BPA rankings

by asufan550 on Jan 6, 2009 9:23 PM EST

Read the site and fanposts daily, but haven't posted before. I was inspired by Cards fan in Chitown's recent post regarding OAR to share a statistic of my own. As we all know, OPS is an excellent statistic, but certainly has flaws. In an effort to more equally weight the components of OPS, I came up with BPA: Bases per Plate Appearance. The name is pretty self explanatory.

I think, with my most recent incarnation, that I've managed properly weight the values of your "normal" stats (H, 2B, BB, etc.), while taking into account SB's, SF's, HBP's, etc as well. Using a league average OBP (adjusted for each player's PA's), I've valued all hits greater than walks, etc based on the greater likelihood additional players will advance due to the live ball.

Pillory me if you must, but any feedback would be highly usefull and appreciated.

Top 100 (Note: My dataset includes all players with at least 50 PA's, so keep that in mind):

Rank	NAME	POS	BPA
1	Albert Pujols	1b	0.788974
2	Chipper Jones	3b	0.680988
3	Rafael Furcal	ss	0.677308
4	Taylor Teagarden	c	0.649303
5	Cody Ransom	1b	0.637312
6	Carlos Lee	lf	0.633368
7	Lance Berkman	1b	0.624951
8	Nelson Cruz	rf	0.624765
9	Manny Ramirez	lf	0.616236
10	Jerry Hairston	ss	0.608195
11	Hanley Ramirez	ss	0.607029
12	Matt Holliday	lf	0.606782
13	Mark Teixeira	1b	0.602254
14	Dustin Pedroia	2b	0.598559
15	Chase Utley	2b	0.597546
16	Ramon Santiago	ss	0.596816
17	Ian Kinsler	2b	0.591132
18	Jose Reyes	ss	0.590194
19	Carlos Quentin	lf	0.587953
20	Aubrey Huff	dh	0.586604
21	Jimmy Rollins	ss	0.581795
22	Grady Sizemore	cf	0.580507
23	Nate McLouth	cf	0.569673
24	Carlos Beltran	cf	0.565696
25	Milton Bradley	dh	0.564401
26	Alex Rodriguez	3b	0.563035
27	Kevin Youkilis	1b	0.561108
28	Mike Fontenot	2b	0.560736
29	Nate Schierholtz	rf	0.556422
30	Brian McCann	c	0.553364
31	David Wright	3b	0.549587
32	Andre Ethier	rf	0.547429
33	Josh Hamilton	cf	0.545704
34	Joe Mauer	c	0.543907
35	J.D. Drew	rf	0.542733
36	David Ortiz	dh	0.542125
37	Johnny Damon	lf	0.539329
38	Justin Morneau	1b	0.538064
39	Aramis Ramirez	3b	0.537246
40	Martin Prado	3b	0.536881
41	Mike Napoli	c	0.536709
42	Chris Dickerson	lf	0.535968
43	Brian Giles	rf	0.534272
44	Ryan Doumit	c	0.530735
45	Oscar Salazar	ph	0.530731
46	Denard Span	rf	0.528153
47	Shane Victorino	cf	0.524339
48	Brian Roberts	2b	0.524062
49	Nick Markakis	rf	0.520519
50	Ichiro Suzuki	rf	0.520371
51	Kazuo Matsui	2b	0.519561
52	Nick Johnson	1b	0.519207
53	Jody Gerut	cf	0.519098
54	Shin-Soo Choo	rf	0.51868
55	Conor Jackson	lf	0.518417
56	Jason Bay	lf	0.516456
57	Russell Branyan	3b	0.516339
58	Vernon Wells	cf	0.515218
59	Ryan Ludwick	rf	0.514982
60	Vladimir Guerrero	rf	0.513712
61	Joey Votto	1b	0.513657
62	Carlos Delgado	1b	0.51076
63	Randy Winn	rf	0.509316
64	Marlon Byrd	cf	0.509283
65	Alfonso Soriano	lf	0.508846
66	Prince Fielder	1b	0.508135
67	Anderson Hernandez	2b	0.507035
68	Curtis Granderson	cf	0.505214
69	Pablo Sandoval	1b	0.503773
70	Jason Giambi	1b	0.503758
71	Jermaine Dye	rf	0.501787
72	Ryan Braun	lf	0.500628
73	Ryan Shealy	1b	0.500173
74	Ben Zobrist	ss	0.499322
75	Placido Polanco	2b	0.498913
76	Stephen Drew	ss	0.494688
77	Juan Pierre	lf	0.493574
78	Ty Wigginton	3b	0.492517
79	Mike Sweeney	dh	0.491776
80	Mark DeRosa	2b	0.491566
81	Miguel Cabrera	1b	0.49053
82	Moises Alou	lf	0.490356
83	Doug Mientkiewicz	ph	0.489349
84	Mike Aviles	ss	0.489276
85	Pat Burrell	lf	0.487752
86	Ryan Spilborghs	ph	0.486826
87	Cristian Guzman	ss	0.485362
88	Raul Ibanez	lf	0.483994
89	Gabe Kapler	cf	0.483319
90	Xavier Nady	rf	0.482576
91	Adam Dunn	lf	0.482244
92	Melvin Mora	3b	0.482014
93	Joe Inglett	2b	0.48178
94	Troy Glaus	3b	0.480363
95	Hank Blalock	1b	0.480266
96	Mike Rivera	c	0.47972
97	Adrian Beltre	3b	0.478715
98	David DeJesus	lf	0.478495
99	Bengie Molina	c	0.478155
100	Jacoby Ellsbury	cf	0.477888

