Announcement

Collapse
No announcement yet.

Progressing Toward Better Stats Thread

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Originally posted by brett View Post
    I was thinking the whole rest of his career. I'm pretty sure his rrOPS+ was virtually identical to his raw OPS+ after leaving Boston.

    103.0 before giving home as road.

    BoggsRestOfCareer-rrOPS+.jpg

    Here's Larry Walker, only full Colorado years 1995-2003....125.1 before giving home as road

    That 1996 SA isn't a misprint. He slugged .800 at home that year. A 212 to 38 sOPS split!

    LWalkerRockiesYears-rrOPS+.jpg
    Last edited by Sultan_1895-1948; 02-21-2015, 06:47 PM.

    Comment


    • Originally posted by Stolensingle View Post

      If that’s the case, I would take that as good evidence that in Boggs’ case, it was more a home park factor than just being at home, and thus support for Sultan’s approach. Of course he was far past his prime when he moved to NY. I’d also like to see his last two years in Boston vs. his first two years in NY, which are more comparable in terms of age.
      I'd think a player who was better at home would have an even greater discrepancy in his decline as he would be more easily marginalized on the road where he didn't have the same advantage.

      As you can see, without giving Boggs his home park on the road, his road numbers are consistent with a 129 OPS+ versus an actual of 144 during the Boston years.
      His home to road park adjusted OPS+ is 159 to 129. This is adjusted such that the average player will produce 100 at home and 100 on the road
      His home to road park adjusted OPS+ after his Boston years are 110 to 104. (His road predicted is overall is 104, his actual is 107).

      It probably is worth looking at the last 2-3 years in Boston where he was declining.

      Sultan will get a more accurate number, but quick math looks like 118 to 100 overall to road predicted in his last 2 years (136 to 100 home road). That is an even greater discrepancy between road predicted and overall than for his entire Bosox career. For his career in Boston his road predicted is 89.6% of his actual. For his last 2 years it is about 85% of his actual. For his post Bosox career it is 97.2%.

      OK as for the neutralized environment, if we have an 8 team league, and we count the 7 road parks as 7/8 weight and the home park as 1/8 weight and adjust the road and home to remove the home field advantage and road field loss of an average player, how is that not neutral. Just an 8 team league doesn't that work?

      Comment


      • OK, let me see if I have this right:

        Code:
        Boggs	rrOBP	rrSA	rrOPS+	OPS+	wrrOPS+
        Boston	1.213	1.024	123	145	129
        NY, etc.1.123	0.9087	103	107	104
        Where wrrOPS+ is after home is added back, i.e., rrOPS+ and OPS+ are averaged after weighting each so that home park is given just 1/8 weight.

        I would call this support for your system. It suggests that Boggs’ production at Fenway was not just a home effect, but a home park effect, because when he moved to NY, the home effect was much reduced. Actually, the effect may be underestimated, because in the modern league, a home park should be weighted less than 1/8.

        I would like to see the same kind of results for a player whose move occurred before he was in such decline, but this is a very good start.

        Comment


        • By the way, I have always speculated that playing half time in Coors hurt players when they went on the road. There is certainly the possibility that Boggs home adjustement negatively affected what he might have done on the road if he had had another home park. I would consider rrOPS+ to be a lower or upper limit on a players true ability to perform, while his actual OPS+ would form the other limit. That is why I personally would still average the two, actual, and rrOPS+ for player evaluations.

          And ideally I would apply that as a modifying coefficient to another stat like Rbat.

          Comment


          • Originally posted by Stolensingle View Post
            I would like to see the same kind of results for a player whose move occurred before he was in such decline, but this is a very good start.
            Pick a few guys.

            Larry Walker transitioning from Montreal to Coors might be worth looking at. Same with Reggie in Oakland and after Oakland.

            Speaking of Oakland, it reminds me of foul territory and it's impact over the course of a season.

            Comment


            • Originally posted by brett View Post
              By the way, I have always speculated that playing half time in Coors hurt players when they went on the road. There is certainly the possibility that Boggs home adjustement negatively affected what he might have done on the road if he had had another home park. I would consider rrOPS+ to be a lower or upper limit on a players true ability to perform, while his actual OPS+ would form the other limit. That is why I personally would still average the two, actual, and rrOPS+ for player evaluations.

              And ideally I would apply that as a modifying coefficient to another stat like Rbat.
              I think you're right back to penalizing guys like DiMaggio or Piazza, and right back to rewarding guys who were lesser hitters across all environments.

              In other words, averaging the two won't impact guys who were around five points or so either way, but it WILL bring the extremely hindered fellas, and the extreme pretender fellas closer. Is that the goal?

