Like us on Facebook


Follow us on Twitter





Page 1 of 2 12 LastLast
Results 1 to 15 of 25

Hybrid View

  1. #1
    Join Date
    Feb 2009
    Location
    Connecticut
    Posts
    1,915
    vCash
    1500

    Positional Adjustments Wrong?

    When looking at the statistic of WAR on Fangraphs, it is clear that the positional adjustments they make are pretty arbitrary.

    Since the offensive and defensive inputs into WAR don’t account for position played in any way, adding together wRAA and UZR without any sort of adjustment will severely underrate players at difficult defensive positions (think short stop, center field, catcher). It’s not that difficult to be a +5 defender at a position like first base, but it takes considerable talent to be a +5 defensive center fielder.

    Here are the full FanGraphs positional adjustments used in WAR:

    Catcher: +12.5 runs (all are per 162 defensive games)
    First Base: -12.5 runs
    Second Base: +2.5 runs
    Third Base: +2.5 runs
    Shortstop: +7.5 runs
    Left Field: -7.5 runs
    Center Field: +2.5 runs
    Right Field: -7.5 runs
    Designated Hitter: -17.5 runs
    Since most players do not play all 162 games, it commonly occurs that they gain around 90-95% of these runs. For example, Pedroia would gain around .23 WAR if he played 150 games in a season at 2B, from the positional adjustment only.

    Based on this, a 2nd baseman having the same season as a 1st baseman (say, 3.0 WAR) will be 2.77 WAR to 1.84 WAR. Almost a full win (.93) difference.

    Based on offensive values only, their positional adjustments are completely off. Take a look at this quick spreadsheet I made:



    5 year Q - the number of qualified players at that position.
    To reach average - The average was right around .330.

    Based on OFFENSIVE only, it is clear that these adjustments are off. Third basemen do not deserve the same adjustment that second basemen receive, for one. In a purely offensive world, it would look something like this:

    SS +20 runs
    2B +12 runs
    C +9 runs
    CF +5 runs
    3B +3 runs
    LF -4 runs
    RF -10 runs
    DH -16 runs
    1B -17 runs

    I just multiplied by 10 to make the numbers look easier. What's important is the relationship between these numbers (proportionally), not the actual run amounts.

    Obviously my results are very primitive and don't take many things into account. But the point I am making is clear. I don't understand how Fangraphs can just decide that one position is harder than another. Is this based on any fact? The amount of leaders in WAR from each position is pretty spread out, so my guess is no. Edit: what I mean by this is that UZR does not really favor a certain position, which would make that position "easier". For example, if 10 first basemen occupied the top 20 defenders, it would make sense to deduct them runs. However, this is certainly not the case, as only 2 are in the top 25 (Gonzalez and Teixeira).

    So basically there could be three reasons that positional adjustments exist.

    1) The simple fact that a replacement 1B is a better hitter than a replacement 2B. However, if this is the case, then it should be based on actual results, not just an arbitrary guess. It is obvious that a replacement 3B is not the same as a replacement 2B, yet they are given the same positional adjustment (+2.5).

    2) To counteract biases in UZR. If corner infielders/outfielders all benefitted from playing an "easy" position (despite the fact that this is purely opinion). Like I mentioned before, this is simply not the case, so this cannot be a logical reason.

    3. A purely opinionated judgment of how hard a certain position is. For example, saying that a CF with a positive UZR is worth x amount of runs more than a LF with the same UZR. Which I doubt is what they intended, because it would be pretty illogical for many reasons, and would not be based on data.

    What do you guys think?
    Last edited by Kenny Krew; 11-08-2012 at 09:29 PM.


  2. #2
    Join Date
    Nov 2009
    Location
    Queens, NY
    Posts
    53,098
    vCash
    1500
    Interesting. I'll wait for the people more knowledgable on sabermetrics to comment on this.

  3. #3
    Join Date
    Nov 2007
    Posts
    39,545
    vCash
    1500
    You are right in the 3B value, it's not equal of 2B. It is similar to CF, but 2B is above both and closer to SS. However C is the same as SS or a bit higher IMO. Also RF is tougher then LF, not easier. Most systems make them even however.

    I think it might be easier to see the positional value at work on BR or Prospectus than FG, but I haven't looked for a long time.


    6/27/09: “We expect [Rondo] to play by the rules and be a leader as a point guard. We need him to be more of a leader,” Ainge said. “There were just a couple situations where he was late this year, I don’t know if he was sitting in his car, but showed up late and the rest of the team was there. We have team rules and you have to be on time. He was fined for being late, he said he was stuck in traffic, and it’s just unacceptable.”

    Some jerks never learn.....