Tweet Comment 46 comments | 8 recs |

Do you like this story?

Comments

i like this...

… don’t have much more to say. although i am curious as to how (and why) you weighted hits and walks differently. for a hitter, a single and a walk should be counted as the same, since whether or not there is a baserunner is essentially random from the hitter’s p.o.v. also, giving walks less weight than hits skews against guys who draws lots of IBBs when runners are on base. if this were a “per-AB” measure than that wouldn’t matter, but since it’s a “per-PA” measure, it does.

i understand that there may sometimes/often be more utility to a hit than a walk, but unless you’re going to go full-bore and include some sort of situational-hitting stat — or at least build a more nuanced term including chance that a runner is on base, chance that he advances, not discounting IBBs, etc. — then i don’t think you should be prejudiced against walks.

but i certainly like the idea.

by kindred on Jan 6, 2009 10:19 PM EST reply actions

whether or not there's a baserunner is essentially random

but on average, it happens in a pretty significant number of times, and a hit advances a runner more than a walk. Though, of coure, balls in play do end in double plays more often too.

They say that it's never too late, but you don't get any younger...

by Valatan on Jan 6, 2009 10:35 PM EST up reply actions

sure...

… but if we’re trying to measure the ability of individuals then it could skew results. for example, if Rick Ankiel bats fourth or fifth and hits a single, there’s a stronger chance that Pujols will be on base to advance than when Kennedy (say) bats ninth, after the pitcher. and whoever the lead-off hitter is will get 150+ ABs every year where he is guaranteed to have no chance to advance a runner, and where a walk is just as good as a single.

by kindred on Jan 6, 2009 11:05 PM EST up reply actions

That's why you'd

use averages that aren’t biased to specific situations. And on average, hits are worth more than walks because they generate more runs.

by haltz on Jan 6, 2009 11:41 PM EST up reply actions

it's not a measure of run creation...

… it’s a measure of bases per plate appearance.

by kindred on Jan 7, 2009 1:02 AM EST up reply actions

It's his stat

He can make it whatever he wants.

by vivaelpujols on Jan 7, 2009 1:16 AM EST up reply actions

Hits get runners

closer to home on average. Ie, more bases.

by haltz on Jan 7, 2009 1:32 AM EST up reply actions

oops

didn’t see your post below.

by haltz on Jan 7, 2009 1:33 AM EST up reply actions

It's trying to figure out who is good at run creation

Just like all offensive stats intentions. Who cares about anything else?

Not afraid to nitpick

by joker24 on Jan 8, 2009 2:30 AM EST up reply actions

oh, and you should definitely post more...

… especially when you’ve got new stats to share!

by kindred on Jan 6, 2009 10:20 PM EST reply actions

This is very good

It seems very thorough and I love the fact that it includes steals because it values the complete ballplayer a little more. Also I like the fact that hits are move valued than walks.

by vivaelpujols on Jan 6, 2009 10:22 PM EST reply actions

this sounds like an attempt at linear weights

on a rate basis. . . unless I’m misinterpreting things.

Nice Robot

by azruavatar on Jan 6, 2009 10:23 PM EST reply actions

Yeah, I read

it and thought wOBA or EqA. Even OPS breaks down to values that are similar values. Anyway, I’m assuming that this is independent of any knowledge of linear weights, so it’s impressive that you came up with this idea. To me anyway.

by haltz on Jan 6, 2009 11:38 PM EST up reply actions

It is an impressive thought process if original.

Lord knows I don’t have the creativity for something like this.