              There's room for another column so if we come up with a "something to consider if you wish" number like you're talking about, I'll follow along.
              Last edited by Sultan_1895-1948; 02-21-2015, 08:05 PM.

              Comment


              • Originally posted by brett View Post
                By the way, I have always speculated that playing half time in Coors hurt players when they went on the road. There is certainly the possibility that Boggs home adjustement negatively affected what he might have done on the road if he had had another home park. I would consider rrOPS+ to be a lower or upper limit on a players true ability to perform, while his actual OPS+ would form the other limit. That is why I personally would still average the two, actual, and rrOPS+ for player evaluations.

                And ideally I would apply that as a modifying coefficient to another stat like Rbat.
                You're certainly not alone on the Colorado road effect, I think a lot of people believe it happens. The possibility that it might apply to other home parks is novel, though, that hadn't occurred to me.

                When you say you would average OPS+ and rrOPS+, I assume you mean by rrOPS+ the value that has home added back, so to speak. I really think Sultan needs to distinguish the two. I would call the value you get by adding rrOBP and rrSA (-1) rrOPS+, and the value you get after the weighting process that adds back home wrrOPS+. (Or maybe call them rrOPS and rrOPS+, respectively). So my understanding is that you would use both OPS+ and wrrOPS+ as limits, and view the hitter's true production as somewhere between them.

                Originally posted by Sultan_1895-1948 View Post
                Pick a few guys.
                As I said before, Bonds and Piazza look like the best test cases out there. Both made moves when they were still in their prime, and for Bonds an added attraction is that he changed home parks once while not changing his home city.
                Last edited by Stolensingle; 02-21-2015, 08:14 PM.

                Comment


                • Except you're dealing with a monster. Literally. Bonds numbers are whack. I will say that I did year by year for just his Pittsburgh years and he gained two OPS+ points.

                  Anyway, awhile back in this thread, we were doing top 5 and top 10 comparisons, but that was only sorting by raw OPS.

                  What I've done here is look at each individual season of rrOPS+ so we can look at top 5 and top 10.

                  I've added a pitchers removed column followed by rrOPS+ which is home as road.

                  In his last season, Piazza was still an above average hitter in a neutral setting.

                  I'll also add that we can see how much the numbers fall when removing pitchers from the league. Imo that same boost (or close to it) should come back the other way, when you consider competing against multiple juicers. Perhaps it's only half of the impact but it matters.


                  PiazzaYearByYear-rrOPS+.jpg


                  Here's Manny, split by two teams, who produced no matter what.

                  The thing is, you can have better home than road, if your road is still very productive compared to the league road. It shows what kind of hitter you are.


                  Manny-Cle-Bos-rrOPS+.jpg
                  Last edited by Sultan_1895-1948; 02-21-2015, 11:08 PM.

                  Comment


                  • So if you take Piazza’s peak years, from 1993 – 2003, in 8/11 of the seasons, his raw OPS+ was < his rrOPS+, by an average of 18.0 points/year. In the other three years, his raw OPS+ > his rrOPS+, by an average of just 6.3 points/year. The eight years when his raw OPS+ was lower include 3/5 of his full seasons with the Dodgers, 4/5 of his peak seasons with the Mets, and the transition year when he played for both teams. In addition, his raw OPS+ < rrOPS+ in each of his last two years, in San Diego and Oakland, and overall in 10 of the 15 seasons in which he played > 50 games.

                    These data thus indicate that Piazza’s relatively better road performance was not dependent on a particular home park, but was evident wherever he played, most particularly in LA and NY, when he was in his best form, and there is a large sample size. This suggests to me that for Piazza, unlike Boggs, the big difference between road and home was due to the general road/home conditions. We can’t say whether he was relatively poor at home or relatively good on the road, but a particular home park does not appear to be the explanation here.

                    So at this point, I think some players, like Boggs, may have gotten an extra boost by playing in a particular home park, but not all players who show a big difference between rrOPS+ and raw OPS+ can be accounted for in this manner. In fact, I'd expect any player who put in significant time in more than one home park would probably turn out like Piazza, because if the home/road difference resulted from a particular home park, it would have to be huge to show up despite all the other years playing in some other home park. DiMaggio, a lifetime Yankee, would be a good candidate to be in the Boggs group, though just because he did play all his years for one team, we can't do this kind of test for him.

