Positional WAR Adjustment - again because it's been awhile

Non-league talk in here. Please make NSFW *links* and not pics.
Ted
Ex-GM
Posts: 5630
Joined: Thu Jun 04, 2015 2:50 pm
Has thanked: 368 times
Been thanked: 378 times

Positional WAR Adjustment - again because it's been awhile

Post by Ted » Fri May 04, 2018 6:48 pm

I'm stuck in a bit of a logic loop here. Someone help me break it.

For any of you newer GM's, I go on a rant about how the positional adjustments to WAR seem too extreme every couple seasons or so. To recap, here they are in terms of runs (and 10 runs roughly = 1 WAR)
Full Season Adjustment
C +12.5
1B -12.5
2B +2.5
SS +7.5
3B +2.5
LF -7.5
CF +2.5
RF -7.5
DH -17.5

Let's throw out catcher, because catcher WAR excludes so many parts of actual catcher value that it's tough.

So a league average hitting SS is worth 2.5 WAR more than a league average hitting DH. (Note, that's compared to all hitters, no position specific hitters). I've noted in the past that this seems to be an antiquated model as there are better hitting players at traditional weaker hitting positions than there were when it was created. I'm not saying there shouldn't be a positional adjustment, just that it's too big. There's actually been some numbers crunched by people way better at thus than me that shows this could be the case.

The other argument I'd make is that if you look at the WAR leaders in any league, 1B and DH seem to be chronically underrepresented. If these adjustments worked, we should have equal representation of various positions in the top 50 players by WAR. Go scan the fangraphs leaderboards, nary a 1B or DH in sight. It's just packed middle infielders and centerfielders. Now, maybe there shouldn't be an equal distribution, maybe better athletes are better hitters and better athletes play skilled positions more often or something like that. I don't know that I believe that.

So, with that setup, I've come to where my head is stuck. If we're valuing more skilled positions because of scarcity of good hitters at those positions, and now 1B and DH aren't worth any WAR because they all hit well (btw, doesn't seem to be the case anymore), then isn't an incremental increase in the hitting ability of 1B and DH actually worth more because you're better somewhere that no one else is taking advantage of, but then that would mean that good getting 1B and DH are scarce, just in a different way with a higher bar, and so on and so forth and blah blah blah

And so I think the problem is this. A flat adjustment to a counting stat that is based on an average performance at a position that is constantly changing seems horrendously flawed. Especially as front offices have become more analytical and have obviously prioritized getting more value out of traditionally weak hitting positions over the last decade and a half. Anyway, does this make sense to anyone else? Feel free to criticize my thinking. I just really don't like these, as many of you are aware, and sick of. They just don't feel right.
Ted Schmidt
Twin Cities Typing Nightmares(2044-present)
California Crusaders (2021-2038)
Image

Ted
Ex-GM
Posts: 5630
Joined: Thu Jun 04, 2015 2:50 pm
Has thanked: 368 times
Been thanked: 378 times

Re: Positional WAR Adjustment - again because it's been awhi

Post by Ted » Fri May 04, 2018 7:02 pm

As I refine this thought a bit more, the flat adjustment doesn't really seem to take into account the availability of players that are better than replacement level at their various positions. It just defines a value relative to a league average hitter at that position. This may actually be a critique of WAR in general.

Let's take say Matt Adams, who is essentially a just above league average hitter and therefore just above replacement level at 1B. Compare him to Joey Votto, who I can't use enough superlatives about. The WAR model just implies increasing value or runs and wins created for each incremental increase in hitting talent (I'm ignoring defensive proficiency for ease of discussion). But players don't exist in an even distribution of hitting talent. OR at least, they don't really appear to. The guys at the top are just SO much better than the average guys. (Which is insane because BAD MLB players are just so, so good compared to the rest of us).

With this in mind, the entire concept of WAR may be fatally flawed as it doesn't account for talent scarcity? Furthermore, if this is the case, then the positional adjustments make little to no sense as you can't just flatly apply a decrease of value to positions with "better hitters" when the reality is that the players at these positions don't exist in linear distributions.
Ted Schmidt
Twin Cities Typing Nightmares(2044-present)
California Crusaders (2021-2038)
Image

User avatar
jiminyhopkins
BBA GM
Posts: 3487
Joined: Mon Jan 31, 2011 2:33 pm
Location: OH
Has thanked: 296 times
Been thanked: 911 times

