Tangotiger Blog

Saturday, January 17, 2015

How do I respond to this guy?

?I received an email, which I will post in its entirety (including the lack of introductions, etc):

Every single weight you have seems arbitrary to me.  Where is the proof?

(0.72xNIBB + 0.75xHBP + 0.90x1B + 0.92xRBOE + 1.24x2B + 1.56x3B + 1.95xHR) / PA

Do you have some regression test of all the data in baseball that revealed these weights to a 95% statistical significance?  What is the dependent variable?  Runs Scored?  After every year, would these weights not change after running the test?  

Running a regression test isn't complicated.  It's first year undergrad engineering stuff.  Why people accept your weights as doctrine with zero statistical evidence is astonishing.

Soooooo.....  where do I begin?  He obviously did not read The Book, where I go into detail as to how this was developed.  And he obviously did not look at the excerpt on The Book site.  He did not google wOBA, as the first link brings you to this Fangraphs page which lists the wOBA weights year by year going back to 1871.

Instead.... he made up non-facts and treated them as facts (aka strawman).  Then used those non-facts as arguments to form his opinion.  In the end, we have a summary opinion with no evidence, which is the definition of bullsh!t.  And there's no reason to argue with bullsh!t, which is why I instead took advantage of his bullsh!t email to at least show a few links about wOBA to those who may not be aware.

Monday, January 05, 2015

Runs Assisted

?Lee continues tracking this for us, which I love.  It really fills the gap between R and RBI.

Lee: can you post the R/event, RBI/event, and RAS/event, where event is each of 1B, 2B, 3B, HR, BB, HB, RBOE, others.

Wednesday, November 12, 2014

How to Create Your Personal WAR

I posted this on Bill James' forum, so I'll just repost it here:.

Note that there is nothing new here at all.  It's the same thing I've been saying when I first starting championing WAR.  But perhaps saying it this way adds more clarity to the process.

Read More

Tuesday, October 28, 2014

How impressive is WAR?

?Poz has a long piece on Bill James.  He quotes Bill on his view of WAR:

Well, my math skills are limited and my data-processing skills are essentially nonexistent. The younger guys are way, way beyond me in those areas. I’m fine with that, and I don’t struggle against it, and I hope that I don’t deny them credit for what they can do that I can’t.

But because that is true, I ASSUMED that these were complex, nuanced, sophisticated systems. I never really looked; I just assumed that the details were out of my depth. But sometime in the last year I was doing some research that relied on these WAR systems, so I took a look at them, and … they’re not very impressive. They’re not well thought through; they haven’t made a convincing effort to address many of the inherent difficulties that the undertaking presents. They tend to get so far into the data, throw up their arms and make a wild guess. I don’t know if I’m going to get the time to do better of it, or if it will be left to others, but … we’re not at anything like an end point here. I assumed that these systems were a lot better than they actually are.

There's things I agree with and things I do not. 

1. I do agree that WAR is not impressive, or at least not impressive looking.  That's the beauty of its design.  For example, look at what WAR is for pitchers at its core:

IP/9 x (lgERA + 1 - ERA) / 10

If your pitcher has a 3.00 ERA in a league of 4.00, and he has 225 IP, you get this:

225/9 x (4 + 1 - 3) / 10 = 5 wins

(That divide by 10 is simply the runs to win converter.)

And here's a little secret: this was invented by... Bill James!  In his classic(*) article on the MVP race with Clemens and Mattingly, he goes through the machinations, including doing that "+1" bit, which is actually the most important part of this equation.  Without the "+1" part, it becomes Wins Above Average, which is how Pete Palmer presented it in The Hidden Game.  The +1 part turns it into Wins Above Replacement.

(*) Most of his articles are classic, so, I'm not really narrowing down the list.

2. As for the not thought through, I do not agree with Bill at all.  They are actually incredibly thought through.  Again, just as an example, the distinction between Starting Pitchers and Relief Pitchers is huge.  This is something that baseball people inherently understand, but that those of us studying the data kind of dismissed or ignored for the longest time. 