                    It's a difficult situation, almost a no-win situation. To show that it's a home park effect and not just a general home effect, you need to look at players who played in more than one home park, but as I just noted, if they played a substantial part of their career in more than one home park, any specific home park effect is likely to be obscured by the lack of such an effect when he played in another home park. Players like Boggs can be identified because they played most of their career in one park, but otherwise it will have to be a very large effect to show up. In fact, it might be that some players who are more or less even on your list, i.e., raw OPS+ ~ rrOPS+, really did have a home park effect, but it's obscured by the lack of such an effect when they played in another home park. You might want to look at McGwire, who seems to have a slight home edge, because he played most of his games in Oakland, but some in St. Louis. If he had a substantial edge when he played at home in St. Louis, but no essential road/home difference in Oakland, it could account for the slight overall appearance of a home advantage.
                    Last edited by Stolensingle; 02-22-2015, 07:21 AM.

                    Comment


                    • Here's The Big Cat year by year....110.6 overall rrOPS+

                      Andres-rrOPS+.jpg
                      Last edited by Sultan_1895-1948; 02-22-2015, 12:32 PM.

                      Comment


                      • Originally posted by Stolensingle View Post
                        You're certainly not alone on the Colorado road effect, I think a lot of people believe it happens. The possibility that it might apply to other home parks is novel, though, that hadn't occurred to me.
                        Now I have realized that Coors IS unique because there is a huge amount of evidence that pitches move differently in CO, especially that fastball move less than half as much, which would tend to create in a hitter the tendancy to swing at close strikes on fastball which he might take with him on the road, and then suffer because the fastballs move 250% as much. Other parks vary primarily by what the ball does after it is hit.

                        But the Coors field effect on road rates seems to be easy to see in guys like Jeff Cirillo, as well as Andre Dawson who became a better road hitter in '98 when he left Colorado.

                        Comment


                        • Originally posted by Sultan_1895-1948 View Post
                          Here's The Big Cat year by year....110.6 overall rrOPS+

                          [ATTACH]146092[/ATTACH]
                          What is odd here is that it looks like Andres was BETTER in road relative numbers from 1993-1998 when he played in Colorado.

                          It looks like 128 rrOPS+ during the Colorado years and 126 actual.

                          He was a much better relative hitter at home versus the road during the years when he didn't play in CO.

                          Sulatn, his road relative OPS+ looks like it matches or exceeds his raw OPS in most seasons, yet his raw OPS+ is 119. He is better on the road in 9 of 17 seasons and virtually identical in 4 or 5 others.
                          Last edited by brett; 02-22-2015, 12:39 PM.

                          Comment


                          • Here are the years sorted by gap between rrOPS+ and raw

                            14.9 - COL
                            7.4 -- SFG
                            6.2 -- COL
                            6.2 -- ATL
                            5.7 -- MON
                            5.3 -- MON
                            4.4 -- MON
                            3.4 -- COL
                            3.1 -- MON
                            -0.1 - STL
                            -0.1 - COL
                            -0.7 - MON
                            -1.4 - ATL
                            -3.1 - MON
                            -5.3 - TEX/SFG
                            -14.8 - COL
                            -25.7 - MON

                            Comment


                            • Here's Walker year by year

                              LWalkerYearByYear-rrOPS+.jpg

                              Comment


                              • Originally posted by brett View Post
                                What is odd here is that it looks like Andres was BETTER in road relative numbers from 1993-1998 when he played in Colorado.

                                It looks like 128 rrOPS+ during the Colorado years and 126 actual.

                                He was a much better relative hitter at home versus the road during the years when he didn't play in CO.

                                Sulatn, his road relative OPS+ looks like it matches or exceeds his raw OPS in most seasons, yet his raw OPS+ is 119. He is better on the road in 9 of 17 seasons and virtually identical in 4 or 5 others.
                                I don’t see that at all. If you look at rrOPS+ vs. OPS+, the pattern appears quite random, with rrOPS+ > OPS+ in nine seasons, less in eight seasons. In CO, greater in three seasons, less in two seasons. The difference is in double digits in only three seasons in his entire career, once in favor of rrOPS+, twice in favor of OPS+. This has all the appearance of a random effect to me, with no underlying home or road bias. If you add up all the differences, 56.6 vs. 51.2, though this isn’t precise, because some rate differences should be weighted more or less, depending on PA in that season.

                                In Walker’s case, there is a definite home bias, rrOPS+ < OPS+ in 13/16 seasons. But this bias is as evident in Montreal as well as Colorado (4/5 seasons there vs. 8/9 in CO), suggesting a general home effect. The effect does appear larger in CO, so that might be taken as evidence of some home park effect. But given that he showed a general home effect in Montreal, perhaps as he got older, and stayed longer in CO, this general effect simply got enhanced.

                                Comment

                                Ad Widget

                                Collapse
                                Working...
                                X