  4. #4
    Join Date
    Feb 2009
    Location
    Connecticut
    Posts
    1,915
    vCash
    1500
    Quote Originally Posted by bagwell368 View Post
    You are right in the 3B value, it's not equal of 2B. It is similar to CF, but 2B is above both and closer to SS. However C is the same as SS or a bit higher IMO. Also RF is tougher then LF, not easier. Most systems make them even however.

    I think it might be easier to see the positional value at work on BR or Prospectus than FG, but I haven't looked for a long time.
    What are you basing this on?


  5. #5
    Join Date
    Nov 2006
    Location
    Riverton,il
    Posts
    15,565
    vCash
    1500
    Quote Originally Posted by Kenny Krew View Post
    What are you basing this on?
    Its hard say its a fact those are harder, but by watching them I think its pretty easy to see. 3B and C seem to be harder on the body then other positions. How often do 1B have to go after a ball like a ss or 2B? They also cover more ground. 3B isnt called the hot corner for no reason, very little time to react to line drives and such. I would say cf also has more ground to cover then and have more plays to make

    Go to fangraphs, leaders, fielding and then sort by plays. Out of the top 60 only one position outside of ss, cf, 2b and rf appears. Alex gordon at lf, and he was 52nd.

  6. #6
    Join Date
    Feb 2009
    Location
    Connecticut
    Posts
    1,915
    vCash
    1500
    Quote Originally Posted by bosox3431 View Post
    Its hard say its a fact those are harder, but by watching them I think its pretty easy to see. 3B and C seem to be harder on the body then other positions. How often do 1B have to go after a ball like a ss or 2B? They also cover more ground. 3B isnt called the hot corner for no reason, very little time to react to line drives and such. I would say cf also has more ground to cover then and have more plays to make

    Go to fangraphs, leaders, fielding and then sort by plays. Out of the top 60 only one position outside of ss, cf, 2b and rf appears. Alex gordon at lf, and he was 52nd.
    I understand that some positions are harder than others. That is obvious, yet there is no way to compare the difficulty of catcher to CF and quantity it. That is literally impossible for a multitude of reasons. What I am saying is that why make it up? Since when is sabermetrics based on opinion?


  7. #7
    Join Date
    Nov 2007
    Posts
    39,545
    vCash
    1500
    Quote Originally Posted by Kenny Krew View Post
    I understand that some positions are harder than others. That is obvious, yet there is no way to compare the difficulty of catcher to CF and quantity it. That is literally impossible for a multitude of reasons. What I am saying is that why make it up? Since when is sabermetrics based on opinion?
    Difficulty?

    There is the frequency of plays made by position.

    There are the body types/athletes that qualify for a given position.

    There are the numbers of games played - say top 50 all time at each position. 3B has less then 2B, SS, 3B. Harder to play? No. Harder on body? Yes.

    Catching is tough subjectively. Lots of guys don't want to play it. Coaches don't want potential pitchers to play it. It's usually thickly built guys 5' 9" to 6' 3". RHT only that have a straight fastball but no other pitchers. Usually fairly slow foot speed.

    The wear and tear are bar none the worse. There are ways to quantify it, but IMO the complete set of understanding of what makes catching what it is MUST contain some subjectives. But that's in an overall opinion. On the field you can compare them to each other.

    BTW FG is absolutely lame in its numbers on C and 1B. BR is better, and it still leaves a lot to be desired. Pitching and hitting are both much easier to quantify.
    Last edited by bagwell368; 11-11-2012 at 08:33 AM.


    6/27/09: “We expect [Rondo] to play by the rules and be a leader as a point guard. We need him to be more of a leader,” Ainge said. “There were just a couple situations where he was late this year, I don’t know if he was sitting in his car, but showed up late and the rest of the team was there. We have team rules and you have to be on time. He was fined for being late, he said he was stuck in traffic, and it’s just unacceptable.”

    Some jerks never learn.....

  8. #8
    Join Date
    Nov 2007
    Posts
    39,545
    vCash
    1500
    Quote Originally Posted by Kenny Krew View Post
    What are you basing this on?
    Metrics I've seen back about 3-6 years ago with basically the batting and fielding history since 1920 behind them.


    6/27/09: “We expect [Rondo] to play by the rules and be a leader as a point guard. We need him to be more of a leader,” Ainge said. “There were just a couple situations where he was late this year, I don’t know if he was sitting in his car, but showed up late and the rest of the team was there. We have team rules and you have to be on time. He was fined for being late, he said he was stuck in traffic, and it’s just unacceptable.”

    Some jerks never learn.....

  9. #9
    Join Date
    Nov 2006
    Posts
    7,039
    vCash
    1500
    I've taken issue with the positional adjustments for a while.

