Nomadsfest Sox Fans

A forum for the old AOL board Sox fans and others.


    LUCKY FLYBALLS

    Share
    avatar
    Deplorable Mark
    Roof Shot

    Posts : 1721
    Join date : 2016-09-16

    Re: LUCKY FLYBALLS

    Post by Deplorable Mark on Tue Jun 13, 2017 2:25 pm

    You would have us believe that all these practical and logical issues were ignored when discussions about tendering Garcia occurred. Because all that counted was Kenny's fervent belief in Garcia's "tools" which had failed to produce for three solid years.  So who's inventing imaginery concepts of what went on in the front office ?


    ****************************


    And you would have us believe the converse.  With the entire front office throwing their hands in the air proclaiming who gives a flying F, we will suck anyway.  Plus this tools stuff is so 1960's jock.


    Bottom line, you don't keep him without a belief that the move will pay off


    PS, I have the HAWK on my side
    avatar
    Deplorable Mark
    Roof Shot

    Posts : 1721
    Join date : 2016-09-16

    Re: LUCKY FLYBALLS

    Post by Deplorable Mark on Tue Jun 13, 2017 2:54 pm

    As to your constant and typically false claim about this 1.8 stuff, it still has not been proven to be fundamentally wrong by the community that truly understands this stuff.   This study (now at 1.7 apparently) was revisited about 4 years and the issue you raised was also covered:

    "Also: it's true that when you have two dependent variables that correlate with each other, the coefficients are unpredictable. But that would show up in confidence intervals and significance. In this case, we have enough data that the mutual correlation is not a problem, that way. "
    http://blog.philbirnbaum.com/2013/05/the-obpslg-regression-puzzle.html

    Bottom line: these guys who understand statistics to a PHD level aren't going to ignore an easy-to-spot flaw in their study and still go on record with their findings.  Somehow you think that because YOU spotted something that no one else in the Statistical community did even though they typically rip apart any study with flaws.  LMAO...


    **************************


    ONCE AGAIN YOU PROOF YOUR COMPLETE LACK OF KNOWLEDGE, NOT MINE


    1.8, according to that journalism drop out you consider some sort of math PhD is the Onbase coefficient divided by the slugging coefficient.  That is mathematical masterbation.  Not serious research.


    BTW, did you even read the above article.  it starts out


    When I wrote that post, I thought I knew the answer ... but I actually hadn't.  So I spent the last few days trying to figure it out.  I actually jumped around among a whole bunch of possibilities, and hit a lot of dead ends ... but I think I finally got somewhere.  Here's the current state of my thinking.  

    As usual, I could be wrong.  I was wrong a few times in the course of working on all this ...




    So apparently you socalled Phd's are still struggling with the issue.  and the part you posted is just pure gibberish to anybody that understands math.

    ***************************

    Another thing, your phony mathematical masterbation is completely useless to a GM.  You can't take away X amount of slugging and replace it with Y amount of on base and expect 70% more

    Finally, where is the proof Phil Birnbaum is a PhD?!?!?!?
    avatar
    Deplorable Mark
    Roof Shot

    Posts : 1721
    Join date : 2016-09-16

    Re: LUCKY FLYBALLS

    Post by Deplorable Mark on Tue Jun 13, 2017 2:59 pm

    Let Roger post enough, and he will eventually hang himself.

    This is from the comment section from his little link that he thought proofed 1.8 means something

    While I don't agree with the regression tactic Anonymous says above (Phil, you have done this before with linear weights), I still just don't understand the "point of OBP vs. point of SLG".

    The ultimate result becomes misleading when we think about these: that a point of OBP is worth more than a point of SLG.

    But what does that actually mean? We know that a 1B (marginal), 2B, 3B, HR are worth more than a BB. By saying that OBP is more important than SLG is nonsense. OBP is nested within SLG, and any point in SLG necessarily increases OBP.

    So again, I think this is an ill-coneived question. These are non-mutually exclusive measures. We want to know on-base skill vs. hitting for power, but they just don't measure these things separately. While it might be an interesting regression exercise, I find it to be a pointless endeavor from a practical standpoint.



