Offensive Ability Rating Rankings for 2009
Some may remember last year when I posted a few articles here about a stat I cooked up over the last two hot stove seasons, called OAR. I brainstormed an idea while reading this encyclopedia of baseball history (which was interesting enough, but for some reason all I could think of was stats and sabermetrics, and how it works, which I'm still not entirely certain, but I have made a lot of progress I think). I got it to the point where it was more accurate, but still not quite what it could be.
Anyway, I realized a few things since I wrote that and started working on the stat again, this time correcting a few things that I think distorted the rankings. For those who didn't read last year, Offensive Ability Rating is an idea to incorporate as many offensive stats as possible into one rating, roughly similar to a quarterback rating. OAR is based mainly on what is very close to Power Factor, an old stat that basically shows how many bases to expect if a person hits the ball, or how efficient they are with their hits. I call this component of the rating BEF, or Base Efficiency Factor. What makes it different from power factor is that I also include stolen bases (but not quite as much as a regular hit)... but they are also penalized by caught stealing. So if someone steals a lot of bases, they can be near the same level as a power hitter in this area of the stat. I also think this accounts for speed to some extent, a valuable asset in a ballplayer (as well as baserunning abilities).
The parallel stat to BEF is called BEA, which is simply multiplying that factor by batting average. While batting average isn't the most important stat, I still think many people pay attention to it and it does have some value in judging a player's offensive performance. I think perhaps that this might be an area that needs revision in the ratings, but I'm not entirely certain, it just seems like a really simple way of seeing how often someone gets and hit and how much power (and to a lesser extent, speed) they have as a hitter.
So that's basically half the stat, the other half is more convoluted.... Eye/Discipline Factor, or EDF measures many things, mainly how often a hitter gets on base when he basically has the bat taken out of his hands. Walks, hit by pitch, as well as when he is told to hit a sacrifice fly, if he can get the run in and an out of course. Also, I wanted to penalize for striking out, since you are basically not putting the ball into play at all, which is not good imo. When you strike out, you don't have the chance of causing an error, etc. Granted you also might not hit into a double play, but I also include a penalty for grounding into double plays in the stat.
Here's where it gets a little weirder, I really wanted to incorporate BABIP into the stat also, so ELF is the next component of OAR, aka Eye Luck Factor. ELF is basically EDF and BABIP combined, or a really weird type of on base stat.
Besides those major components of the stat, I also included pitches per plate appearance, which is sort of a tie breaker if two players are really closely rated. Before I go further, I just want to state that I don't think this is the most scientific or advanced stat, it's more like just for fun and to start conversation about different aspects of baseball. Hopefully some good philosophical or critical discussion of stats in general will come of this. But without further adieu, here's the OAR rankings for 2009:
1. Albert Pujols: 158.27
2. Chase Utley: 144.93
3. Kevin Youkilis: 144.31
4. Jason Bay: 144
5. Adam Dunn: 143.5
6. Adrian Gonzalez: 143.37
7. Prince Fielder: 142.93
8. Ben Zobrist: 142.74