Nice Robot

by azruavatar on Jan 6, 2009 11:41 PM EST up reply actions

Just you wait

Azruavatar v 2.0 is going to blow some freaking minds.

hecanthithecanthithecanthithecanthit

by Alxfritz on Jan 8, 2009 12:18 AM EST up reply actions

Good work, devil

I, for one, would like to see more about your methodology. How did you value those extra bases? Did doubles get greater than twice the weight of a walk b/c of the likelihood someone might score from 1st? I’m also not crazy about including sacrifice flies since, in many cases, it’s not at all clear that those batters “gave themselves up.” All that said, I like what you’ve done here and would be interested in reading more about it to help me understand why Pujols is worth .79 bases per PA.

by chuckb on Jan 6, 2009 10:25 PM EST reply actions

Thanks for the feedback

Yes, doubles, triples, and home runs all get more value than a walk. My newest method (I’ve been working on this off and on for roughly a year) of accomplishing this is to add the probability of one runner on, two runners on, and three runners on to the amount of total bases each hit/whatever provides, and then multiply that by my perceived likelihood that a runner will gain more than one base from the hit. For instance a double would be worth 2*(1+(avgobp*1.33)+(avgobp2*1.33)+(avgobp3*1.33)). I decrease the value of a walk/ibb/hbp(same as walk) by then multiplying each of the likelihoods my admittedly arbitrary prediction that a pitcher is less likely to walk a man with runners on first, first and second, etc: 1*(1+(avgobp*.5)+(avgobp2*.25)+(avgobp3*.125)). As I said, those are arbitrary numbers, and probably a far less than perfect way of doing things, but one that enables the outcome (BPA) to tell a more complete story.
Sac flies and hits are included because the idea of BPA is to determine how many bases a batter contributes to the team. This is also why gdp’s, CS, and pickoffs are included.
I also have attempted to penalize strikeouts (but not to the extent that some would) by using the following calculation: -(k/(1+(avgobp^3)). Again, a somewhat arbitrary calculation, but one that I think makes the stat more complete.
Note: the numbers I originally posted are slightly off due to an error in my calculations. Numbers for notable Cardinals are:
Pujols: .8513
Glaus: .5304
Ankiel: .4893
Molina: .4948
Ludwick: .5712
Greene: .2854
Kennedy: .4368
Schumaker: .4939
League average (adjusted for PA’s): .4601

by asufan550 on Jan 6, 2009 11:44 PM EST up reply actions

okay...

… i think i misunderstood at first. so the measure is TEAM bases per INDIVIDUAL plate appearance? Thus, a double that scores a runner from first is worth 5 total bases, discounted by the probability that a runner is actually on first (which is less than 1.00).

am i reading you right? if so, then treating walks differently from hits makes sense, as does adding value to sac flies. although you still need to correct for IBB and HBP, i think, since those are out of the batter’s control.

by kindred on Jan 7, 2009 1:09 AM EST up reply actions

Yes. I do treat IBB’s differently(much lower ratios). HBP’s, however, are currently treated the same as a walk. I’m no so sure that they are completely out of a batter’s control. See Biggio and Quentin – guys that have always had high HBP numbers.

by asufan550 on Jan 7, 2009 2:37 AM EST up reply actions

yeah

I think HBP should be same as BB

4B - beer baseball bands blog

by Cards Fan in Chitown on Jan 7, 2009 1:05 PM EST up reply actions

Can you explain

this to me: “(k/(1+(avgobp^3))”

what does avgobp mean? I think you are taking too large a chunk out for strikeouts (which combined with less GIDP, should be fairly negligible when it comes to the players TB or runners advancing).

by haltz on Jan 7, 2009 1:45 AM EST up reply actions

will post complete cards’ roster tomorrow for those interested.

by asufan550 on Jan 7, 2009 2:39 AM EST up reply actions

awesome

4B - beer baseball bands blog

by Cards Fan in Chitown on Jan 7, 2009 1:04 PM EST up reply actions

interesting to see the players with fewer PAs

but where’s Mather? he had more than 50 right? Barton? etc

I think you have a better grasp of the math involved than I do, at least it seems that way… some of the stuff you mentioned I’m not totally following, but hey, I didn’t do the best job of describing mine either. I’m a bit leery to divulge my formula, but I suppose I should.

I also like the idea of giving more value to hits than walks, even if just a little bit more. interesting that you’ve been working on this for a year! we should collaborate perhaps, I only started after the season was over, and it has become apparent to me that this project is going to take more work than I thought; although my favorite part was the creativity involved and naming the stat. glad I could inspire, and I’d like to hear a more in-depth description. I’ll probably go into the components of my stat a little more when I get the time; right now I’ve been writing this:

this line is dedicated to '09

by Cards Fan in Chitown on Jan 7, 2009 2:21 AM EST reply actions

Pujols number is stunning

I assumed he would be near the top (if not #1). But the gap between his numbers and the next player (Jones) is amazing…

by JWO on Jan 7, 2009 8:54 AM EST reply actions

Unfortunately, so is Greene's.

by vinniefromjersey on Jan 7, 2009 10:50 AM EST up reply actions

yeah

same in my OAR stats, he was like 10 points higher than everyone else, and the ranking was pretty tight for the rest of it.