    Apparently the entire SABR community isn't lining up to proclaim 1.8 is some kind of Holy Grail.  Just like the KARK claimed in 2008 when he called bullshit on the journalism drop out Roger thought was a statistical genius.

    and I bet Roger still hasn't looked up the meaning of multicolinearity
    avatar
    Deplorable Mark
    Roof Shot

    Posts : 1721
    Join date : 2016-09-16

    Re: LUCKY FLYBALLS

    Post by Deplorable Mark on Tue Jun 13, 2017 3:49 pm

    Further proof 1.7 is simply wrong.


    If one were to except this theory, one would be forced to conclude the a player with a avg/obp/slg line of 280/360/460 is worth the same as 280/330/510 assuming all else is equal.


    If both have 500 PA, the first player has a runs created of 74 and an RC27 of 6.22.  the second player, the one with the lower obp, but higher slg, creates 79 runs with an RC27 of 6.36


    Clearly, you'd rather have the 2nd player.  So I guess the 50 point slg advantage more than made up the 30 point onbase decifit.  Now was it fair to use a 50 point slg advantage.  Well 30 times 1.7 equals 51.


    So once again, one of the people who like to deliberately disagree with the KARK and claim they are the ones that believe in science apparently don't understand the science as well as the KARK understands the science.
    avatar
    Deplorable Mark
    Roof Shot

    Posts : 1721
    Join date : 2016-09-16

    Re: LUCKY FLYBALLS

    Post by Deplorable Mark on Tue Jun 13, 2017 4:13 pm

    It gets even better folks.  Change player A to 260/360/460 and player B to 300/330/510


    Player A drops to 72 runs created and 6.04 RC27 while Player increases to 81 and 6.49 RC27


    How can this be with no change in the OBP-SLG differiental?!?!?!?!?


    I guess hits most be more valuable than walks!!!!  A complete contradiction of the 1.7 bullshit.
    avatar
    rmapasad
    Roof Shot

    Posts : 2423
    Join date : 2009-04-06
    Location : Northridge, CA

    Re: LUCKY FLYBALLS

    Post by rmapasad on Tue Jun 13, 2017 6:06 pm

    Deplorable Mark wrote:It gets even better folks.  Change player A to 260/360/460 and player B to 300/330/510
    Player A drops to 72 runs created and 6.04 RC27 while Player increases to 81 and 6.49 RC27
    How can this be with no change in the OBP-SLG differiental?!?!?!?!?
    I guess hits most be more valuable than walks!!!!  A complete contradiction of the 1.7 bullshit.


    Fundamental problem. You're using the old, old Bill James RC formula which while not discredited completely hasn't been used for years.  As noted in Fan Graphs way back in 2008:
    " Tonight we’ve completed the phasing out of Runs Created (RC) and Runs Created per 27 Outs (RC/27). In their place you will find two run stats based off wOBA, which is a sounder metric than the previously used Runs Created. "

    So you're using an unsound RC/27 metric as being the measure of how well another metric works.  Not much proof here.
    avatar
    rmapasad
    Roof Shot

    Posts : 2423
    Join date : 2009-04-06
    Location : Northridge, CA

    Re: LUCKY FLYBALLS

    Post by rmapasad on Tue Jun 13, 2017 7:47 pm

    Deplorable Mark wrote:And you would have us believe the converse.  With the entire front office throwing their hands in the air proclaiming who gives a flying F, we will suck anyway.  Plus this tools stuff is so 1960's jock.
    Bottom line, you don't keep him without a belief that the move will pay off
    PS, I have the HAWK on my side

    Hawk ?  Please.  I can see Kenny gilding the lily about this story to Hawk and Hawk was more than happy to be his on-air promoter. Sure, Kenny loved Garcia at first due to tools.  He loved Tank at first too.
    But Kenny OK'ed cutting Tank after years of failure and IMO he would have had no problem letting Garcia go either if the Sox were contending this year and needed a better RFer to do that.
     