We just couldn't explain that a 3.50 ERA by a starting pitcher was far better than a 3.50 ERA by a relief pitcher, and it goes beyond just volume of innings.  Keith Woolner was one of the first to bring this up over a decade ago, and others followed suit, me included, notably in The Book.  This is research that evolved over time to the point where I gave it a rule, the Rule of 17, which basically says that a relief pitcher gets 17% more K, allows 17% fewer HR, allows 17 points fewer in BABIP, and 17% fewer runs (walks are flat).

There's the standard thing we do with park effects, as well as the difference in AL/NL talent ,so that "lgERA" is really adjusted for all that.  Some even go so far as to look at the actual opponents and their fielders to further adjust that lgERA.  (Note, when I say ERA I really mean RA/9, but ERA is so ubiquitous a term.  Which is also another advance, that we focus on runs allowed, not the made-up earned runs.)

3. The wild guess could be something that's true, but I wouldn't say it's a wild guess so much as it's a necessary guess, an educated guess, a guess to move the discussion forward.  Some examples are BABIP, which we really don't know how to split up very well, or at least, in a way that we can explain it well enough.  If I say to regress Kershaw's 2014 BABIP will be based on his BABIP in 2013 and 2012 and 2011, that looks really confusing.  Even if I try to tell you that simply to understand his 2014 performance on its own.  It's really really hard.  So, I just say: split the difference and assume his responsibility of the BABIP is halfway between his observed performance and league average.

Another one we have to handle is relief pitchers and leverage.  Again, to move the discussion forward, we credit the reliever not with the Leverage Index that he actually faced, but rather halfway between that and the (by definition) league average of 1.  It's part of a concept called chaining, that if that reliever wasn't there, some other reliever would have taken his place.  But much like Ozzie Smith's fielding is leveraged at SS (he's involved in more plays than in LF) or Rickey Henderson's hitting is leveraged at leadoff (and so he gets FAR more PA than the average hitter), we can't completely discount the talent associated with the leveragable opportunities.

***

So let me just say that for purposes of making sure the metric is not a black box, to make sure the metric is accessible, to make sure that anyone could calculate their own version of WAR, the framework is flexible enough to allow that to happen.

We do not want it complex, or (too) nuanced, or too sophisticated.  We want it so that anyone can build a house, and WAR gives you that blueprint.  The potential saberist out there is now empowered and is given a path to build an even better house.  The foundation is there.

We can say that a house is not impressive, or we can say that a house is incredibly impressive.  Either way, WAR has been able to cut through the idea that we need something complex to be able to explain something as complex as baseball.  

Tuesday, September 23, 2014

Why do we evaluate pitchers primarily by runs they allow, but not batters by runs they score?

I sent this to Bill James in response to a discussion he's having on his blog.

With regards to a "run" credited to a pitcher: it's actually shared with his fielders, but just that the pitcher is the primary owner.  This would be similar to a "run" credited to a batter, when it's shared with the guys who follow him.  But we evaluate batters, for the most part, on their components.  In essence, we're not giving the batter full credit for the runs they actually scored, when evaluating their overall performance.  To that end, if two pitchers had the same RA/9, but one did it with a .240 BABIP and the other did it with .340 BABIP, we might be more inclined to think that the guy with the .240 BABIP had more help from his fielders.?

Sunday, September 21, 2014

Negative Runs Created

?There's a thread here asking about negative wRC+ and wRC. 

The point that is missing is the context of runs. You can have 9 guys that are all league average, and this team will score 4.5 runs, which is 0.5 runs created per player. You can replace one guy with a pitcher-as-batter, and keep the other 8 guys the same. This team will score 4.0 runs. Since the 8 guys were each worth 0.5 runs created when the 9th player was someone like them, we reason that they are STILL worth 0.5 runs when a pitcher-as-batter is in their midst.

And if the 8 guys are worth 0.5 runs created, that makes them worth 4 runs. Since the team of them plus the pitcher-as-batter scored 4 runs, we deduce the pitcher-as-batter created 0 runs... even though he did actually get on base, and score runs, and drove in runners.

And if you had Ben Sheets batting, this team would score 3.9 or 3.8 runs, and so, he'd deliver "negative" runs.

***

And this is the hard part: if you had 9 Ben Sheets batting, they would certainly score more than 0 runs, and certainly not negative runs.

So, you really have to establish the ASSUMPTIONS for the stat. One set of assumptions leads to Sheets being negative runs. Another leaders to Sheets being positive.