9. Carlos Pena: 142.14
10. Alex Rodriguez: 141.03
11. Joe Mauer: 140.12
12. Joey Votto: 139.27
13. Lance Berkman: 138.71
14. Shin-Soo Choo: 137.63
15. Jason Werth: 137.57
16. Nick Swisher: 137.33
17. Derrek Lee: 137.12
18. JD Drew: 136.51
19. Mark Texeira: 136.28
20. Ryan Howard: 135.36
21. Russell Branyon: 134.29
22. Mark Reynolds: 133.33
23. Nick Johnson: 132.44
24. Todd Helton: 132
25. Brad Hawpe: 131.1
26. Troy Tulowitzky: 129.44
27. Hideki Matsui: 129.38
28. Chipper Jones: 129.08
29. Justin Morneau: 128.51
30. Ryan Braun: 128.41
31. Bobby Abreau: 127.52
32. Matt Holliday: 127.47
33. Raul Ibanez: 127.09
34. Hanley Ramirez: 126.86
35. Mike Cameron: 126.8
36. Justin Upton: 126.36
37. Jack Cust: 126.16
38. Jason Bartlett: 126.15
39. Dan Uggla: 125.7
40. Evan Longoria: 125.61
41. Adam Lind: 125.47
42. Johnny Damon: 125.47
43. Jason Kubel: 125
44. Andre Ethier: 124.39
45. Nelson Cruz: 123.79
46. Casey Blake: 123.64
47. Ryan Zimmerman: 123.59
48. Ian Kinsler: 123.12
49. Torii Hunter: 123.05
50. David Ortiz: 123.02
51. David Wright: 122.78
52. Chone Figgins: 122.71
53. Kosuke Fukudome: 122.54
54. Miguel Cabrera: 122.44
55. Dexter Fowler: 122.41
56. Pablo Sandoval: 122.4
57. Luke Scott: 121.81
58. Nate McLouth: 121.54
59. Brian Roberts: 121.54
60. Marco Scutaro: 121.28
61. Adam Laroche: 121.23
62. Victor Martinez: 121.03
63. Kendry Morales: 120.43
64. Grady Sizemore: 120.18
65. Paul Konerko: 120.04
66. Derek Jeter: 118.51
67. Denard Span: 118.34
68. Chris Coglan: 117.96
69. Scott Rolen: 117.59
70. Matt Kemp: 117.33
71. Carl Crawford: 117.33
72. Michael Cuddyer: 116.33
73. Jermaine Dye: 115.74
74. Curtis Granderson: 115.73
75. Dustin Pedroia: 115.62
76. Brian McCann: 115.57
77. Michael Bourn: 115.42
78. Michael Young: 114.9
79. Shane Victorino: 114.8
80. Felipe Lopez: 114
81. Billy Butler: 113.27
82. Nick Markakis: 111.63
83. Jacoby Elsbury: 111.51
84. Ryan Ludwick: 111.43
85. Orlando Hudson: 110.64
86. Marlon Byrd: 110.13
87. Stephen Drew: 110.13
88. David Dejesus: 108.59
89. Mark Derosa: 108.57
90. Ichiro Suzuki: 108.53
91. Brandon Inge: 108.22
92. Luis Castillo: 108.2
93. Adam Jones: 108
94. Nyjer Morgan: 107.77
95. Alberto Callaspo: 107.73
96. Yunel Escobar: 107.58
97. Asdrubal Cabrera: 107.42
98. James Loney: 107.15
99. Jorge Cantu: 107.14
100. Hunter Pence: 106.74
105. Alfonso Soriano: 106.43
106. Skip Schumaker: 105.55
117. Colby Rasmus: 101.55
Btw, Miguel Tejada still sucks: 92.1 (#137)
Next, I just wanted to show some leaders in the different component stats:
- BEF leaders are Carlos Pena, then Mark Reynolds, Russell Branyon, Nelson Cruz, Ryan Howard, Raul Ibanez, Ian Kinsler, Albert Pujols, Nick Swisher, Prince Fielder, Adrian Gonzalez, Adam Dunn, & Jason Werth
- BEA leaders: Albert Pujols, Prince Fielder, Joe Mauer, Derrek Lee, Ryan Howard, Joey Votto, & Mark Texeira
- EDF leaders: Nick Johnson, Albert Pujols, Chase Utley, Chipper Jones, Lance Berkman, & Adrian Gonzalez
- ELF leaders (heh): Nick Johnson, Albert Pujols, Kevin Youkilis, Chase Utley, Shin-Soo Choo, Adam Dunn, Prince Fielder, Lance Berkman, Ben Zobrist, Chipper Jones, Todd Helton, Chone Figgins, Adrian Gonzalez
- P/PA leaders (or wear down the pitcher award): Jason Bay & Jason Werth (4.5 pitches per plate appearance!), Kevin Youkilis, Nick Johnson, Adam Dunn, Casey Blake, Todd Helton, & Nick Swisher
- the unlucky slowass awards go to (GDP leaders): Miguel Tejada, Yadier Molina, Evan Longoria, & Hunter Pence (Skip and Colby hardly grounded into any DPs
- stolen base leaders: Jacoby Ellsbury, Michael Bourn, Carl Crawford, Nyjer Morgan, Chone Figgins, BJ Upton
- big whiffers: Mark Reynolds, Ryan Howard, Jack Cust, Adam Dunn (btw, Yadi struck out the least of anyone!)
- take one for the team award: Chase Utley, Shin-Shoo Choo, Jason Kendall, Brandon Inge, Kevin Youkilis
- the walk-takers: Adrian Gonzalez, Adam Dunn, Albert Pujols, Prince Fielder, Chone Figgins, Chipper Jones
So basically, I hope it's more accurate than last year's OAR rankings, and that some insights might be gained on seeing what players are doing overall offensively... instead of procrastinating longer, here's the formula I came up with (as a creative-type bass playing accounting clerk non-statistician dude):
EDF=BB+HBP+.8*SF-.9*GDP/PA+.8*SO
ELF=EDF*4+BABIP/2
BEA=BEF*BA
BEF=TB+.8*SB/H+CS
OAR=EDF+ELF+BEA+.1*BEF+PPA/30*100
I came up with that basically as an educated guess/guesstimate, I'm sure it has some pretty loud distortion in the numbers... I think it would be interesting to see if this is more accurate than last year's version of OAR, or if my revisions are helping. Also, whoever wants to help out with this project, be my guest. I think it would be cool to see after park adjustments after another revision or two. What do you guys think?
60 comments
|
5 recs |
Do you like this story?
Comments
thanks man
I’d really appreciate it!
by Cards Fan in Chitown on Jan 28, 2010 1:55 PM EST up reply actions
if it seems like I have made some progress
and the stat is more accurate (I think you said last year’s was pretty much average, like 75% accurate or something, I don’t totally get what the testing is you are doing)…. do you think that it would be easy to use the formula on other years of data? I’m sure you have a much easier way than what I did, heh. lots of calculator button pushing.
by Cards Fan in Chitown on Jan 28, 2010 2:00 PM EST up reply actions
He's determining how close your formula is to determining offensive value by how closely it maps to real runs scored/conceded records
things like wOBA match up very well. Things like BA less so.
RELEASE THE CENTIQUID!!!!
by Felonius_Monk on Jan 29, 2010 5:30 AM EST up reply actions
In fairness, wOBA is fit to those same metrics, so it's a bit of an unlevel playing field
there are a ton of free parameters one could tweak in the above to get a better fit to runs
by brackenthebox on Jan 29, 2010 7:34 AM EST up reply actions
Right, a regression run estimator would naturally have the best fit to whatever sample you are testing it on
by vivaelpujols on Jan 30, 2010 6:23 AM EST up reply actions
which is why building a regression by just putting a bunch of free parameters in your model
doesn’t tell you much at all. For one example, Ptolemy’s theory of planetary motion more accurately predicted the planets’ motion than Copernicus’s. Only it did so at the expense of many more free parameters.
They say that it's never too late, but you don't get any younger...
I agree
However, wOBA (linear weights) isn’t necessarily regression. A regression estimates the value of each parameter based on how they combined to create team, or game, or inning runs. Linear weights looks at the actual value of each event based on change in run expectancy. It’s really a perfect stat.
by vivaelpujols on Jan 30, 2010 8:15 PM EST up reply actions
That still doesn't get you out of proper model building
There is no way to just say ’I’m going to add a bunch of inputs without thinking’ and then go and construct a meaningful answer. You need some sort of logic behind how those things should theoretically combine.
I am very deeply skeptical of any model that linearly combines inputs like OBP and SLG, etc.
They say that it's never too late, but you don't get any younger...
your first paragraph
is very similar to the argument Tango et all make for linear weights. It has logical / meaningful underpinnings.
I assume the second paragraph was a general indictment, not necessarily aimed at wOBA? wOBA can be approximated by linearly combining SLG and OBP, but is truly a linear combination of the set of events that make up offense (more or less anyhow).
by stevesommer05 on Jan 31, 2010 8:49 AM EST up reply actions
The linear weights in wOBA still are based on a shaky model
They assume that the form of a player’s value is of the form
Value = c1 * OBP + c2* SLG.
This is problematic for a few reasons:
1) if I double a players OBP and SLG, a formula like this tells me that I am doubling that players’ value. I contend that you would far more than double that players’ value.