    But even if he did defend Garcia why would he use the WORST argument possible (great tools) which can be simply answered by : if he has such great tools why he's been such a lousy player ?   Smarter to just show: 1- limited financial risk  2- we don't have anybody better and it's not worth spending money on a big FA  3- some chance his power might finally come around due to his body type and lot to gain if he finally does.
    Those sound like much more sensible and convincing arguments than the tools bullshit.
    avatar
    Deplorable Mark
    Roof Shot

    Posts : 1721
    Join date : 2016-09-16

    Re: LUCKY FLYBALLS

    Post by Deplorable Mark on Tue Jun 13, 2017 10:24 pm

    rmapasad wrote:
    Deplorable Mark wrote:It gets even better folks.  Change player A to 260/360/460 and player B to 300/330/510
    Player A drops to 72 runs created and 6.04 RC27 while Player increases to 81 and 6.49 RC27
    How can this be with no change in the OBP-SLG differiental?!?!?!?!?
    I guess hits most be more valuable than walks!!!!  A complete contradiction of the 1.7 bullshit.


    Fundamental problem. You're using the old, old Bill James RC formula which while not discredited completely hasn't been used for years.  As noted in Fan Graphs way back in 2008:
    " Tonight we’ve completed the phasing out of Runs Created (RC) and Runs Created per 27 Outs (RC/27). In their place you will find two run stats based off wOBA, which is a sounder metric than the previously used Runs Created. "

    So you're using an unsound RC/27 metric as being the measure of how well another metric works.  Not much proof here.


    runs created discredited

    LMAO!!!!!!

    Hey shit for brains, when you get to change your weights every year, of course wOBA will be more accurate.  BTW, where is your wOBA calculation?

    Or does your lying stupid ass already realize it yields the same result and you are just praying I don't bother to do the math

    BTW, how good can wOBA be?   I didnt see anything in the formula to account for all those lucky ground ball hits you like to bitch about.

    You ain't got a clue douchebag
    avatar
    Deplorable Mark
    Roof Shot

    Posts : 1721
    Join date : 2016-09-16

    Re: LUCKY FLYBALLS

    Post by Deplorable Mark on Tue Jun 13, 2017 11:15 pm

    rmapasad wrote:
    Deplorable Mark wrote:And you would have us believe the converse.  With the entire front office throwing their hands in the air proclaiming who gives a flying F, we will suck anyway.  Plus this tools stuff is so 1960's jock.
    Bottom line, you don't keep him without a belief that the move will pay off
    PS, I have the HAWK on my side

    Hawk ?  Please.  I can see Kenny gilding the lily about this story to Hawk and Hawk was more than happy to be his on-air promoter. Sure, Kenny loved Garcia at first due to tools.  He loved Tank at first too.
    But Kenny OK'ed cutting Tank after years of failure and IMO he would have had no problem letting Garcia go either if the Sox were contending this year and needed a better RFer to do that.
     
    But even if he did defend Garcia why would he use the WORST argument possible (great tools) which can be simply answered by : if he has such great tools why he's been such a lousy player ?   Smarter to just show: 1- limited financial risk  2- we don't have anybody better and it's not worth spending money on a big FA  3- some chance his power might finally come around due to his body type and lot to gain if he finally does.
    Those sound like much more sensible and convincing arguments than the tools bullshit.

    Sounds more like you are trying to convince yourself.

    Double talk all you want.  Your argument still boils down to 

    WE THINK HE SUCKS BUT ARE TOO LAZY TO LOOK FOR ANYBODY BETTER.

    Yep, that not only great logic but a wonderful attitude as well.
    avatar
    Jack Brickhouse
    Andy the Clown

    Posts : 852
    Join date : 2014-04-03

    Re: LUCKY FLYBALLS

    Post by Jack Brickhouse on Wed Jun 14, 2017 6:49 am

    WIN NOW-RELOAD!!
    avatar
    Deplorable Mark
    Roof Shot

    Posts : 1721
    Join date : 2016-09-16

    Re: LUCKY FLYBALLS

    Post by Deplorable Mark on Wed Jun 14, 2017 8:24 am

    BTW, where is your wOBA calculation?


