WAR Basics

[1905 Giants]

So can someone give me a quick rundown of how to best utilize WAR?

So far what I've heard is 40+ WAR careers and 5+ for a season, but other than that how do I best use it when rating players?

[willshad]

Don't use it at all for 1) catchers 2) good fielding third basemen, shortstops or second basemen 3) guys who had huge home/road splits.

[JDanger]

Well, firstly I wouldn't listen to anyone who says anything critical about anything without asking for evidence.

Secondly, B-Ref has this little reminder above each Player Value table:

8+ MVP,
5+ All-Star,
2+ Starter (average),
0-2 Sub,
< 0 Repl

[Bothrops Atrox]

Well, firstly I wouldn't listen to anyone who says anything critical about anything without asking for evidence.

Secondly, B-Ref has this little reminder above each Player Value table:

8+ MVP,
5+ All-Star,
2+ Starter (average),
0-2 Sub,
< 0 Repl

Not sure if I would use this scale for modern pitchers. The 10th best NL and AL pitching WAR each year is around 4, not 5. So if you think the 10th best pitcher in each league is an All-Star...

[Ben Grimm]

This is what I don't get...

Why isn't WAR simply the oWAR+dWAR? Why is it close but not quite for in almost every incident? Using Brandon Phillips (just a random name) as an example over his past 5 years...

       OFF  DEF  WAR  OFF  DEF  O+D  DIF
                       %    %         %
2008   1.5  1.6  2.8  53%  57%  3.1  10%
2009   2.3  0.6  2.5  92%  24%  2.9  14%
2010   2.7  1.4  3.7  73%  38%  4.1  10%
2011   3.9  1.0  4.7  83%  21%  4.9   4%
2012   2.2  1.6  3.4  65%  47%  3.8  11%

So, okay. I guess WAR is somehow obtained in some combination of both offense and defense with offense having a bit more value. But it's not the same value throughout as seen in the '09 & '11 seasons. Combining '09s 2.3 oWAR with 0.6 dWAR gives a result of 2.9 WAR. But somehow it's only 2.5 (a 14% difference). Drop down two seasons and it's 3.9/1.0 which would total 4.9 (which is close) but the "true" WAR is 4.7 - only a 4% difference.

I can understand if somebody was to say that oWAR counts as 65% of the total and dWAR makes up the other 35%, but that's clearly not the case. I've been searching thru the endless threads and I can't seem to find anything that addresses this. Any info would be extremely helpful.

[Jackaroo Dave]

This is what I don't get...

Why isn't WAR simply the oWAR+dWAR? Why is it close but not quite for in almost every incident?
. . . .
. . . .
. . . .
I've been searching thru the endless threads and I can't seem to find anything that addresses this. Any infoi would be extremely helpful.

oWAR is all WAR minus fielding. dWAR is fielding WAR and position adjustment. So if you add dWAR and oWAR, you double count the position adjustment. I can't say precisely why the position adjustment appears in both o- and dWAR, but it is useful for some comparisons. For example, if you wanted to compare defense by itself for defenders at different positions, you'd want to take into account their positions. Similarly, if you wanted to compare offense alone, you'd want to keep track of the difference in offensive replacement value for the two different positions.

[Ben Grimm]

oWAR is all WAR minus fielding. dWAR is fielding WAR and position adjustment. So if you add dWAR and oWAR, you double count the position adjustment. I can't say precisely why the position adjustment appears in both o- and dWAR, but it is useful for some comparisons. For example, if you wanted to compare defense by itself for defenders at different positions, you'd want to take into account their positions. Similarly, if you wanted to compare offense alone, you'd want to keep track of the difference in offensive replacement value for the two different positions.

Okay, I'll take a look into that and see if it clears things up for me. To be honest, I know very little how WAR works other than from reading posts and some blogs and such. But it seems to be recognized more each year and I don't want to be ignorant of it or how it's created. Thanks for the reply

[Jackaroo Dave]

Apropos of this, especially your frustration in looking for a really basic answer--which I have shared many times with Fangraphs and BBREF--I was wondering if people posting to the stats forum would be interested in a team project putting together a primer for contemporary statistical analysis: something that would explain, say, how linear weights are derived, then how they are used to develop oWAR, then branching out to the other aspects of WAR, etc.