And one set of assumptions will lower the RC of the OTHER batters depending on who is their teammate, while the other assumption tries to maintain independence.

Choose your poison, but don't tell the other person that he's wrong. You aren't arguing results, but rather reasonableness of assumption

Friday, September 12, 2014

BsR and BsR

?I didn't realize the Fangraphs has chosen to use BsR to represent baserunning, meaning runs from non-batter events (like SB, CS) which they use wSB as notation, and runs from batter events (like taking extra base), which they use UBR as notation.

The wSB notation is in reference to the "w" in the "weighted" metrics inspired by wOBA.  The "U" in UBR is inspired by MGL's UZR, who provides, I think, the UBR numbers.  It's a nice notation for the w and U metrics.

But then, they combine the two into BsR.  But longtime saberists know that BsR is the notation that Dave uses for his BaseRuns (as described by Patriot).

I don't know how Dave feels about it, but it gives me a chance to bring up BaseRuns, easily one of the least appreciated and most important metrics around.  If giving it a new notation, or finding some different notation for Fangraphs' version of BaseRunning is the result, all the better.

You guys are creative.  What have you got for me?

Monday, September 08, 2014

Improving WAR by admitting things we don’t know into its own bucket

Dave has a good article here that talks about Passan's article.  The first couple of readers have interesting points, so, I'd like to go off on a tangent to those.

We all know that offense is made up of getting on base and moving runners over and not making an out.  But, what if the ONLY thing we had was OBP and runs scored?  If a team was say +200 in runs scored relative to league average, we would NOT be able to get that as an estimte using only OBP, even with random variation.  That's because if you are +200 in runs scored as a team, you probably had tons of HR too.

So, team runs scored might have one SD = 50 runs, but using only OBP, I might get one SD = 35 runs.  How do I make up the shortfall?  Well, I can create a multiplier and simply give out 1.4 runs for every 1 run estimate I get from OBP.  That forces things to add up.  Or, I can simply say "I dunno", and for every shortfall, I simply put it in some team bucket, where I need to allocate those runs ONLY WHEN I CAN ESTIMATE IT.

I might end up with an Andre Dawson being a BELOW AVERAGE hitter (because I'm not aware of his HR), but I'd also have a bucket for the Expos and Cubs that shows I'm short a few runs.  But I just don't know if Dawson gets those or some other hitter.  Because as I said, in this case, I don't know how many HR he hit.  I might have an IDEA in watching him hit, but I can't tell.

And the same applies for fielding.  Maybe I can say that Andrelton Simmons saved 10 runs as a "fer sure" compared to an average SS.  And maybe the Braves have another 30 runs to allocate.  I just don't know who to give it to.  That's one way to handle this.

But are people happy to do this?  They AREN'T happy with FIP, because this is EXACTLY what FIP does in WAR at Fangraphs.  Indeed, they'd rather allocate those "I dunno" runs to the pitcher in some way.  Maybe partway, via BABIP, but not all the way.

So, that might be a solution, to give out some runs, but not all runs.

You make me laugh!  But, be funnier still!

About 15 years ago, Howard Stern was reading a report from some paid expert consultants of his competitors, who were trying to figure how to beat Howard Stern in the ratings.  The basic conclusion was: be funnier.  Baba Booey likened it to baseball, making the analogy that to be a better hitter than Mark McGwire (his example), you just need to hit more HR.

And so we come to this overall good article by Jeff Passan(*), but who in making his point went to the typical absurd point: saberists need to do better than the current versions of WAR.  It's the same thing as complaining that a comedian doesn't make you laugh ALL the time, or that Peyton Manning throws 30% incompletions, that you need to checkin two hours early for that transatlantic flight that flies at under the speed of sound, or that sometimes, your spouse really does have a headache.  What are your options?  

(*) In the comments later today, I will correct the misinformation in his article.

The current versions of WAR, at Fangraphs and Baseball Reference ARE the best the public has.  If you think they are not good enough, fine.  But, what are your options?  To not use them at all?  To rely on... what?  Yapfests?  You have a legion of people who have dedicated their free time to develop the best metrics we can.  That has to be good enough.  That they can be improved doesn't mean you get to discard them. 