    *************************************


    YO ROGER, I SAVED YOU THE TROUBLE


    player A 260/360/460  72RC .361wOBA
    player B 300/330/510  81RC  .365wOBA


    player B still comes out ahead, contrary to your little 1.7 theory!!!!


    As an added bonus, from fran graphs:


    Runs Created is a good estimator of how many runs a team should have scored in a given season. When applied to players, it is somewhat less accurate though still a useful estimator of a player’s actual production.



    that doesn't sound like its been discredited!!!!!!


    Once again, you mislead the board because your apparently don't fully understand sabrmetrics and your math skills are inferior to the KARK's
    avatar
    rmapasad
    Roof Shot

    Posts : 2423
    Join date : 2009-04-06
    Location : Northridge, CA

    Re: LUCKY FLYBALLS

    Post by rmapasad on Wed Jun 14, 2017 1:04 pm

    runs created discredited   LMAO!!!!!!   Hey shit for brains, when you get to change your weights every year, of course wOBA will be more accurate.  BTW, where is your wOBA calculation?
    Or does your lying stupid ass already realize it yields the same result and you are just praying I don't bother to do the math
    BTW, how good can wOBA be?   I didnt see anything in the formula to account for all those lucky ground ball hits you like to bitch about.
    You ain't got a clue douchebag  >.

    I am tired of you constantly turning stuff into ridiculous 14 year old level pissing contests instead of calmly trying to get at the truth. 



    I ran all teams from 2010-2017 - Runs, OBP, Slug, OPS, W/OBA, RC and 1.8 * OBP + Slug.


    1.  Your little calculation gave 20 pts of overall OPS more to the "better SLug" side. Of course if one side gets 20 more OPS pts. the way you jiggered it, they'll score more runs. That wasn't the way these studies were done.  The question is: 300/360/500  better than 300/330/530 ?  -

     I found four teams that had identical OPS's from 2010-2017.

     D-Backs 256/321/411 = 732 OPS    5292 runs scored

     Orioles 255/313/419   = 732 OPS    5216 runs scored


    Twins         254/319/396   = 715 OPS   5142 runs scored

    White Sox  255/315/400   = 715 OPS   5004 runs scored



    So teams at equal OPS with the better OBP and worse Slug scored more runs. To be fair, Runs Scored are "corrupted" by baserunning issues (SB. CS, GDP) so not relying heavily on this comp.




    2. Here are the correlations.  I took the STDEV's of each category to the average in that category and then cross-referenced the STDEV's to each other.


    OBP*1.8 + Slug v. w/OBA - correlation was very good .09 STDEV... Almost a perfect match. 

    RC to W/OBA - not good correlation at all at  .81 STDEV
    RC to Actual Runs Scored - Same bad correlation of .82 STDEV.   Which is wierd since RC takes baserunning stuff into account.  Park adjustments in RC account for a good share of that variance.
    then:

    W/OBA to Runs Scored - good correlation at .26 STDEV.   W/OBA uses park adjustments too.
    1.8* OBP + Slug to RS  - good correlation at .28 STDEV.


    I understand why W/OBA seems like a better metric than Runs Created.


    3. But now I have to show one other correlation: raw OPS
    Raw OPS v. W/OBA   - very good correlation at .07 STDEV
    Raw OPS v. Runs Scored - good correlation at  .26 STDEV
    Raw OPS v. Runs Created - bad correlation at . 84 STDEV



    4. BUT and this is what floored me -   good old fashioned OPS seems to work.  Not a whole helluva lot.  Perhaps this substantiates that 1 pt of OBP is no better than 1 pt of Slug  and vice-versa, but quite honestly I have neither the time or inclination to re-read all the W/OBA and OBP*1.8 stuff.   It's not that any of that stuff is INVALID.  It's not.  It just doesn't seem that it adds much to the party that OPS doesn't.
    Maybe in the higher power environment of 2002-2007 when all these new "advancements" on OPS came about OBP added more runs, but recently it doesn't seem to.
    avatar
    rmapasad
    Roof Shot

    Posts : 2423
    Join date : 2009-04-06
    Location : Northridge, CA

    Re: LUCKY FLYBALLS

    Post by rmapasad on Wed Jun 14, 2017 2:20 pm

    Deplorable Mark wrote:
    Sounds more like you are trying to convince yourself.

