.2	2.3	0.041
John Dawson	54.2	2.30	3.20	1.2	2.0	0.037
Cristo Muñóz	60.2	1.48	2.90	1.3	2.2	0.037
Troy Reynolds	60.2	2.23	3.45	1.1	2.1	0.035
Sergio Herrera	66.2	2.02	3.44	1.0	2.3	0.035
Scott Everard	84	2.14	2.97	1.9	2.9	0.035

[Jwalk100]

I use rWAR too. This is confirmation for me, thanks!

[RonCo]

There's a Phil Cole sighting.

[Lane]

Nice, thank you

[Knucklehead254]

There's a Phil Cole sighting.

Why you gotta do this to me

[pandan]

Good post. I think it's good to use both.

I generally look at fWAR for how good the pitcher should've been, and rWAR for how good their season was in actuality. Anecdotally, I tend to observe greater fluctuations in the latter season-by-season.

[cheekimonk]

Why calculate rWAR/G for starter and rWAR/IP for RP? Seems that "per IP" would be a direct comparison.

[Bob Breum]

Why calculate rWAR/G for starter and rWAR/IP for RP? Seems that "per IP" would be a direct comparison.

My thought here is that it rewards starting pitchers who pitch deeper into games. rWAR is a counting stat, so two otherwise equivalent pitchers who differ in stamina will also differ in rWAR/game. Their rWAR/inning, however, will be identical.

[cheekimonk]

Why calculate rWAR/G for starter and rWAR/IP for RP? Seems that "per IP" would be a direct comparison.

My thought here is that it rewards starting pitchers who pitch deeper into games. rWAR is a counting stat, so two otherwise equivalent pitchers who differ in stamina will also differ in rWAR/game. Their rWAR/inning, however, will be identical.

That makes sense except for 2-way players who split rotation to & relief

[DanGladdenFC]

Very useful information, thank you. It's good for absolutely something, I guess?