Can they get better?  Yeah, sure.  No one is saying otherwise.  But, are you going to wait until they are better? Because in the end, you have to use SOMETHING, RIGHT NOW.  And whatever that something is (other than taking a blend of the two), it'll be worse than what Fangraphs and Baseball Reference gives you.

You aren't funnier than Louie CK, though someone eventually will be.  You can't throw better than Clayton Kershaw, though eventually someone will.  And the transporter hasn't been invented. Yet.  Until then, fWAR and rWAR is the best we've got.  Using anything else is a step back.

Wednesday, August 13, 2014

wOBAcon

?I use this stat, or some form of it, all time time. SF is the obvious missing element, and ROE (reached on error) as well.

Friday, July 18, 2014

wOBA Intro

?Good job by Neil in describing wOBA.  A few of my responses:

IBB issue:

Suppose you calculate wOBA as has been described, and your player ends up with a wOBA of .320.

Now, suppose that you insist on including IBB. We agree it shouldn’t have the full weight of a walk (.70), but we also agree it’s better than an out (.00). Assume that the weight for the IBB is .32.

What happens to this player’s new wOBA if we include IBB in both the numerator (with the .32 weight) and the denominator (as a full PA)? Nothing! His new wOBA stays exactly at .320.

And if you think about it, what wOBA implicitly does with the IBB is to give it a DYNAMIC weight that is exactly equal to the hitter’s wOBA. So, Bonds’ IBB is worth .50 while the #8 hitter’s IBB is worth .25.

And it makes perfect sense if the IBB is win-neutral with respect to the hitter at the plate.

***

HBP v UBB:

As has been noted, HBP occurs more often when it can do more damage than the UBB, namely with a runner on 1B. UBB are issued disproportionately with 1B open.

***

ROE issue:

When the defense gets an error, this is a BAD thing for the defense. You can argue it’s far more bad for the fielder than the pitcher, but nonetheless, taking the defense together as a whole, it’s bad for the defense to allow a runner to reach base on error.

And for the defense, whether they allow a runner to reach base on error or by a single, it’s a similar kind of damage.

Defense is pitching+fielding.

Since offense is the exact mirror of defense, whatever is bad for the defense is good for the offense. Regardless of the amount of talent it takes for the batter to reach base via error or via hit batter, the fact is the offense BENEFITS.

***

"Making an out doesn’t change his wOBA, except for the slight increase in PA."

EXCEPT?  That's a huge exception!  Take for example a hitter with a .320 wOBA on 600 PA.  That would mean he has "192" in the numerator and 600 in the denominator.

Now, give this player 20 walks.  That adds 14 to the numerator (now at 206) and 20 to the denominator (now at 620).  His wOBA is now .332.

Now, give him 24 outs.  Numerator stays at 206 and denominator at 644.  wOBA is .320.

So, adding 20 walks has been cancelled out by adding 24 outs.

And that's perfectly fine.

Monday, June 16, 2014

wOBA and BaseRuns

This article does a good job in showing that linear methods (like wOBA and its inspiration of Linear Weights) can't be used for Teams.  wOBA-squared does a much better job.

In any case, the answer is BaseRuns.  For teams (and pitchers) use BaseRuns.  For hitters, use wOBA.  For hitters in a particular environment, use the "Theoretical Team" construction and a "with or without you" process (google Patriot and that term, and you'll get what you need).?

Monday, May 26, 2014

Explaining OPS

You'll laugh, you'll cry.  From Patriot.

Friday, May 16, 2014

WAR as a local road, until we get to the highway

There are many legitimate questions on WAR.  Questions.  Not conclusions.  In this thread, the original thread starter simply comes to conclusions.  In there though are some legitimate questions, but there are also some untruths.  Normally, I'd go through the 22-page thread, and pick out the legitimate questions or correct the falsehoods.  I'd do that NOT for the benefit of the original commenter.  After all, he has shown himself to prefer summary opinions without evidence.  Nothing you can do with that person.  I would do it for the benefit of everyone else, so we can have a dialogue and exchange ideas.  And move forward.

Anyway, I will leave it to the Straight Arrow readers to cull through that long thread, and pick out whatever is said in there that interests them, and I'll comment here.