    Double talk all you want.  Your argument still boils down to 

    WE THINK HE SUCKS BUT ARE TOO LAZY TO LOOK FOR ANYBODY BETTER.

    Yep, that not only great logic but a wonderful attitude as well.

    Bullshit.  It doesn't boil down to the Sox were "too lazy" to look for somebody better.
    It's that somebody better was going to be TOO EXPENSIVE for a team that wasn't going into the season as a serious contender.  
    You were hollering all winter about the Catcher job. Were they TOO LAZY to find a better Catcher ?  How about CF ?   Tilson had 2 big league AB's and a so-so minor league record. Sox were willing to go with him, with May/Engel/Leury as fallbacks.

    Yet we are somehow to believe this fairy tale that even though Garcia didn't play much RF last year and hit for shit the Sox were still enthused about him taking the RF job because he's promised to work out and has this great body ? 
    Or do we believe that the Sox treated RF as they did the other HOLES they had going into 2017 like Catcher and CF ?  Namely - we're not going to spend $ on FA's, we're going to play who we've got. G
    [size=16]ive Garcia and Liriano their final chances to prove something, throw Asche into the picture also.  All low-cost guys in mid-late 20's - maybe one of them will emerge.  End of story.  


      





      
    [/size]
    avatar
    rmapasad
    Roof Shot

    Posts : 2423
    Join date : 2009-04-06
    Location : Northridge, CA

    Re: LUCKY FLYBALLS

    Post by rmapasad on Wed Jun 14, 2017 3:17 pm

    Deplorable Mark wrote:BTW, where is your wOBA calculation?
    *************************************
    YO ROGER, I SAVED YOU THE TROUBLE
    player A 260/360/460  72RC .361wOBA
    player B 300/330/510  81RC  .365wOBA
    player B still comes out ahead, contrary to your little 1.7 theory!!!!


    As an added bonus, from fran graphs:
    Runs Created is a good estimator of how many runs a team should have scored in a given season. When applied to players, it is somewhat less accurate though still a useful estimator of a player’s actual production.
    that doesn't sound like its been discredited!!!!!!
    Once again, you mislead the board because your apparently don't fully understand sabrmetrics and your math skills are inferior to the KARK's

    1. Serious question, not intended as a gotcha.  Both RC and w/OBA are linear weight formulas that use actual events (HR, BB, singles,etc /PA).  How do you substitute OBP and Slug to get a runs result ?
    Couldn't find the formula for it.  Maybe you used your own - that's fine.  Just would like an explanation.

    2. Although I see what you did in saying that 50 pts of Slug v. 30 pts of OBP is roughly a 1.7 multiple of OBP, that isn't the formula which is OBP*1.7 (or 1.Cool + Slug.   Meaning that OBP when multiplied by 1.7 becomes 56% of the value of OBP+ Slug instead of 44% of the traditional value of OBP+ Slug. 
    Then you exacerbate all this by incrementing overall OPS 20 pts. by subtracting 30 pts OBP and adding 50 pts Slug. whereby your Player B's OBP is only 38% of OBP + Slug.    Not an apples to apples comp with the actual formula.

    3.  I am puzzled by the RC issue.  I ran STDEV of RC to Runs, OPS and found disconnects.  Again, the RC formula heavily uses park factors so team like Colorado and Texas get big dings on their RC v. actual runs. 
    But w/OBA and OPS relate well to Runs Scored and to each other.

