SB Nation User Blog: ZeppelinDZ

2009 Royals BABIP for batters

ZeppelinDZ — Fri, 11 Sep 2009 21:09:46 -0000

I really didn't want to wait til the end of the year to skew the data with September call-ups so I'll post this now for your consumption.

For those who don't know, a simple google search on the subject will familiarize you with the concepts here as much work has been done on the subject in just the last year. Several fairly in depth models have been developed but I choose just to keep things simple with a aLD+bGB+cFB model. It isn't as good as the others, but it will still be pretty close and a lot easier to compute. If someone wants to run one of the better models for comparison, I would highly encourage it.

Lastly, this isn't a list of who is good/bad on the team, its an adjustment. Billy Butler is and will be a great hitter, even if he was lucky this year. The list below the jump.

50 PA in 2009 min.

Name          LD%      BABIP   xBABIP    Diff
Billy Butler      18.90%    0.341    0.297   0.044    Aviles was ~.065 last year for comparison
Mark Teahen    20.10%    0.341    0.304    0.037
Miguel Olivo       15.70%    0.311    0.277 0.034    Profiles as a higher BABIP guy with higher K and HR rates
A. Callaspo      16.00%    0.305    0.279    0.026
Mitch Maier    17.60%    0.311    0.289   0.022
Alex Gordon     8.50% 0.246    0.233     0.013    The 8.5 LD% is the worry here, just hasn't driven the ball at all
David DeJesus    19.50%    0.308    0.301     0.007
Jose Guillen    14.20%    0.266    0.268    -0.002
W. Bloomquist   19.80%    0.298    0.302    -0.004
Mike Jacobs    18.40%    0.289    0.294    -0.005
John Buck           16.30%    0.258    0.281    -0.023
Y. Betancourt      17.30%    0.255    0.287    -0.032
Brayan Pena 21.80%    0.280 0.315    -0.035    Encouraging that his BA has room to rise further
Coco Crisp       17.60%    0.247    0.289    -0.042    Another reason to try to bring him back next year
Tony Pena   2.80% 0.132    0.196    -0.064    2.8% Oh My Gawd, I had no idea
Mike Aviles        18.90%    0.226    0.297    -0.071    Can this guy just be normal for a year?
Luis Hernandez 20.50%    0.227    0.306    -0.079

Clarifying BABIP

ZeppelinDZ — Thu, 12 Feb 2009 19:37:55 -0000

Just the luck, a major area of baseball statistical interest for me gets a major post by Mellinger and I’m out of town with the Fanpost already falling down the main page. Well, I want to respond to a couple misconceptions about BABIP and I felt the comments section of our community response wouldn’t do it justice. This is partly a summation of other’s ideas, partly my take on it, so feel free to correct me if you believe I have misstated some part.

First and foremost, BABIP of batters IS a skill. No credible researcher that has studied it will argue against that. Mellinger’s take that good hitters have high BABIP’s is absolutely correct… it is also irrelevant to how hitter’s luck is measured with BABIP. Now, there is still ongoing research to understand it better, but generally most people will agree that BABIP is mainly a measure of a batter’s ability to hit the ball hard interacting with the opposition’s defense. (A quick note, the determination of this is backwards, I have started with a conclusion here then move to evidence, but past research done correctly has answered this for us and simplifies this explanation)

When we talk about pitchers, the fact that they face, over the course of a season, all different types of batters with differing abilities to hit the ball hard tends to average their BABIP out to a mean value, based on their own park and defense. That’s why people tend to focus on the over/under .300 BABIP as a measure of luck. There are always outliers that have to be considered, but this is still a key assumption of DIPS theory. I won’t get into pitchers anymore because it has been written about much more extensively elsewhere.

But back to hitters. We can further break down the results of batted balls from hitting the ball hard to making solid contact and power of the swing and again, interacting with defense (+ other aspects + ‘error’). However, over the course of a season, players tend to play against the "average" defense, so it cancels itself out when comparing player to player. So if we want to understand what happens to a batted ball, we need to focus on finding a statistical proxy to "solid contact" and "bat power" and whatever "other effects." Most baseball fans will instantly be able to answer part of this, a line drive seems to fit part of that perfectly. Well, the statistics agree, line drives are highly correlated to BABIP. Now work done by myself, Studman of THT, Bendix and Dutton of BAT, and MattS of The Good Phight blog have all added other potential instruments to use in determining what causes batted balls to result in hits for those other effects. These variables can then be regressed to fit a mathematical model to explain what causes BABIP.

So where does luck come in? Well, regressions rarely yield 1 +1 = 2 type answers. They inherently have error cause by the unknown, i.e. 1 +1 + E = 3 (known + unknown = BABIP). Mathematically, E is very easy to quantify, but intuitively, it is harder to explain what it means. This E is what many of us consider luck. It’s some unexplained force that is causing batted balls to behave outside our understanding. Now, without getting into some heavy duty philosophy on fate, choice, and control, people will take that and define it as they will. But in the end, the fact is batters have not been shown to be able to control that error. Like I said though, this is far from the final word, more research is required.

The case against Mike Aviles is the E was enormous in 2008. Judging by the normal proxies, he wasn’t really doing anything amazing with respect to "solid contact" and "bat power." Above average? Yes. League leader? Not even close. Now, as a Royals fan, I hope my statistical methods are completely wrong and Aviles is league MVP next year, but it requires a lot of people to be very wrong about what causes hits to be hits.

Sorry if this is long, but I felt Mellinger's post was more due to mis-understanding than anything and this isn't a simple issue.

Freakonomics Blog on Sports Economy

ZeppelinDZ — Fri, 09 Jan 2009 20:39:35 -0000

Freakonomics Blog on Sports Economy

Doesn't answer too many questions, but relevant to things none the less.

Aviles and PrOPS

ZeppelinDZ — Mon, 06 Oct 2008 14:25:53 -0000

I thought about just adding a comment to the Aviles/Sickels thread, but as I think Mike Aviles will be the single most talked about person this off-season for RR, I wanted this to be a new post.

So, a lot of you read my look at xBABIP and the royals, but a central theme about some comments was what it really means. When comparing actual versus expected values, how does that translate to more traditional stats which actually tell you about on-field performance. i.e. what does a BABIP-xBABIP of .60 mean?

Enter PrOPS (I'm gonna call it props to save wear and tear on my shift key). Props is short for predictive OPS. It was developed by economist JC Bradbury a couple years ago. Basically, it takes a bunch of peripheral rate stats and turns them into a luck neutral, semi-park neutral OPS number. here is the article introducing it.

It seems the exact formula is proprietary as far as I can tell, I have tried to reverse engineer it and have a working version w/o park factors, but since THT
reports the stat, its easy to look up. Now, I really like the concept of this. Its not perfect, but as a normalizing agent, it is my favorite stat I have found to see true performance.

Props isn't necessarily a predictive stat. However, if you can forecast the components that make up the calculation, it can be used in a way. I might do a little of that over the off-season.

Now, I said this post was about Mike Aviles, but I'm gonna be really quick on his analysis.

2008 props: .737

OPS - PrOPS = .096

that .096 would be the largest difference of ANY qualified batter this year.

Is a 740 OPS bad for a middle infielder? no. Can he improve? sure, but here are the fundamental components of OPS, which ones do you think he gets better at for his age 28 2009 season? What about a year of bad luck or a sophomore slump?

Line drives per batted ball
Groundball-to-flyball ratio
Walk rate
Hit-by-pitch rate
Strikeout rate
Home run rate
Home park of the player

I have said this about a dozen times now, I hope Aviles proves me wrong, but the 2009 Mark Quinn award seems well within reach.

P.S. for good news. props like gordon, dejesus, and shealy

Sexson waived...again

ZeppelinDZ — Fri, 15 Aug 2008 19:27:42 -0000

Sexson waived...again

"But the dramatic holes in the first baseman's swing made him too easy an out."

2 teams think he's crap now...is there anybody who still thinks we should sign him?

more BABIP thoughts

ZeppelinDZ — Fri, 08 Aug 2008 16:11:27 -0000

kcdc1 recent post on the inherit problems with applying xBABIP to hitters lead me to do some work on if certain hitter types produce different results from batted balls. Obviously as technology improves, individual analysis will be possible, but for now, I still try to group players in well defined profiles. The data I’m using is from 2007 and includes only qualified batters, so this is an analysis of MLB regulars only. Also, home runs and infield flies are properly accounted for this time.

I preformed a number of regressions, but I’ll spare most of the details as too technical for most, but if you are interested in them and/or the data email zeppelindz@gmail.com and I’ll help you out.

Power hitters:
I sorted players by both HR/AB and HR/FB and derived a new xBABIP equation for varying degrees of power and the results were unexpectedly conclusive.

Power hitters generate significantly more hits from ground balls and significantly less from fly balls. There was a jump in line drives as well, but it wasn’t as conclusive.

K Rates:
Sorted by K rate and had pretty much the same result as power hitters.

The conclusion on this, I believe, is fairly straight forward. These guys hit balls hard, causing more grounders to get through for hits (as postulated by kcdc1). However, hard hit fly balls result in fewer hits. This is likely caused by expectations. Opposing defenses know power hitters and play them deep frequently which takes these deep hits away at high rates. Also, just postulating here, power hitters likely hit fewer bloop singles, which classify as fly balls.

Speed:
I created a speed statistic using stolen bases, triple to double ratio, and infield hits to sort players. It’s somewhat arbitrary, but looking at the players used, I think it got a list of speed players for a valid result.

Surprisingly, speed seemed to have little effect. The rate of hits from ground balls actually went down very slightly (probably statistical noise).

Clutch:
I couldn’t resist, Fangraphs had a clutch stat which I sort players on to see its effect.
Well, no changes.

Clutch is independent of hit type it seems (remember, HRs are factored out which might change things).

As for the royals, this changes very little from my first post. The royals are great at having average players. Guillen barely fits in as a power hitter compared to the guys I analysis. Gathright, although very fast, doesn’t use his speed effectively to really be considered in the same category as some of my speedster sample. Bottom line, no result substantially change from, but there would be a few players who cross a threshold from say, slightly lucky to neutral or something like that. After the season is over, I’ll go back through with some of these findings and get a better idea of our roster and luck.

Questions, comments, or any additional player types you want tested?

KC Royals and BABIP

ZeppelinDZ — Mon, 04 Aug 2008 19:07:12 -0000

Let’s take a look at some Royals hitters and how their luck and other factors have fared over the last few seasons.

I’m an economist that does forecasting for a living, so all these stats are a labor of love. Plus, I have become completely sold on this 3 true outcomes approach to baseball. Basically, I’m looking at the difference between BABIP and expected BABIP (an avg of a couple different methods to calculate it) to look for luck, which are the unlabled stats given. Also, I’m looking at line drive rate variance to see if expected BABIP is a reasonable estimation. I also like walk/k rates and isolated power (if all this sounds a lot like PECOTA minus its comparables, it should).

Quick note: of course young guys can improve, but this is a historical stat take on things to see how reliable those stats are for us to judge our guys on.

Mark Teahen:
06: 0.037    very lucky    .290/.357/.517
07: 0.041    very lucky    .285/.353/.410
08: -0.015    unlucky       .243/.312/.381

Trends: solid walk rate, high K rate, unpredictable power, below ML avg LD%
Bottom Line: current numbers more likely to be the "real" Teahen, I'm really down on him after looking at those past BABIPs

John Buck:
05: -0.009    neutral       .242/.287/.389
06: -0.019    unlucky          .245/.306/.396
07: -0.030    very unlucky    .222/.308/.429
08: 0.008    neutral       .247/.323/.407

Trends: consistent production ind. of luck, good power, OK walk rate, high k rate, low LD%
Bottom Line: League average catcher offensively and very consistently such, Royals can win with that.

Alex Gordon:
07: 0.006    neutral        .247/.314/.411
08: 0.007    neutral        .252/.343/.406

Trends: increasing walk rate, developing power, high k rate, avg LD%, huge L/R splits
Bottom Line: Gordon's future lies on developing his incredible tools, learning to lay of the slider away from leftties

Joey Gathright:
05: 0.027    lucky       .276/.316/.340
06: -0.024    unlucky        .238/.321/.292
07: 0.020    lucky       .307/.371/.342
08: -0.007    neutral       .251/.302/.267

Trends: declining walk rate, very low and improving K rate, no power with some speed based doubles, variable but likely very low LD% with extreme GB tendancies
Bottom Line: 07 looks fluky with both luck and unusually high LD%, if that was his breakout year, he has regressed badly with career lows in several key areas.

Jose Guillen:
05: -0.012    neutral             .283/.338/.479
06: -0.047    very unlucky    .216/.276/.398
07: 0.034    lucky       .290/.353/.460
08: -0.014    neutral       .261/.291/.456

Trends: has forgotten how to take a walk as a Royal but was ok before, average K rate, great power, below avg LD%
Bottom Line: If he remembers how to walk at his career rate, legit batter, 91 OPS+ OF/DH until then.

Mike Aviles:
08: 0.067    wow lucky    .340/.365/.550

Trends: low walk rate, low K rate, legit power, avg LD%
Botom Line: flashing legit offensive tools, but contining the .340 average is a luck based pipedream. Age 27 rookie season is worrisome on development curve.