    You've clearly got a few positions that are harder to play at a high level than others. It doesn't take a lot to be an average first baseman. Since we're talking about baseball players, it's not a stretch to say that almost any player has the hand-eye coordination to do the basics of the job well enough; catching balls thrown by other fielders and holding their foot on a base is what that amounts to. Lefties have an advantage on throws from 1B, but a lot of guys have shown (Derrek Lee, Albert Pujols, etc.) that if you're athletic enough, you can make even right handedness not be a disadvantage.

    Catcher is the hardest position on the field, just because they have so much extra work to do on every single pitch. Calling a pitch, catching it, taking the beating they do day in and day out makes it very difficult.

    Shortstop is next, but is the difference between a replacement level shortstop and a replacement level 3B and 2B so much that they should be given a whole lot more credit for it? I don't think so, personally.

    Left field is considered easier than any other position than 1B, but I think that largely has to do with the predominance of RH hitters in baseball which makes the way the ball spins off the bat a little easier to handle in LF than in RF (balls hit to RF off of a RHH tend to tail more than balls pulled to LF).
    Replacement level outfielders probably just play the corners, but a guy who can play CF will also play the other two at times, so that's another group that is hard to distinguish between.

    I'm with Tom Tango in thinking that if you're talking about a replacement level player (WAR is Wins Above Replacement, after all) you're looking at someone who is on the bubble of making a roster. Catchers usually just catch. First basemen usually just play first. Center fielders can play any of the three positions and Infielders on the bubble are usually utility men who play both 3B and 2B, and sometimes SS.

    So, I think you have to put it down to 4-5 pools rather than delineating between each position. C, 1B, IF, OF, DH. You can get a little more creative with it and do something like C, 1B/LF, 2B/3B, SS/CF, DH too, since that probably equates more with reality than anything.

    Part of the problem with this is that positional adjustments are designed to counteract selection bias in UZR. Essentially, the way UZR works is players are judged based on the performance of other players at that same position. It's baseline comes up to be runs above average rather than replacement. So, you have guys who are good defenders in LF, and their UZR scores are off the charts because their LF peers tend to be pretty bad, in comparison. Basically, you're the tallest guy in the room until someone taller walks in. On the other side, a good defensive player at SS doesn't get rewarded with as much value for their performance because the talent pool of the league at that position is generally so good. That doesn't tell us much about the difficulty of the position or what the replacement level would/should be for the position defensively.

    The guy who created UZR, Mitchell Lichtman, has stated that UZR wasn't intended to be used in WAR the way it is, and it certainly shouldn't be listed with a decimal point. Saying someone was worth 5.6 runs with the glove in a given year is pointless, because the +/- on that 5.6 is probably 5 runs or more, in and of itself. The margin for error with UZR, especially in one year samples is extremely high.


    Another way to look at it is to say players who play the positions they play are being put there to maximize the team's leverage on balls in play. You want your best defenders in positions that are more likely to get the ball (C, SS, CF). You mitigate the risk of poor defense by putting guys who are slow, uncoordinated, or just bad, at positions like LF, 1B, DH so that they don't hurt you as much.

    Positional adjustments are there to make sure that the averages of those positions comes out roughly equal for each position because teams don't just throw their DHs out at SS some days for the hell of it.

    To bring a long post to a close, I personally don't see much value in delineating between each position, individually. I think there are player pools that one could use that would be either as effective, or more effective in achieving the goal of WAR. There is a lot of math behind figuring out what those numbers should be, and I'm nowhere near qualified to undertake something like that. And again, a huge issue I have with the defensive side of WAR, anyway, is that it's set against average production/performance/whatever rather than replacement level. People tend to forget that and make claims like "Carl Crawford was worth 2 wins with his glove, alone in 2010!" The key is to figure out what the replacement level player would do in LF and then adjust according to that. Then, positional adjustments wouldn't even be necessary. Given the limitations of defensive metrics, however, any kind of accuracy in such a number would be limited.
    Last edited by Milnertime; 11-13-2012 at 05:28 AM.

  10. #10
    Join Date
    Jun 2008
    Location
    Baltimore now, but born and raised on the south side of Chicago.
    Posts
    6,824
    vCash
    1500
    Would it make sense to adjust the position values based on the level of play in the league each year. For example when there were a slew of good SS back in the 00s. Tejada, Garciapara, Jeter, and Arod. Plus a bunch of guys who were plying well (good not great) Rentaria, Alex Gonzalez, Eckstein, Valentin, furcal. Shouldn't a SS value go down, because they are easier to replace?

    I really think the value should be judged not by replacing them off the minors, But a situation where 0 is average play in the league for that position. This just makes more sense to me... especially if I were thinking about it from a GM position.

    Quote Originally Posted by MrPoon
    man with hair like fire can destroy souls with a twitch of his thighs.