2) SLG and OBP are not independent variables. They detectably correlate, both abstractly and in fact. While a regression model can, at the expense of added complexity in the model, account for this cross-correlation, but a model like the above cannot
3) This still starts without a clear theory about how the outcomes of ABs result in runs. It is just a more sophisticated (and more empirical ) version of the heuristic arguments that Bill James always used. There’s not anything wrong with that, it’s just not ‘irrefutable truth’ in the way that many statheads would argue. Like I said above, Ptolemy was more accurate at predicting the planets’ orbits than Copernicus.
They say that it's never too late, but you don't get any younger...
This is incorrect
They assume that the form of a player’s value is of the form
Value = c1 * OBP + c2* SLG.
Combining OBP and SLG is a way to approximate wOBA (linear weights per PA scaled to OBP); however, it’s not the correct way and not the way that FanGraphs does it, or how B-R computes batting runs.
Linear weights treats each individual event separately. They look at the average change in run expectancy following each event. That is the definition of context neutral value. Read these:
http://www.tangotiger.net/runscreated.html
http://www.tangotiger.net/rc2.html
http://www.tangotiger.net/rc3.html
by vivaelpujols on Jan 31, 2010 2:36 PM EST up reply actions
Nope
From the second page:
R = (H-HR+BB) x score rate + HR
You can statistically regenerate that to be a linear combination of SLG and OBP, and then you statistically determine the score rate. That the score rate is determined with the average number of players on base, and the average number of outs, and for a specific year and league gives you a more specific model, but it does not give you a more robust model. You are still building the assumption that value is a linear combination of total bases and on base percentage.
They say that it's never too late, but you don't get any younger...
So then I don't really see the problem
The presence of score rate in the BaseRuns formula allows it to be dynamic to differing run environments. And Linear Weights is not built from a regression without logical underpinnings, it is build from the exact change in run expectancy following an event. The fact that it can be reproduced with OBP and SLG (and that’s not even true because it includes SB and CS) is not in itself a detractor of the stat.
Please show how linear weights breaks down, why it lacks logical underpinnings, and why it isn’t the definition of context neutral value.
by vivaelpujols on Jan 31, 2010 3:54 PM EST up reply actions
Because someone who triples league or team or whatever
average is more than one and a half times as good as someone who doubles average. Linear models tend to be really good at approximating performance near an average, but tend to be really poor at evaluating outliers.
They say that it's never too late, but you don't get any younger...
Could you do a comparison of the 2009 top 100 by OAR vs the top 100 by wOBA
obviously with some minimum PA limit? It’d be interesting to see how close the two are.
RELEASE THE CENTIQUID!!!!
yeah
I was thinking about doing that… although at first glance they were pretty different. I think including all the other stuff like stolen bases really changes the dynamic
by Cards Fan in Chitown on Jan 28, 2010 1:56 PM EST up reply actions
wOBA includes stolen bases
As well as a speed variable which measures how good a hitter is at (say) going from 1st to 3rd on a single (which is more important in most cases than SB%), at least in the FanGraphs variant. I believe the original version in The Book didn’t. It includes the majority of things you included here, except more esoteric (and, I’d imagine, without much proof, relatively low-value) things like P/PA (which is probably taken into account to some extent by other parts of the formula, e.g. OBP and BB rate) and GIDP (which, to at least some extent, is partly out of a batter’s ability to control).
And re: stolen bases, does your formula work in such a way that a CS removes about 3 times as much value as a SB adds? I think that’s the general logic.
RELEASE THE CENTIQUID!!!!
by Felonius_Monk on Jan 29, 2010 5:36 AM EST up reply actions
huh
I had no idea that wOBA included stolen bases
by Cards Fan in Chitown on Jan 29, 2010 1:14 PM EST up reply actions
This is awesome!
I applaud your efforts. I am by no means a stat guy, but am working on it, and your bit seems solid to me! One thing I notice- at the top you mention accounting for speed and/or stolen bases, and that someone who uses this ability well could rank near the same level as power hitters. At first glance I would not be able to say that your ranking system rewards them very much, but this may be due to the fact that there’s not really anyone that utilizes this part of the game as well as history (and other things) might dictate they could. I suppose this part of a ranking system would be very subjective as to how highly it is weighted/rewarded. I may be way off point here, but those were my thoughts when looking through the rankings.