4B - beer baseball bands blog

by Cards Fan in Chitown on Jan 7, 2009 1:05 PM EST up reply actions

thought this deserved another rec

4B - beer baseball bands blog

by Cards Fan in Chitown on Jan 7, 2009 3:06 PM EST reply actions

For those still interested

Another slight change to the formula (told you it was a work in progress). Cardinals and their ranks(out of 567):
Molina: .4909 (101)
Pujols: .8046 (1)
Kennedy: .3951 (272)
Ryan: .3355 (372)
Greene: .2540 (455)
Glaus: .4934 (96)
Ludwick: .5293 (58)
Stavinoha: .1515 (505)
Barton: .3387 (368)
Ankiel: .4516 (157)
Schumaker: .4499 (163)
Just for kicks:
Miles: .4420 (180)
Lopez: .3803 (297)
Izturis: .4537(!) (152)
Roberts: .5378 (47)
Adjusted Avg: .4222

Entire list, along with subsequent updates and other invented stats, will be up at overzealousfan.blogspot.com in the near future.

by asufan550 on Jan 7, 2009 11:57 PM EST reply actions

I take it Iz2's stat padding SBs help?

hecanthithecanthithecanthithecanthit

by Alxfritz on Jan 8, 2009 12:19 AM EST up reply actions

it's starting to look like Izturis wasn't as bad as we thought he was

4B - beer baseball bands blog

by Cards Fan in Chitown on Jan 8, 2009 12:59 AM EST up reply actions

If you don't mind my asking

are you developing this for a particular source? Your blog, website, book…someone else’s? I’m just curious.

The Personal Finance Playbook.

by Toddius on Jan 8, 2009 12:45 PM EST up reply actions

his blog is at

http://overzealousfan.blogspot.com

4B - beer baseball bands blog

by Cards Fan in Chitown on Jan 8, 2009 1:49 PM EST reply actions

meant to reply to Toddius

doh

4B - beer baseball bands blog

by Cards Fan in Chitown on Jan 8, 2009 1:49 PM EST up reply actions

thanks

I’ll check it out.

The Personal Finance Playbook.

by Toddius on Jan 8, 2009 4:28 PM EST up reply actions

Any way you could throw up a link for a spreadsheet file of this?

I’m not interested in the process, that’s what you’ve come up with and I don’t want to steal your work, but I’m interested in throwing the values into my fantasy baseball cheat sheet Excel spreadsheet for 2009, along with the Marcel projections, Bill James, etc.

"I just wish that the late Harry Caray were still around so I could hear him mispronounce 'Kosuke Fukudome' every fukun' night" -- Dennis Miller

by fourstick on Jan 9, 2009 2:56 PM EST reply actions

Im just curious...

… in the last set of stats you provided, I assume, the strikeout penalty is included. How much does it affect/help Yadi’s and Pujols’ BPA?

Yadi swings and hits a high fly ball... Endy Chavez goes back, to the track, to the wall... ITS A GUNNER!! Yadi gives St. Louis the lead in the top of the ninth!

by Paulspike on Jan 12, 2009 2:45 PM EST reply actions

yeah

I wonder how much a strikeout should penalize an offensive player’s stats…

4B - beer baseball bands blog

by Cards Fan in Chitown on Jan 12, 2009 11:37 PM EST up reply actions

Formula is still being tweaked

However, in that incarnation, Pujols (well, everybody) was penalized far too many TB’s (52) due to an error in my calculations. The actual number should be 2, leaving him with a BPA around .8826. FWIW, Yadi should be at .5055. Average changed to .5815.
I’m working on the formula tonight. Will have updated numbers at my blog shortly.
Thanks, everyone, for all the interest and input. I’m hoping to also have a RC stat up shortly based on these numbers.

http://overzealousfan.blogspot.com

by asufan550 on Jan 13, 2009 9:58 PM EST up reply actions

Alright. It's done for the foreseeable future.

I found some helpful statistics in my spreadsheet that I had failed to utilize before. I believe this is the most accurate figure yet.
Cards:
Molina: .5440 (344)
Pujols: .9460 (2)
Kennedy: .5340 (360)
Ryan: .4857 (432)
Greene: .4947 (424)
Glaus: .6861 (100)
Ludwick: .7990 (18)
Stavinoha: .3276 (533)
Barton: .5628 (302)
Ankiel: .6959 (90)
Schumaker: .5792 (264)
Just for kicks:
Miles: .5623 (303)
Lopez: .5522 (324)
Izturis: .5344 (359)
M. Young: .5763 (274)
Adjusted Avg: .6150

Turns out I was over-penalizing strikeouts by quite a bit (in my eyes). I used these numbers to come up with a RC/27 stat that ended up pretty close to James’, and pretty close to average RPG last year. Both of these facts make me quite pleased and lead me to believe that I’ve created an accurate, useable statistic.

http://overzealousfan.blogspot.com

by asufan550 on Jan 13, 2009 11:28 PM EST reply actions