    4. I'd have to re-read the old 1.8 studies to find out why they thought the uber-premium on OBP makes it so much better than straight OPS.    I really haven't used 1.8* OBP + Slug, it's bulky.  Same with the calculation for W/OBA.   Not that they are invalid.
    OPS seems to work well on a team level, and it's easier to use.  Plus even it's used more regularly now on major league scoreboards so the ordinary fan has actually heard of it.
    avatar
    Deplorable Mark
    Roof Shot

    Posts : 1721
    Join date : 2016-09-16

    Re: LUCKY FLYBALLS

    Post by Deplorable Mark on Wed Jun 14, 2017 9:54 pm

    Serious answer to #1

    Your basic equation (which btw isn't linear or weighted) is

    Runs = TB * (H + BB)/(AB + BB)

    Thats roughly TB times OBP

    TB can be rephrased as AB * SLG

    Therefore Runs can be quickly estimated as AB * SLG * OBP

    BTW, James went with should a basic formula because that is all the info he had at the time.  I don't believe The Sporting News even bothered with OBP in the 70's.  Also, the beauty of the basic formula is that it never changes.  It is the perfect mathematical way to sum up the relationship between batting and scoring runs.  It is the e = mc2 of sabermetrics.  So it really pissed me off that you claimed this thing of brilliance and beauty was discredited.
    avatar
    Deplorable Mark
    Roof Shot

    Posts : 1721
    Join date : 2016-09-16

    Re: LUCKY FLYBALLS

    Post by Deplorable Mark on Wed Jun 14, 2017 10:06 pm

    Serious answer to #2

    Pete Palmer in his classic book THE HIDDEN GAME OF BASEBALL show that multiplying OBP and SLG is a better estimate of runs than merely added them together

    However OPS is so much easier to calculate when you do all this on pencil and paper like one had to prior to the home computer

    Today, not only do you have the means to easily multiply them, but you are also adding operations like park effect and league context which nobody can do quickly in their head.  So why continue adding the 2 when multiply is more accurate.
    avatar
    Deplorable Mark
    Roof Shot

    Posts : 1721
    Join date : 2016-09-16

    Re: LUCKY FLYBALLS

    Post by Deplorable Mark on Wed Jun 14, 2017 10:20 pm

    Serious answer to #4

    1.8 is nonsense.  It derived from the coefficients of OBP and SLG being in the same regression.  But OBP and SLG partial measure the same thing.  Why do you think wOBA breaks it out into the individual components?  It's to avoid double counting.  1.8 is skewed by multicolinearity.  I can't believe anybody takes it serious.  Judging by the discussion section on the Phil Bandburn blog, a lot don't.

    Now that doesn't mean OBP isn't more important than SLG, because it is.  The reason it is more important is that getting on base means you didn't make an out.  In reality, the point of hitting isnt hitting, it is out avoidance.
    avatar
    rmapasad
    Roof Shot

    Posts : 2423
    Join date : 2009-04-06
    Location : Northridge, CA

    Re: LUCKY FLYBALLS

    Post by rmapasad on Thu Jun 15, 2017 2:08 pm

    Deplorable Mark wrote:Serious answer to #4

    1.8 is nonsense.  It derived from the coefficients of OBP and SLG being in the same regression.  But OBP and SLG partial measure the same thing.  Why do you think wOBA breaks it out into the individual components?  It's to avoid double counting.  1.8 is skewed by multicolinearity.  I can't believe anybody takes it serious.  Judging by the discussion section on the Phil Bandburn blog, a lot don't.

    Now that doesn't mean OBP isn't more important than SLG, because it is.  The reason it is more important is that getting on base means you didn't make an out.  In reality, the point of hitting isnt hitting, it is out avoidance.

    Not being argumentative but IIRC the way these studies tested run differentials were:
    Team A   300/360/500    v.   Team B    300/330/530.
    Since BA remained static there was no OBP/SLG interrelationship as Team A's OBP was higher due to BB's (doesn't affect SLG) and B's SLG was higher due to more hits being for extrabases (doesn't affect OBP).

    This may be why Birnbaum answered "Anonymous" by basically saying multicolinearity wasn't an issue. 
    Once more, it seems that guys who've been using regression models would be smart enough to avoid an obvious and fundamental flaw like that.