Tony Pena Jr.:
06: -0.047    Very unlucky    .227/.261/.341
07: -0.006    neutral                   .267/.284/.356
08: -0.117    God hates TPJ .148/.169/.196

Trends: very low walk rate, high K rate, low and dropping iso power, declining LD%
Bottom Line: I think 07 Pena is closer to the real offensive performance to expect, but does his defense offset what Aviles brings? RR.com has spoken

David DeJesus:
05: 0.010    neutral        .293/.359/.445
06: 0.005    neutral        .295/.364/.446
07: -0.014    neutral        .260/.351/.372
08: -0.013    neutral        .301/.361/.457

Trends: Good walk rate, very low and declining K rate, some power, great LD%
Bottom Line: The real deal! 07 was caused by low LD% unlike rest of career. Very excited about him.

Billy Butler (BAM BAM):
07: 0.014    neutral        .292/.347/.447
08: -0.001    neutral        .271/.331/.394

Trends: good walk rate, low K rate, developing power, low LD%
Bottom Line: Once he learns to drive the ball like he did in the minors, he becomes an elite hitter.

Ross Gload:
06: 0.024    lucky       .327/.354/.462
07: -0.007    neutral       .288/.318/.441
08: -0.027    unlucky        .275/.321/.346

Trends: low walk rate, very low K rate, lower than normal iso power - avg other years - doubles based, good LD%
Bottom Line: alot better than some people make him out to be, but below avg power, especially HRs, for a 1B limits his value, future is a ML bench

Mitch Maier:
sample size

Bottom Line: needs ABs to see who he is

Estaban German:
06: 0.073    extremely lucky    .326/.422/.459
07: -0.008    neutral    .264/.351/.376
08: -0.024    unlucky       .244/.300/.336

Trends: down year in general, historically, good walk rate, lowish K rate, light doubles power, avg LD%
Bottom Line: tough 08, but 07 numbers look repeatable. 350 OBP with low K rate can play, but still looks like super sub at best.

Alberto Callaspo
07: -0.077    very unlucky    .215/.265/.271
08: -0.019    unlucky    .290/.349/.330
small samples

Trends: avg walk rate, very low K rate, no power, good LD%
Bottom Line: I expected more power from him, but right now, a German clone in value. Future depends on what position Aviles ends up at.

Miguel Olivo:
05: -0.028    unlucky       .217/.246/.367
06: 0.007    neutral        .263/.287/.440
07: -0.002    neutral        .237/.262/.405
08: 0.024    lucky          .261/.293/.477

Trends: low walk rate, high K rate, good power, low LD%
Bottom Line: all his value stems from his ability to hit the long ball, which plays OK for a catcher. Good back-up, marginal starter

Mark Grudzielanek
05: 0.000    neutral        .294/.334/.407
06: -0.005    neutral        .297/.331/.409
07: 0.027    lucky    .302/.346/.426
08: 0.002    neutral        .299/.345/.399

Trends: Greek God of consistency, slightly below avg walk rate, low K rate, light power, exceptional LD%
Bottom Line: he has to breakdown at some point right?

need stat help (eBABIP)

ZeppelinDZ — Tue, 29 Jul 2008 21:07:26 -0000

quick and dirty question...
does anyone know (or know source) of the eBABIP equation?

I have a link to a guy who quotes .763LD% + .265GB% + .131FB% as the 2006 version.

I ran a regression on 2004-2008 batted balls and got .463LD% + .242GB% + .261FB%
but i know of some errors on that, but I didnt think they were that big of ones to get such a different result.

I have been running some stats on our boys looking for signs of hope, and I am trying to get a better idea of how lucky a team we are.

early results... TPJ very unlucky, but still bad

Kyle Davies

ZeppelinDZ — Wed, 18 Jun 2008 14:07:33 -0000

Kyle Davies

with all this mike aviles love going around, I think another player is getting overlooked for his hot streak. in 4 starts this year, kyle davies has allowed 1, 1, 1, and 1 run.

his walk rate is still high, and SO's are down a tad, but his hit rate is excellent and his overall strike% isn't terrible. still needs plenty more starts to really judge him, but he may be the hottest player on the team right now.

take that crappy 27 year old ss's with lifetime ops+ of 174!!

berroa and salary

ZeppelinDZ — Mon, 09 Jun 2008 16:27:45 -0000

berroa and salary

there was some debate over who paid the salary. Basically, everybody (ap, kc star, la times) all said royals picked up the salary, but the dodgers team site said differently. well, they amended it.

"The Dodgers will be responsible for only the remaining pro-rated portion of the Major League minimum salary for Berroa, not the $3.5 million remaining from his multi-year contract, as earlier reported."

royals got about .75 mil (buyout and pro rated min salary) and a crappy prosect for berroa