Not limited to WAR, but I thought that might be a good place to start, because it has so many moving parts yet is in common use.

Another important part would be a compact reference guide, so that answers to questions like Ben's would be easier for the readers to find themselves.

I know a lot of us feel that we have a slippery grip on some of the concepts, and with a little research--maybe with some guidance--we could get a much firmer grasp. This would also be an opportunity to reach out to those who just get fed up or use the difficulty of contemporary stats as a way to escape dealing with their insights.

You wouldn't have to know a whole lot, just a little more than the person you'd be writing for. We could set our own pace and be as ambitious or as laid back as we individually chose.

I think it would be fun and could be productive, and a lot easier than trying to do it on one's own. I know several people here who have the interest and the ability, and there are others who might not want to be involved but could vet the project and offer critiques and suggestions.

Is anybody interested?

[Ben Grimm]

I got my first primer on WAR here: Understanding Ryan Howard's WAR. But it was in '09 I believe and things have changed a bit.

BTW Jackaroo, I love the idea of a primer that can be stickied here built by you guys who actually know how WAR is developed. I can't offer much more to it at this time other than maybe asking for a "dumbed-down" idea or example for those of us who'd be in the remedial class.

Like I mentioned, I normally wouldn't care, but it seems like this is a direction we're heading when analyzing players and I'm sure many would like to have the ability to get out of the dark.

[Jackaroo Dave]

I got my first primer on WAR here: Understanding Ryan Howard's WAR. But it was in '09 I believe and things have changed a bit.

BTW Jackaroo, I love the idea of a primer that can be stickied here built by you guys who actually know how WAR is developed. I can't offer much more to it at this time other than maybe asking for a "dumbed-down" idea or example for those of us who'd be in the remedial class.

Like I mentioned, I normally wouldn't care, but it seems like this is a direction we're heading when analyzing players and I'm sure many would like to have the ability to get out of the dark.

Or make an informed choice to stay where they are. Thanks for your encouragement. If this gets off the ground, there would be no more valuable contribution than pertinent questions and requests for clarity.

[Ben Grimm]

Or make an informed choice to stay where they are. Thanks for your encouragement. If this gets off the ground, there would be no more valuable contribution than pertinent questions and requests for clarity.

I really love the idea you're heading. If I can be of any help in any avenue, just let me know. While I may not be able to contribute to the pure mechanics of WAR, wOBA, RC, and such, I can help in researching things to help you guys put it together if needed.

[Sultan_1895-1948]

Found this graph online from SABR.

Anyone know why the WAR numbers don't match up to baseball-reference? For example, Ruth in 1927 shows 12.5 on bbref but here (year 14 on the graph), it's like 8.3

WARsabrGRAPH.JPG

[Jackaroo Dave]

Found this graph online from SABR.

Anyone know why the WAR numbers don't match up to baseball-reference? For example, Ruth in 1927 shows 12.5 on bbref but here (year 14 on the graph), it's like 8.3

The years on the graph appear to be ordered from high to low. So Ruth would have 8.3 WAR in his 14th best year, not his 14th year.

However, bbref has no 8.3 WAR listed. They list 1933 as Babe's 14th best, with war of 6.2. In 1931, a couple of steps up, he has 8.1.

This disparity could be because the calculations for bbref war were changed within the last couple of years, or because the graph is using a different flavor of WAR, perhaps from Fangraphs or from Seamheads. The article really should reference the source of the stats.

[Sultan_1895-1948]

The years on the graph appear to be ordered from high to low. So Ruth would have 8.3 WAR in his 14th best year, not his 14th year.

Oh ok thank you. I assumed it was progressive seasons because of the lower label, which is obviously stupid. Brain fart moment

[Sultan_1895-1948]

Nothing to do with WAR but this graph is pretty cool

OPSgraph.jpg