    To try to test as much of this for myself as possible,  I tested teams'  2012-2016 numbers using
    1) 1.8 * OBP+ Slug   2) OBP*SLG    3) straight OPS

    Assuming that w/OBA is now the new gold standard here's how each correlated to w/OBA.  
    - 1.8 *OBP + Slug  had a .36 STDEV to w/OBA   and   OBP*Slug had a .37  - virtually no difference.
    - 1.8 * OBP had an even stronger relationship OBP * Slug = .22 STD DEV
    - OPS had a weaker relationship to W/OBA

    However, I changed the formula to 1.25* OBP+ Slug and found that tightened things up even more:
    - 1.25 *OBP + Slug = .33 STD DEV to w/OBA
    - 1.25 *OBP + Slug = .14 STD DEV to OBP*Slug

    So in effect 1.8*OBP and particularly 1.25* OBP seem to have virtually the same result as OBP*Slug.
    The correlation got worse in fact when I used 2* OBP + Slug
    I don't pretend to have the sophistication of the stat guys who ran their original models, and I clearly may be missing something.  But to me all this says is that the 1.8 OBP stuff, while it wasn't invalid, was a tempest in a teapot.  It doesn't seem to add much to the party.  Maybe that's why it's passe now.
    avatar
    Deplorable Mark
    Roof Shot

    Posts : 1721
    Join date : 2016-09-16

    Re: LUCKY FLYBALLS

    Post by Deplorable Mark on Thu Jun 15, 2017 3:34 pm

    rmapasad wrote:
    Deplorable Mark wrote:Serious answer to #4

    1.8 is nonsense.  It derived from the coefficients of OBP and SLG being in the same regression.  But OBP and SLG partial measure the same thing.  Why do you think wOBA breaks it out into the individual components?  It's to avoid double counting.  1.8 is skewed by multicolinearity.  I can't believe anybody takes it serious.  Judging by the discussion section on the Phil Bandburn blog, a lot don't.

    Now that doesn't mean OBP isn't more important than SLG, because it is.  The reason it is more important is that getting on base means you didn't make an out.  In reality, the point of hitting isnt hitting, it is out avoidance.

    Not being argumentative but IIRC the way these studies tested run differentials were:
    Team A   300/360/500    v.   Team B    300/330/530.
    Since BA remained static there was no OBP/SLG interrelationship as Team A's OBP was higher due to BB's (doesn't affect SLG) and B's SLG was higher due to more hits being for extrabases (doesn't affect OBP).

    This may be why Birnbaum answered "Anonymous" by basically saying multicolinearity wasn't an issue. 
    Once more, it seems that guys who've been using regression models would be smart enough to avoid an obvious and fundamental flaw like that.

    To try to test as much of this for myself as possible,  I tested teams'  2012-2016 numbers using
    1) 1.8 * OBP+ Slug   2) OBP*SLG    3) straight OPS

    Assuming that w/OBA is now the new gold standard here's how each correlated to w/OBA.  
    - 1.8 *OBP + Slug  had a .36 STDEV to w/OBA   and   OBP*Slug had a .37  - virtually no difference.
    - 1.8 * OBP had an even stronger relationship OBP * Slug = .22 STD DEV
    - OPS had a weaker relationship to W/OBA

    However, I changed the formula to 1.25* OBP+ Slug and found that tightened things up even more:
    - 1.25 *OBP + Slug = .33 STD DEV to w/OBA
    - 1.25 *OBP + Slug = .14 STD DEV to OBP*Slug

    So in effect 1.8*OBP and particularly 1.25* OBP seem to have virtually the same result as OBP*Slug.
    The correlation got worse in fact when I used 2* OBP + Slug
    I don't pretend to have the sophistication of the stat guys who ran their original models, and I clearly may be missing something.  But to me all this says is that the 1.8 OBP stuff, while it wasn't invalid, was a tempest in a teapot.  It doesn't seem to add much to the party.  Maybe that's why it's passe now.


    Not to be argumentive, but the things I looked at that appear to be some kind of regression DO avoid the multicolinearity problem by NOT using OBP and SLG, but instead using the individual components.

    I agree with "Anonymous".  1.8 is useless regardless of how accurate.  As a GM, I cannot substitute a X number of points of OBP for Y number of points of SLG.

    Sponsored content

    Re: LUCKY FLYBALLS

    Post by Sponsored content


      Current date/time is Sun Sep 24, 2017 9:08 am