As I said on Twitter: WAR makes reasonable assumptions and constructs a methodology based on the data available.  WAR has its limitations, but only because of what it (and all of us) don't know.

As a result, the WAR framework (as opposed to the various implementations) is the best thing we've got.  Just because you want to be able to drive at 100kph doesn't make the local roads somehow an inferior surface.  That's all you've got in order to get to that highway.?

Friday, April 25, 2014

WAR, using only R+RBI

The following is based on data as found on Fangraphs, 2010-2013, minimum 1200 PA.  There were 248 players that qualified, enough for 30 teams to field a starting lineup, plus another 8 left over for DHing.  I only used R plus RBI, per 700 PA.  The leader was Miguel Cabrera with 245 R+RBI per 700 PA.  The average was 164 R+RBI.  The "replacement level" was 124.6 R+RBI.  That's the minimum we'd expect of an MLB hitter.  Twelve players peformed below this level, but most of them provided positive fielding value.  As the leader of this group, we have Brendan Ryan with 117 R+RBI per 700 PA, or negative 7 relative to the replacement level R+RBI.  But he also had +61 runs of fielding value above average.  So, we can't be beholden to R+RBI without considering a player's fielding value. 

But, to the extent that that's what people do do, simply look at R+RBI, and pay lip service to everything else, we're going to move forward on this idea.

To put this on the WAR scale, we'll simply multiply this R+RBI above replacement by .071. Here are the top 10 in R+RBI per 700 PA, placed on the WAR scale:

8.6           Miguel Cabrera

 7.7          Carlos Gonzalez

 7.1          Jose Bautista

 7.0          Ryan Braun

 6.9          Josh Hamilton

 6.9          Troy Tulowitzki

 6.6          David Ortiz

 6.5          Allen Craig

 6.3          Mike Trout

 6.3          Curtis Granderson

So, this is what the mainstream fan wants to see from actual WAR.  They want to see numbers that look like that for these players.  For many, they sort of match.  Here for example is where they don't match at all.  These players all average under 1 WAR per 700 PA, and yet have more than 3 "WAR" per 700 PA (as calculated using only R+RBI):

5.9           Ryan Howard

 4.2          Mark Reynolds

 4.1          Vladimir Guerrero

 4.0          Michael Morse

 3.9          Jason Kubel

 3.7          Raul Ibanez

 3.5          Adam Dunn

 3.2          Garrett Jones

 3.2          Delmon Young

 3.1          J.P. Arencibia

On the flip side, these are guys that WAR loves (min 4.5 WAR per 700 PA), but don't have the R+RBI to back it up (max 3.5 "WAR" based on R+RBI).?

2.3           Chase Headley

 2.4          Carlos Ruiz

 2.5          Brett Gardner

 2.9          Yadier Molina

 3.0          Shane Victorino

 3.0          Brian McCann

 3.2          Ben Zobrist

 3.2          Jason Heyward

 3.3          Joe Mauer

 3.4          Carlos Gomez

Thursday, April 24, 2014

Negative Losses

When I was still in Montreal, I created a won-loss system for hockey players.  I was kind of proud of it.  The way it works was this: you have 20 players, the average? team wins 40 games, and so, the average player would be worth 2 wins (and 2 losses).  That would mean the average player would have 4 "games".  So, the perfect player would be 4-0, as a team of such players would win 80 and lose 0.  The problem I quickly came across was when I tried to do it with Wayne Gretzky.  I'd have Gretzky as worth some 5 to 10 wins above average.  But my system wouldn't allow any player to exist beyond being 2 wins above average.

Then I figured that since Gretzky played twice as often as the average forward, his "space" is not 4 games, but 8 games.  That still though left me to cap him off at 4 wins above average (8-0 record).  So, I abandoned the idea of a cap, and simply allowed him to be worth 12 W and -4 L. 

I then thought maybe I could add 4W and 4L on his record, to make him 16-0.  That took care of the negative losses, but now I made his "space" 16 games rather than 8.  And I'd have to remove 4W and 4L from the rest of his team.  I wasn't necessarily against the idea, but the whole thing started to lose its lustre for me.

Anyway, I sent someone a link to the Willie Mays page, so you can see the Gretzky issue in a baseball context and for a whole career.  He noted that it's hard to wrap your head around negative losses, and I agree.  I sell the idea on the basis that Mays and Gretzky were so good that they negated some of the losses of their teammates. 