Re: Positional WAR Adjustment - again because it's been awhi

Post by jiminyhopkins » Fri May 04, 2018 7:07 pm

This may just be a layman's take on the situation, but why can't we just separate offensive and defensive war? IIRC baseball reference does this. Just wondering. Maybe ootp won't let us make that distinction.
GM, 2051 and 2053 JL 4TH WILDCARD Phoenix Talons (2029-??), BBA
CARETAKER GM, 2053 GBC CHAMPION Tokyo Pearls (2053 - 2056)
GM, THE GREATEST MINOR LEAGUE TEAM OF ALL TIME Toledo Liberty
Vic Caleca Team News Award Winner: 2051, 2054, 2057

Ted
Ex-GM
Posts: 5630
Joined: Thu Jun 04, 2015 2:50 pm
Has thanked: 368 times
Been thanked: 378 times

Re: Positional WAR Adjustment - again because it's been awhi

Post by Ted » Fri May 04, 2018 7:13 pm

jiminyhopkins wrote:This may just be a layman's take on the situation, but why can't we just separate offensive and defensive war? IIRC baseball reference does this. Just wondering. Maybe ootp won't let us make that distinction.
Nope, I like this. I don't really use WAR anymore. I use some offensive thing like wRC+ and some other defensive metric. Then it is up to me to maintain a sense of how scare a certain value is relative to the rest of the league. If I wanted to be more methodical, I could spreadsheet it. I just like to overthink things.
Ted Schmidt
Twin Cities Typing Nightmares(2044-present)
California Crusaders (2021-2038)
Image

User avatar
RonCo
GB: JL Frontier Division Director
Posts: 19815
Joined: Sat Nov 14, 2015 10:48 pm
Has thanked: 1982 times
Been thanked: 2902 times

Re: Positional WAR Adjustment - again because it's been awhi

Post by RonCo » Fri May 04, 2018 7:29 pm

Is helpful to think about why WAR came about in the first place.
GM: Bikini Krill
Nothing Matters But the Pacific Pennant
Roster

User avatar
jiminyhopkins
BBA GM
Posts: 3487
Joined: Mon Jan 31, 2011 2:33 pm
Location: OH
Has thanked: 296 times
Been thanked: 911 times

Re: Positional WAR Adjustment - again because it's been awhi

Post by jiminyhopkins » Fri May 04, 2018 7:33 pm

I like wins and RBI.

But seriously, it does seem like everyone (and I mean the whole baseball universe) is war happy. It's why Mario Deortez and Bartolo Ortiz will never win an MVP award. FOr most people its war and thats it.

In the GBC (ancient history I know), we seemed to evaluate players just fine without war. Oh well. Maybe we can just turn it off in ootp and force everyone to truly scout these guys the old fashioned way, lol. :mrgreen:
Last edited by jiminyhopkins on Fri May 04, 2018 8:10 pm, edited 2 times in total.
GM, 2051 and 2053 JL 4TH WILDCARD Phoenix Talons (2029-??), BBA
CARETAKER GM, 2053 GBC CHAMPION Tokyo Pearls (2053 - 2056)
GM, THE GREATEST MINOR LEAGUE TEAM OF ALL TIME Toledo Liberty
Vic Caleca Team News Award Winner: 2051, 2054, 2057

agrudez
Ex-GM
Posts: 7681
Joined: Tue Nov 20, 2012 10:30 am
Has thanked: 21 times
Been thanked: 47 times

Re: Positional WAR Adjustment - again because it's been awhi

Post by agrudez » Fri May 04, 2018 7:39 pm

WAR is really useful as a quick, succinct way of describing a player (particularly for articles - ie. "player X is playing great with a 4 WAR this season" vs. "player X is playing great with [insert 8 different stats here]) and a useful tool in comparing players across the same position (ie. This 3B is consistently +1 WAR over my current 3B so that is probably a projectable upgrade worth pursuing) once context is taken into account (IP, FIP, BABIP, oBABIP, individual iso rates, outlier ZR, park factors, etc.). I don't believe in using it to compare any batter vs. any other batter (for the reasons you ascribe in the OP - a DH is intrinsically worth less than a 2B due to scarcity and positional viability, but is a .700 OPS 2B with mediocre defense more *useful* than an .850 OPS DH - even if the former is harder to find than the latter?).
League Director: Kyle “agrudez” Stever*
*Also serves as chief muckraker
-Ron, 2025 media guide

Image

Ted
Ex-GM
Posts: 5630
Joined: Thu Jun 04, 2015 2:50 pm
Has thanked: 368 times
Been thanked: 378 times

Re: Positional WAR Adjustment - again because it's been awhi

Post by Ted » Fri May 04, 2018 8:03 pm

Agree with you Marko about everyone being WAR happy. It's almost to the point of pitcher wins and RBI. Given that it's significantly more complicated than they are, it's perhaps even more prone to misuse. And I think that's why it annoys me so much. I get the reason why it was made, Ron. I think. But I don't think that's how it's being used. I also think that it's probably not sophisticated enough even for its original purpose (which I assumed to be a way of getting a context neutral measure of a player's value). What it comes down to me is that the idea of assigning linear weights to the value of various outcomes then figuring run scoring probability and translation to wins is perhaps very wrong minded in its origin.

