THE BOOK--Playing The Percentages In Baseball

For sake of completeness, there is a pretty big difference between the two versions of WAR on the pitching side, but that’s not relevant to measuring Hamilton’s value.  fWAR is based on xFIP, if I recall correctly, while rWAR or b-rWAR is based on ERA.

Michael: I refer you to the link I put in the main thread.

@Det Mike ...

I keep brining that issue up at FG, using Adam Wainwright as the classic example.

His ERA is consistently 0.5+ runs lower than his FIP. Quite simply, he’ll give up more walks, and less hits. Like this year, something like 17 more walks than Halladay, 50 fewer hits.

Anyway, the difference in approach “costs” AW50 1-2 WAR a year at FG.

At FG, this same issue caused DA to average the various defensive metric systems with weights, and come up with an “average rating”. Carl Vrawford had a big difference among the systems as well.

I like the average approach. Rather than just picking the one we like the best, or the one that rates our favorite player the highest (like me and BR’s WAR ... favoring AW50), averaging them together would seemingly be the best approach, specifically because no single metric gives the complete pitcher. Averaging, seemingly, minimizes the cons of each systems, while retaining its strengths (i.e., a highly rated defender among most systems, is going to have a high average ... rather than just basing it off of a single system, which might be the high/low extreme).

If we’re, as stat community, are going to put so much value in WAR ... using it to select our MVP, the best/worst players, trade evaluations, etc ... then there needs to be more done to minimize the dependence on any single metric or any one system. That seems to be uneccesarily self-limiting.

fWAR is based on xFIP, if I recall correctly, while rWAR or b-rWAR is based on ERA.

FanGraphs WAR is based on FIP, not xFIP. 

There is no “right way” to do pitcher WAR.  Sean’s way has problems.  Our way has problems.  A combination of the two would have problems.  Extracting pitcher performance from defensive performance is just really hard. 

Neither is right or wrong.  They’re just different.

Adam Wainwright’s career rWAR is 18.1.  His career fWAR is 17.4.

#5

Yes, but that is with the higher replacement production given to Wainwright in FG.  Take away the extra replacement level increase due the the lower baseline and adjust down for Waino pitching in the NL compared to the AL (which FG does not do) and the gap is much larger.  Presumably that gap is due to FG not considering any BABIP skill that exists (agreed with Cameron that it is virtually impossible to locate the skill with that few BFs), situational/timing issues, controlling the running game, DP’s, XBH prevention, and all of the other secondary run prevention skills.  That is why I always say that if I want to see a pitcher’s WAR in a season or 3, I pick FGs before the pitcher has enough BFs for all of the pitching/defense codependent aspects of run prevention to be identified as skill.  If I want to look at guys with long careers, I look at rWAR since the co-dependent run prevention aspects have had enough BF’s to reach levels I am comfortable with.  Way too many guys have too much influence over situational pitching/timing, DP’s, controlling the running game, BABIP, XBhit prevention, wild pitches, etc. to ignore them at a long-career level.

I don’t think the differences in replacement level/league adjustments are enough be able to say the difference in methodology is really 1-2 wins per year.  For the it to be that high, the difference between fWAR and rWAR would have to be higher than 1-2 runs per year.  Otherwise, you would have replacement level pitchers in one being close to average in the other (as a general rule, not as extreme cases).

"Adam Wainwright’s career rWAR is 18.1.  His career fWAR is 17.4.”

Oh sure, bring facts into the discussion. *grin*

While I do not know of all the aspects involved in the different WARs, nor how small variations, affect the overall WAR.

I basically looked at this year, and the difference in WAR was about 0.8 (now it’s 0.6), and just jogged it through my mind, and figured it would be around 1.0 difference by the end of the year. Then I noticed that his ERA is often .5+ less than FIP, and threw out a number.

To be accurate, 0.5 to ~1 would have been a much better estimate than 1-2 *but still wrong, see below).

Also, the difference in ERA and FIP (or E-F) is really only significant since 2008, so the effect I am talking about really only covers 3 years. He’s been an SP for 4. 2007 his ERA and FIP were only different by 0.2

When you look at the actual numbers, it turns out I’m “full of crap”, and unfortunately, it’s not the first time.

Year --- E-F --- fWAR --- rRAR
2008 --- 0.58 --- 2.20 --- 2.20
2009 --- 0.48 --- 5.70 --- 6.00
2010 --- 0.79 --- 5.60 --- 5.60

Turns out it doesn’t make a difference. Thanks. I’ll not make the same arguement again. *grin*

err, the difference currently is 0.0, not 0.6. Bad me.

More mistakes. I’m really just polluting the site at this point.

AW50’s fWAR is 5.0, rRAR is 5.6

So, this year there is a ~significant~ difference, nothing like Hudson’s, Lee’s, or Liriano’s though.

I’m out. Heh Heh.

#7

I think (estimation here) it is about .4 runs per year for a NL pitcher.  So about 22 fgRep. to 20 rRep. for a current AL pitcher and maybe 24 fgRep. to 20 rRep. for a current NL pitcher.  Certainly enough to add a WAR or two to Wainwright’s totals.  So that would bring the WAR gap to around 2, or near 20 runs prevented that Rally picks up on that Fangraphs does not due to the factors I laid out earlier.  Now how much of that is Wainwright vs. luck?  Obviously, we can’t tell yet - nor can we tell at this point which WAR is a better indicator of Wainwrights skill, But the replacement level is making a little difference.  Nobody suggested 1-2 wins per year.

I was responding specifically to this line from post #3:

“Anyway, the difference in approach “costs” AW50 1-2 WAR a year at FG.”

Which it seems was just a hasty estimation that CircleChange has changed his mind on after looking at more of the data.  I’m fine with saying it is 2 wins over his career.

The problem with looking at the difference between ERA and a component run estimators scaled to ERA is unearned runs. The league average UER/R for pitchers during Wainwright’s career (weighted by his IP in each season) is 0.075. But Wainwright has had a 0.085. So he gives up a greater percentage of his total runs as unearned runs.

Here is the problem with judging a player solely on metrics. Anybody who has ever seen Hamilton play defense knows he is an above average to possibly very good defender. He has 60 speed, an 80 arm, and 55-60 range. I appreciate the saber side of baseball, but judging a player’s worth takes more than numbers. I’ve seen more guys held at third or scared to take second on a double with Josh in the outfield than any other player because of his arm. I don’t see those intangibles anywhere in the formulas.

Josh Hamilton, 2010 UZR: +5.8.  And guess what? It’s all in the ARM portion of the rating.  He’s +6.0 runs in preventing runners from advancing or throwing them out, and -0.2 runs in getting to balls with his range. 

This is a very weird complaint, given that the metric is agreeing with everything you say.

I’ve seen more guys held at third or scared to take second on a double with Josh in the outfield than any other player because of his arm. I don’t see those intangibles anywhere in the formulas.

Those aren’t “intangibles”.  Those look pretty tangible to me.  If you are suggesting that these things should be counted, I agree.  Counting something is tangible.

Something is not “intangible” because it’s not being counted.

David, what he is doing is typical of people who have summary opinions with no evidence: they see a topic, and use that as a launching pad to say whatever they want to say based on whatever flimsy beliefs they have.

In this case, Josh Hamilton is +6 in his career in the ARM rating of UZR, which is pretty good.  The leaders over the last 3 seasons according to UZR are: Hunter Pence, Jeff Francoeur, Nick Markakis, Adam Jones, Jayson Werth, BJ Upton.  At the bottom of the pile, Brian Giles, Brad Hawpe, JD Drew, Grady Sizemore, Adam Dunn.