I think it really helped Chase Utley
but so did his on-base skills
by Cards Fan in Chitown on Jan 28, 2010 1:57 PM EST up reply actions
I think the "real" value of a SB and a CS is pretty well established
obviously in real life it depends greatly on context, but I believe the average SB isn’t worth very much in real run value (certainly far less than getting on base once, and I believe less than getting from 1st to 3rd on a single), and is worth about 3 times less than a CS is to the fielding team (i.e. you need somewhere approaching 3/4 of your attempts to “work” to be a + base-stealer, otherwise you’re hurting your team; as something of an aside, I’ve always thought that level was a bit too high, because AFAIK it doesn’t take into account how often pickoff and CS attempts “create” errors (they’re just counted as a successful steal, when in reality they will add extra bases), as well as the incalculable value that having a base-stealer on 1B creates in terms of how the pitcher pitches and what pitches the catcher calls).
I think there are much harder elements of base-running to quantify; like I said above, the “other” advantages (other than purely advancing one base) of having a speedy guy, which increase errors, and potentially may impact the pitcher negatively, and (a consistently under-rated element of baserunning) the ability of runners to judge hits and GB etc. well and take extra bases (say, going first-to-third, or sneakily taking 3B from 2B on a GB to the SS) better than average. I still don’t think people have got a very good handle on those elements, and I don’t think it’s unreasonable to say that traditional wOBA and WAR probably slightly under-rates elite baserunners.
RELEASE THE CENTIQUID!!!!
by Felonius_Monk on Jan 29, 2010 5:42 AM EST up reply actions
I want to see the OAR for Bonds' best season.
And Ruth’s. Maybe a few other amazing historical players.
I'm living in an age that calls darkness light...
good idea
I want to see how it compares to Albert Pujols this year.
by Cards Fan in Chitown on Jan 28, 2010 1:57 PM EST up reply actions
Great work.
And I’m glad to see so many support your hard work, unlike the way this community totally abandoned the stat I worked hard to create.
And I just don’t get it either. You can see my stat on the back of every baseball card, yet this “community” will never get behind SO.
But, I guess I have to realize that while my work has influenced the Baseball Writers in years past, who look very favorably upon my stat SO, which we all know stands for Statistically Outstandings. Each year, the truely enlightened vote for Ryan Howard (who always has one of the highest SO rankings each season) for MVP, it will never achieve the accolades it truely deserves here at VEB.
Long live SO and long live Ryan Howard for being so damn Statistically Outstading!
Think of how stupid the average person is, and realize half of them are stupider than that.
well
there is a contingent that does think it is a colossal waste of time. 16% of voters!
by Cards Fan in Chitown on Jan 29, 2010 1:15 PM EST up reply actions
I've also seen it recorded as "K"
presumably for “Kingly”
by brackenthebox on Jan 29, 2010 1:22 PM EST up reply actions
In all seriousness, why 'K'?
striKeout?
They say that it's never too late, but you don't get any younger...
from wikipedia (I promise I didn't write the entry)
The use of “K” for a strikeout was invented by Henry Chadwick, a newspaper journalist who is widely credited as the originator of the box score and the baseball scorecard . . . The letter “S” was used to coin “sacrifice” so Chadwick decided to use “K”, with “K” being the last letter in “struck.” Chadwick also invented many other baseball scoring abbreviations, such as using numbers to designate player positions.
by brackenthebox on Jan 30, 2010 4:39 PM EST up reply actions
when i was younger
i would think that the “K” looked like a batter fully extended swinging, and a backwards “K” looked like a batter left with the pine on his shoulder, so to speak.
R.P.O.F.Y.M.
I kinda thought it had something to do with the 3 lines it takes to make the letter "K".
like 3 strikes? 3 lines.
Think of how stupid the average person is, and realize half of them are stupider than that.
interesting
Obviously you put some work into this….nice.
At first blush it seems like the guys that this stat favors really strongly are power hitters. Most of the speedy/slap hitter types are further down.
I have a kinda gut reaction to that as well
Adam Dunn at 5 and Ichiro at 90.
If I was putting together a lineup I’d take Ichiro over Dunn and not even think about it very much, which might be my problem.
Maybe that’s just me.
On the whole though, these rankings look more or less accurate.
They say sing while you slave but I just get bored
by Scarecrow7775 on Jan 30, 2010 8:40 AM EST up reply actions
Ichiro's better overall
but Dunn’s a much better hitter I’d say.