So, according to The Indis, Willie Mays has 210 W and -27 L (that's negative 27).  In Bill James notation, that's a 210+27 record (rather than 210 - -27). But, if I add an equal number of W and L in those negative seasons, his career record is 244 W and 7 L (a 244-7 record).

Presentation-wise, understanding-wise, what makes more sense to you?

Wednesday, April 23, 2014

Stop treating errors as if the batter made an out

I'm all for counting the different ways for a batter to reach base, and keeping those totals separate.  So, hit batter, unintentional walks, singles, doubles, triples, HR... and reaching on error (ROE).  But when it comes time to aggregating things, and you have hits and walks on one side, and outs on the other side, and  you have to decide where to aggregate the ROE?  Well, there's only one place for them, right in the bucket where the batter did NOT make an out.

As hard as it is to believe, a batter that reaches base (ROE) LOWERS his on base percentage. 

And don't talk about the "mistake" of the fielder? as if the fielder is not part of the opposition.  We count the mistakes of the pitcher don't we?  The hitter played a role.  We're going to not count something in hockey, football, basketball, or soccer because of the opponent's position on the field who made the mistake?

Monday, April 21, 2014

Are catchers poorly represented by WAR?

The WAR framework has always had an issue with catchers.  It's hard to figure out exactly how to rank them.  In a nutshell, the total number of wins given out to all MLB catchers should be roughly equivalent to 1/8.5 of all wins given out to all other non-pitchers (the extra .5 due to DH and PH and PR).  The idea is that catchers is as important a position as any other position.  This idea is further supported that catcher salaries is in line with this fraction.  (Or at least, it was, the last time I checked several years back when I was developing the WAR model.)

This is an issue I first reconsidered when a BBWAA voter(*) told me that he voted straight along WAR lines for the Hall of Fame.  Which made me happy, since it gave the stat a "hidden" legitimacy.  And as RallyMonkey (who created the WAR implementation at Baseball Reference that the voter used) said, it's like he had a HOF vote.  Even better is that he'd get all the credit, without incurring any of the blame.

(*) Ken Davidoff, who has agreed to be named.

Anyway, since Piazza was 11th on the WAR list at the time, the voter was being assailed for leaving him off his ballot.  Catchers, more than any other position, I am pretty flexible in going with the idea that we should adjust the positional adjustment an extra 0.25 or 0.50 wins per season.  If that means an extra 4 to 8 wins for star catchers over their career, maybe that makes more sense.

One way to check is to look at salaries paid.  After all, salaries represent the perceived value, and WAR represents the perceived production.  Ideally, these two should correspond strongly.  There are obvious reasons why salary is not a great metric.  First, we have that the pre-free agent class of players are severely underpaid compared to the free agent class of players, given the same level of talent.  We're talking a multiple of about 5x.  Then, it's possible there is a bias for or against catchers as well.

In any case, I'm not here to do anything comprehensive.  I'm getting the issue started, and as Michael Lewis said of Bill James: he prefers the honest mess than the tidy lie.  And I'm going to leave you with a mess.  I'll give you the cleaning supplies though, so feel free to ask questions below.

So, this is what I did.  I took the two best catchers of this past generation, Mike Piazza and IRod.  (To tidy this up, you should expand it to include more catchers, handling each catcher, one at a time.) 

Piazza had 59 rWAR (WAR at Baseball Reference) and IRod had 68.  IRod started his career in 1991 and ended in 2011 (Piazza was 1992-2007).  So, I looked for all players who had no more than 71 WAR and no fewer than 56 WAR (i.e., +/- 3 wins of our catchers).  I limited it to those whose first season was no earlier than 1989 and no later than 1994 (i.e., +/- 2 years of our catchers), and whose last season was no earlier than 2005 and no later than 2013.  That left me with 5 players: Manny, Lofton, Edmonds, Sosa, Olerud.  Those 5 averaged 63 WAR, compared to the 64 WAR of our two catchers.

WAR is saying that our two superstar catchers are equivalent to these 5 star players.  I agree, it smells wrong, but just because it stinks doesn't mean it's not right.