Try to win games with a bunch of 2011 Brendan Ryans (248/.313/.326, 4.2 bWAR). It won't work. You won't be able to score at all. I know I'm cherry picking a player here, but I'm not even sure that WAR is a good representation of his individual value. Removing sequencing removes the damage that an absolute black hole does to you lineup. Considering that you have to have multiple unlikely outcomes in a row to score a run, any metric that doesn't account for the change in likelihood of those multiple outcomes occurring in subsequent order is very suspect. I don't know how you would account for it, but as "complicated" as WAR is, it's probably frighteningly over-simplistic as an actual measurement of player value. The more an more I look into it, it seems more like a mathematical model that has tried to capture a variety of values and succeeds in resembling the whole, but more due to a formula designed to do just that than one that accurately represents the fine details making that whole.

As Kyle stated, it's okay for a quick reference talking point, but to really try to tie it to a single season of player value is probably a gross misuse.
Ted Schmidt
Twin Cities Typing Nightmares(2044-present)
California Crusaders (2021-2038)
Image

agrudez
Ex-GM
Posts: 7681
Joined: Tue Nov 20, 2012 10:30 am
Has thanked: 21 times
Been thanked: 47 times

Re: Positional WAR Adjustment - again because it's been awhi

Post by agrudez » Fri May 04, 2018 8:35 pm

Well, first off, if you had 9 Brendan Ryan clones with that same batting split across your diamond, even if their defensive performance at each position was constant (ie. in OOTP's world they all had the same ZR for their respective position), their WAR at the various positions would be vastly lower. A +10 ZR and .63 OPS at SS is valuable, but that same batting slash in LF is not - and this is adequately accounted for in WAR. In that scenario, I suspect you'd get good WAR at SS and CF, decent at 2B and C (nowhere close to the 4.2 cited, but still positive), bad at 3B, LF and RF (probably approaching 0) and dreadful at 1B, DH and LF (likely pretty far in the negative). Honestly, I wouldn't be surprised if the aggregate WAR of that lineup was approaching 0 (1B, DH and LF cancelling out CF, SS, 2B and C) which would make them a 30-40 win pace lineup on their own merits.
League Director: Kyle “agrudez” Stever*
*Also serves as chief muckraker
-Ron, 2025 media guide

Image

Ted
Ex-GM
Posts: 5630
Joined: Thu Jun 04, 2015 2:50 pm
Has thanked: 368 times
Been thanked: 378 times

Re: Positional WAR Adjustment - again because it's been awhi

Post by Ted » Fri May 04, 2018 8:44 pm

agrudez wrote:Well, first off, if you had 9 Brendan Ryan clones with that same batting split across your diamond, even if their defensive performance at each position was constant (ie. in OOTP's world they all had the same ZR for their respective position), their WAR at the various positions would be vastly lower. A +10 ZR and .63 OPS at SS is valuable, but that same batting slash in LF is not - and this is adequately accounted for in WAR. In that scenario, I suspect you'd get good WAR at SS and CF, decent at 2B and C (nowhere close to the 4.2 cited, but still positive), bad at 3B, LF and RF (probably approaching 0) and dreadful at 1B, DH and LF (likely pretty far in the negative). Honestly, I wouldn't be surprised if the aggregate WAR of that lineup was approaching 0 (1B, DH and LF cancelling out CF, SS, 2B and C) which would make them a 30-40 win pace lineup on their own merits.
I meant more like a version that had that portion of their WAR from defense at each position. The 1B would have a better batting line, but would still be a subpar hitter for 1B, and an excellent batter. That team still wouldn't score any runs. You just can't have that many subpar bats and win any games.
Ted Schmidt
Twin Cities Typing Nightmares(2044-present)
California Crusaders (2021-2038)
Image

agrudez
Ex-GM
Posts: 7681
Joined: Tue Nov 20, 2012 10:30 am
Has thanked: 21 times
Been thanked: 47 times