RELEASE THE CENTIQUID!!!!
by Felonius_Monk on Feb 1, 2010 5:54 AM EST up reply actions
this year Ichiro is lower than usual because he hardly walked at all
probably other reasons too for his low ranking
by Cards Fan in Chitown on Feb 1, 2010 2:40 PM EST up reply actions
for comparison
Bonds 2001 season (which didn’t have p/pa info for, so I just gave him the average of that) he put up OAR of 216. fuck
by Cards Fan in Chitown on Feb 2, 2010 8:22 PM EST reply actions
Vivaelpujols
hope I’m not bugging you, but have you had a chance to run the test on this stat yet? pretty interested in seeing the accuracy of this…. I have an idea to tweak the formula that might make it more accurate (basically lowering the on-base percentage portion of the stat). if it’s a major chore don’t worry about, but I was going to run the stat on some other big seasons like Bonds’ 2001 to compare players.
by Cards Fan in Chitown on Feb 2, 2010 9:20 PM EST reply actions
Oh, I'm sorry, I've been really busy
I should be able to get to it sometime tomorrow.
by vivaelpujols on Feb 2, 2010 10:25 PM EST up reply actions
cool
no rush, just getting anxious/bored
by Cards Fan in Chitown on Feb 2, 2010 11:21 PM EST up reply actions
I just...
I can not imagine Bobby Abreau being worth more offensively than Matt Holliday.
How in the holy hell
has the #19 guy on this list not found a job yet? Seriously?
Can Colby round out our new MV3?
Do you mean #21, Russell Branyan?
Because I’m pretty sure Teixeira has a job.
"Forget it, spants. It's Chinatown." - tom s.
I don't get why teams always seem to undervalue that guy
does he have horrible defense? I don’t think so…. does he hit well above average. definitely… do they know something abou thim we don’t? does he eat babies?
by Cards Fan in Chitown on Feb 4, 2010 2:02 PM EST up reply actions
Injuries play a part
also maybe lack of teams who need a 1B/DH type. Positionally inflexible players can struggle in a market where there isn’t a big demand for that position. At least OFers can always slot in somewhere.
Orlando Hudson seems to have struggled to find a job two years in a row, despite being an above-average 2B.
I still wouldn’t mind Branyan as a bench lefty/3B vs righties sometimes option, but I suspect he wants to play everyday, and it seems he wants more money than he’s really worth as well.
RELEASE THE CENTIQUID!!!!
by Felonius_Monk on Feb 5, 2010 8:37 AM EST up reply actions
is he just not that good at 3B?
by Cards Fan in Chitown on Feb 5, 2010 12:35 PM EST up reply actions
Last time he played there (Brewers) he was somewhere in the region of average according to UZR
but that was two years and a knee surgery or two ago. I think with his age and injury issues a low-impact position would be better for him, so probably he’s more a 1B/DH now. That said, yeah, I still think he could handle 3 starts a week at 3B. He’s probably no worse than DeRosa there.
RELEASE THE CENTIQUID!!!!
by Felonius_Monk on Feb 8, 2010 4:49 AM EST up reply actions
just read an article at fangraphs
it’s about the White Sox and their need for a regular DH… wonder if they can sign Branyon
by Cards Fan in Chitown on Feb 5, 2010 8:54 PM EST up reply actions
i was under the impression
ozzie wanted to rotate guys at dh, which is why they didn’t take thome back.
"Some days I feel like the hypotenuse in a love triangle; others as if my lucky number is pi."
yeah that's exactly what I thought
especially given they’ve got two or three really poor defensive guys, and a good-hitting 1B (Konerko) who’s getting a bit old and might need a day or two off the field at some point.
I suppose you could argue that Branyan can play the field too, still, and he could make the odd start at 1B or 3B, so I guess it would work.
RELEASE THE CENTIQUID!!!!
by Felonius_Monk on Feb 8, 2010 4:51 AM EST up reply actions
Yah...
but when your rotation is Andruw Jones, Konerko, Juan Pierre, and Mark Kotsay, signing Russell Branyan to DH and give Konerko days off at 1B would seem to be a smart idea. Although Carlos Delgado would fit that bill too.
Can Colby round out our new MV3?

by 


