The lifetime earnings of IRod was 123MM$ and 120MM$ for Piazza.  If these 5 guys averaged around 115-130MM$, then it's a validation for WAR (or at least, one more notch).  If they averaged more than 130MM$, then maybe WAR overvalues catchers, since it suggests they have the same value but are being paid too little.  (Again, the idea is that you are paid for performance, but we have the noted mess up there to clean up.)  If they averaged under 115MM$, then maybe WAR undervalues catchers.

And how much did our five guys earn?  It goes from a high of 207MM$ for Manny to a low of 61MM$ for Lofton.  The average of our five noncatchers is 109MM$.

This suggests that perhaps WAR for superstar catchers needs to be bumped up by about 10%, or some 6-7 wins.

The next step is to repeat this process, but add 0.50 wins / 700PA for our catchers, and then see what noncatchers are in their new comp groups, and repeat the above.

Again, I can't stress enough how messy this is.  We need more aspiring saberists to clean this up.

Thursday, April 17, 2014

wOBA Primer

Terrific job by studes to getting inside wOBA, and explaining it for mass consumption.

***

I'll add one thing about the presentation of wOBA compared to the presentation of Linear Weights (LWTS): I've never had to explain negative values.  Even though wOBA is completely dependent on the LWTS model, the user is never left scratching his head with what to do with one guy who is -10 runs in 600 PA and another guy who is -7 runs in 300 PA. 

And similarly, you could compare wOBA to various forms of RC (including wRC), and you don't have to be confused with one guy having 82 RC in 600 PA and another having 46 RC in 300 PA.

Why do we have this issue?  It's what I call the single-dimensional presentation problem.  What we REALLY want is to present the data in TWO dimensions.  For example, IP and ERA (or RA9).  Those are perfectly presented, one as a "rate" stat, and the other as a quantity stat.  Do we REALLY want to throw those two pieces of information away, and instead present "runs relative to average" or "runs relative to replacement"?  We have one guy with a 3.00 ERA and 108 IP and another with a 3.75 ERA and 216 IP, in a league of 4.00 ERA.  In terms of "runs relative to average", the first guy will come in at +12 and the second guy at +6.  In terms of "runs relative to replacement", the first guy is +24 and the second guy is +30.

So, which is the better presentation, relative to average or relative to replacement?  Beats me.  Everyone has their own question, and by collapsing the two dimensions (quality and quantity) into one dimension, you lose half of your audience.

BUT, what if instead I present TWO dimension, and I say that the pitcher's "support neutral" W-L record is 7-4 for one guy, and 12-11 for the other guy.  This now opens up possibilities for both groups to have a discussion.  One guy can suggest, "hey, the first guy won 3 more than he lost", while the other guy might say "ok, but the second guy got 5 more wins and 7 more losses, so the first guy will need a backup who has to be at least as good as that".  And if this backup would be 4 and 8, then we can see that the 7-4 record, while impressive, doesn't have enough quality to make up for the hole that the lack of quantity opened up.

So, doubling back to wOBA.  By putting it on a rate scale, we are now forced to ALSO consider PA.  Linear Weights and Runs Created, by presenting the number as some sort of combination of quality and quantity leave us asking if that combination is sufficient to answer our questions.  And half of the time, the answer is no.  Hence, the best thing to do is leave things in two-dimensions as much as you can.

Friday, April 11, 2014

Simple team-level +/- metrics

Before he was snatched up by the Redsox, Tom Tippett would provide a simple team-level assessment on his Diamond-Mind site.  He would count all the bases that the batters on a team got (TB+BB+HB) and subtracted what the defense gave up.  A plus/minus of sorts.

Tony Blengino offered something along the same lines, but focused on K and BB only.  As you guys know, the K-BB differential is one of the easiest and most illuminating metrics we have of pitchers.  It's so simple and powerful that it is as good, or better, as a predictor of future pitcher performance than any other metric that you will find.  Tony therefore simply expanded this idea to look at it on an offense minus defense perspective.  A plus/minus of sorts, but limited to just this particular subset of play.

I'm a fan of the approaches that Tippett and Blengino espouse, both for the simplicity of its presentation, but power of its message.

It's why I also am a fan of Bill James' presentation of K and BB numbers on a scale of W and L.?