Re: Positional WAR Adjustment - again because it's been awhi

Post by agrudez » Fri May 04, 2018 9:05 pm

Well, 1B defense doesn't really do much for WAR even when it's superb, so a 4.2 WAR 1B is probably crushing it offensively in this hypothetical. Same for DH (obviously) and the corner OFs, at a minimum.
League Director: Kyle “agrudez” Stever*
*Also serves as chief muckraker
-Ron, 2025 media guide

Image

User avatar
RonCo
GB: JL Frontier Division Director
Posts: 19815
Joined: Sat Nov 14, 2015 10:48 pm
Has thanked: 1982 times
Been thanked: 2902 times

Re: Positional WAR Adjustment - again because it's been awhi

Post by RonCo » Fri May 04, 2018 9:11 pm

WAR's best use is probably to compare players at the same position across eras. Which is really what it was designed for.

Of course, as Ted said, it's often used for other things...and it works fine for those things as loose estimates. The position adjustments are attempts to generalize value across positions, but yes, it's flawed. The creators of WAR were pretty clear than you shouldn't use it as a precision measure. But yeah, I think it's fair to say that when player A posts 1.5 WAR more than player B, player A is probably more valuable. Shrug.

It's also probably true that every MLB team has their own statisticians fiddling with data in their own way to adjust the impact of things--measurement of defense in particular is changing rapidly. So the publicly available WAR is probably not what actual baseball teams are using.

And let's not even get into whether you trust what OOTP is doing to calculate it, which I do not. I don't trust either ZR or WAR to be direct. I'm certain they are useful measures, but I am not so willing to say they are what the baseball world knows as ZR and WAR. But they are what we have, so we use them as shorthand lingo. It also means that we each (as BBA GMs) can use whatever we want and try to create a competitive advantage through our own gut reactions.

All that said, WAR is like any other composite stat (As runs created and even the linear weighted stats,. which I tend to like, too), useful in its own way, but easy to stretch too far.
GM: Bikini Krill
Nothing Matters But the Pacific Pennant
Roster

User avatar
RonCo
GB: JL Frontier Division Director
Posts: 19815
Joined: Sat Nov 14, 2015 10:48 pm
Has thanked: 1982 times
Been thanked: 2902 times

Re: Positional WAR Adjustment - again because it's been awhi

Post by RonCo » Fri May 04, 2018 9:25 pm

Here's a simple plot of all BBA players WAR vs. their wRC+ (PA = 150+)
WARvsWRC-plus.PNG
WARvsWRC-plus.PNG (23.3 KiB) Viewed 983 times
GM: Bikini Krill
Nothing Matters But the Pacific Pennant
Roster

User avatar
RonCo
GB: JL Frontier Division Director
Posts: 19815
Joined: Sat Nov 14, 2015 10:48 pm
Has thanked: 1982 times
Been thanked: 2902 times

Re: Positional WAR Adjustment - again because it's been awhi

Post by RonCo » Fri May 04, 2018 9:28 pm

In a sense, a sabermetrics guy saying a "6" WAR is better than a "5" WAR guy is about like a what a scout says when they say a "65" guy is better value than a "60" guy ... whatever.
GM: Bikini Krill
Nothing Matters But the Pacific Pennant
Roster

User avatar
RonCo
GB: JL Frontier Division Director
Posts: 19815
Joined: Sat Nov 14, 2015 10:48 pm
Has thanked: 1982 times
Been thanked: 2902 times

Re: Positional WAR Adjustment - again because it's been awhi

Post by RonCo » Fri May 04, 2018 9:37 pm

Ted wrote: And so I think the problem is this. A flat adjustment to a counting stat that is based on an average performance at a position that is constantly changing seems horrendously flawed.
It is flawed, though "horrendously" is a judgement call. It's also flawed (as you noted) that as offense rises across skill positions the adjustments may well need to be tweaked. I real a very old article by Bill James himself (hallowed be thy name), in one of his very early Baseball Annuals (back when he was doing those beautiful books), where he suggested that WAR and that kind of stat--since they are essentially predicated on the idea of player's base skill sets NOT changing) could actually be used to understand the rate at which players improve from era to era...because obviously, they DO change as training and nutrition and technology helps them develop. This meant that as you saw data in WAR make less sense it was an indicator of how fast that development was improving players...though that's damned hard to "prove" for anyone who isn't employed full time to figure it out. :)

So, yeah, it's flawed. But the concept is useful, IMHO.
GM: Bikini Krill
Nothing Matters But the Pacific Pennant
Roster

Post Reply Previous topicNext topic

Return to “Off Topic Chatter”

Who is online

Users browsing this forum: No registered users and 4 guests