  11. #11
    Join Date
    Nov 2006
    Posts
    7,039
    vCash
    1500
    Quote Originally Posted by flips333 View Post
    Would it make sense to adjust the position values based on the level of play in the league each year. For example when there were a slew of good SS back in the 00s. Tejada, Garciapara, Jeter, and Arod. Plus a bunch of guys who were plying well (good not great) Rentaria, Alex Gonzalez, Eckstein, Valentin, furcal. Shouldn't a SS value go down, because they are easier to replace?

    I really think the value should be judged not by replacing them off the minors, But a situation where 0 is average play in the league for that position. This just makes more sense to me... especially if I were thinking about it from a GM position.
    The reason that's not done is because there is already offensive value included in WAR, for one, and UZR already does what you're suggesting to an extent.

    Here's what fangraphs says about it:

    Traditionally, offensive position adjustments have been popular, which aligns the positions by adjusting on the basis of the difference in offensive runs. However, due to the variability in offensive performance from year to year, that can lead to miscalculations, such as believing that an NL 2B and an NL SS were equal in 2008 because they had the same batting line. Clearly, shortstops are better defenders than second baseman, and we have to reflect this in their value.

  12. #12
    Join Date
    Jun 2008
    Location
    Baltimore now, but born and raised on the south side of Chicago.
    Posts
    6,824
    vCash
    1500
    Quote Originally Posted by Milnertime View Post
    The reason that's not done is because there is already offensive value included in WAR, for one, and UZR already does what you're suggesting to an extent.

    Here's what fangraphs says about it:
    I guess it depends on What you want to measure. If you want to measure something absolute that happened in the past . Player A was X valuable last year then this makes sense. But now that I think about it that is kind of a pointless assessment. This might have use for GM awards but that is about it.

    If you want to project out, and compare across years then a yearly positional adjustment doesn't make much sense, because it's not an accurate assessment of future value. **** I wish I had more time to futz with this stuff.

    Quote Originally Posted by MrPoon
    man with hair like fire can destroy souls with a twitch of his thighs.

  13. #13
    Join Date
    Feb 2009
    Location
    Connecticut
    Posts
    1,915
    vCash
    1500
    ^Yet it is unclear between 2B and 3B (for example) who is the better defender. They require different skillsets. And honestly, most people that are too slow for SS go to 3B, and most people who don't have the arm for SS go to 2B. It's almost impossible to say which is harder.

    So despite the average 3rd baseman being a much better hitter than the average 2nd baseman, they are even. This makes no sense. I hate to repeat myself, but it is only the obvious distinctions that can be justified between positions.


  14. #14
    Join Date
    Nov 2007
    Posts
    39,545
    vCash
    1500
    Quote Originally Posted by Kenny Krew View Post
    ^Yet it is unclear between 2B and 3B (for example) who is the better defender. They require different skillsets. And honestly, most people that are too slow for SS go to 3B, and most people who don't have the arm for SS go to 2B. It's almost impossible to say which is harder.

    So despite the average 3rd baseman being a much better hitter than the average 2nd baseman, they are even. This makes no sense. I hate to repeat myself, but it is only the obvious distinctions that can be justified between positions.
    2B take more chances then 3B. They have to play with their back to runners, and handle fancy footwork around 2B. They lack the range to their right and arms of SS's. But have to have more range than 3B.

    Quick comments on "defensive spectrum" - a Bill James idea:

    http://www.rcjhawk.us/baseball/spectrum/

    http://www.baseballprospectus.com/ar...rticleid=16033


    6/27/09: “We expect [Rondo] to play by the rules and be a leader as a point guard. We need him to be more of a leader,” Ainge said. “There were just a couple situations where he was late this year, I don’t know if he was sitting in his car, but showed up late and the rest of the team was there. We have team rules and you have to be on time. He was fined for being late, he said he was stuck in traffic, and it’s just unacceptable.”

    Some jerks never learn.....

  15. #15
    Join Date
    Nov 2006
    Posts
    7,039
    vCash
    1500
    Quote Originally Posted by bagwell368 View Post
    2B take more chances then 3B. They have to play with their back to runners, and handle fancy footwork around 2B. They lack the range to their right and arms of SS's. But have to have more range than 3B.

    Quick comments on "defensive spectrum" - a Bill James idea:

    http://www.rcjhawk.us/baseball/spectrum/

    http://www.baseballprospectus.com/ar...rticleid=16033
    Yeah, I think the easy way of thinking of it is who gets the ball hit to them the most. Generally, they're the better defender (or, more is demanded of them to be good at their position).

    I don't think it's exactly that cut and dry, but that's the most simple way of condensing my novel from above.

Page 1 of 2 12 LastLast

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •