Announcement

Collapse
No announcement yet.

What sabremetric dataset includes productive outs?

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

  • #31
    Originally posted by drstrangelove View Post
    I'm not sure how many seasons one needs to drastically cut the luck factor, but I think it's highly unlikely one seasons does it. If you look at the productive outs stat across even a handful of random players you will find that players don't get equal amounts of productive out chances each year.


    If you limit yourself to just seasons with 600 or more PAs

    a) Beltre has had seasons with between 42 and 71
    b) Beltran 53 to 76
    c) Cabrera 44 to 75
    d) Encarnacion 35 to 54
    e) Escobar 35 to 50
    f) ARod 52 to 100

    It's pretty clear that this stat is not random: some players get far more chances than others, while every player has personal seasons that are far more than other seasons.

    And this completely ignores the results of the productive outs. ARod was successful in 20 to 44%, Beltre 19 to 42%, Beltran 24 to 51%, and Cabrera 23 to 53%. Even these players seem to be "great" some years and "horrible" other years, which implies that seasons are far too short to remove the randomness of this stat.

    Using ARod as an example, one year he had 100 productive out chances, the next year 55. One year he was successful at 44%, the next year 22%.

    Beltre was successful in successive seasons at 22%, 32%, 15%, 19% and 37%.
    Cabrera was successful in successive seasons at 34%, 53% and 33%.
    Beltran was successful in successive seasons at 28%, 42% and 23%, then later 24%, 42% and 26%.

    If players' "success" rates go from 32% to 15%, 44% to 22%, 53% to 33%, 42% to 23%, in back to back seasons, it's obvious that this is stat has a lot of luck involved unless we believe batters go from far above average to far below average all the time.


    What ever skill this measures, the "success and failure rate" is far, far more random than batting average, which we all know is already partly luck. The opportunity rate is random as well. So yes, the stat looks like it is far more random than you can expect to be corrected by 500 at bats.

    And of course we know that the success rate is at least affected by the running skill and speed of the man who is scoring which is itself something the batter can't control and is therefore luck. It makes a huge difference if the runner is Mays or Jackie Robinson, versus Frank Howard or Ryan Howard.
    Come on with the luck in batting average, are you speaking of an entire season, a career, luck will probably even out.
    You can get lucky in a short series, no way to ever tell how much luck plays in a 500 at bat season and a career, forget it.
    Prove it, can't be done.

    Comment


    • #32
      Originally posted by scorekeeper View Post
      Mebbe I’ve forgotten or just missed seeing it, but what is the universal definition of a productive out?
      According to BBRef (which comes from Elias)

      1) successful sacrifices by a pitcher with one out
      2) advancing any runner with none out
      3) driving in a base runner with the second out of the inning

      So if the runner on second base is 25 year old Willie Mays, you are skillful when you hit a grounder, but if you hit the exact same grounder and the runner on second base is 42 year old Willie McCovey, you are a failure.

      If you hit a hard grounder that nearly gets to center field, but Ryan Howard trips or is held at third, you are failure, but if it's a routine grounder to short stop and Jackie Robinson takes off on contact and scores, you are skillful.

      A short pop fly to center that Billy Hamilton scores on: skill +1; a long fly to center that Frank Howard is thrown out on, skill -1. If the left fielder has a weak arm or is lazy, +1 to skill; hustles and has a strong arm, -1 to skill.

      If the runner is on third to score on a deep grounder to short because he's Raines, or Henderson, or Morgan, or Cobb and stole second and third, +1 to skill, if the runners is still on second and can't score on the same deep grounder to short because he wouldn't or couldn't steal, -1 to skill.
      Last edited by drstrangelove; 02-19-2016, 01:42 AM.
      "It's better to look good, than be good."

      Comment


      • #33
        Here is Ichiro's batting average:

        2001-.350
        2002-.321
        2003-.312
        2004-.372
        2005-.303
        2006-.322
        2007-.351
        2008-.310
        2009-.352

        -29, -9, +60, -69, +19, +29, -41, +42

        His batting got worse in 2002, 2003, suddenly got better in 2004, but then suddenly got worse in 2005, 2006, but got better in 2007, but got worse in 2008, then got better in 2009.

        George Brett
        1979-.329
        1980-.390
        1981-.312

        Really high in 1980 +61 points, but then got really worse in 1981, -78 points.

        Tony Gwynn
        1986-.329
        1987-.370
        1988-.313

        +41, -57


        Ted Williams
        1940-.344
        1941-.406
        1942-.356

        +62, -50

        1956-.345
        1957-.388
        1958-.328

        +43, -60

        Hank Aaron
        1958-.326
        1959-.355
        1960-.292
        1961-.327

        +29, -63, +35

        1970-.298
        1971-.327
        1972-.265
        1973-.301

        +29, -62, +36

        Tris Speaker
        1915-.322
        1916-.386
        1917-.352
        1918-.318
        1919-.296
        1920-.388

        +64, -34, -34, -22, +92

        TY Cobb
        1919-.384
        1920-.334
        1921-.389
        1922-.401
        1923-.340
        1924-.338
        1925-.378

        -50, +55, +12, -61, -2, +40

        Rod Carew
        1975-.359
        1976-.331
        1977-.388
        1978-.333

        -28, +57, -55

        Helimann
        1920-.309
        1921-.394
        1922-.356
        1923-.403
        1924-.346
        1925-.393
        1926-.367
        1927-.398
        1928-.328

        +85, -38, +47, -57, +47, -26, +31, -70

        Dick Groat
        1959-.275
        1960-.325
        1961-.275

        +50, -50

        Joe Torre
        1969-.289
        1970-.325
        1971-.363
        1972-.289

        +36, +38, -74

        Ashburn
        1957-.297
        1958-.350
        1959-.266

        +53, -84

        Willie McGee
        1984-.291
        1985-.353
        1986-.256

        +62, -97




        My point isn't that BA is all luck, or even 25% luck. But that BA jumps around enough season to season for us to know that 20-40 points a year (about 10-12%) is likely luck. One player who is naturally a .260 hitter might hit .300, while another player who is naturally a .340 hitter, might also hit .300. Over the course of a season, some of this variability will get reduced, but it isn't zeroed out for everyone, because given enough players, it won't (isn't) hard to find players who still show huge variability. Yes, maybe, just maybe, over a long enough career it might average down to something small.

        Productive outs shows a far, far wider variability.
        Last edited by drstrangelove; 02-19-2016, 01:30 AM.
        "It's better to look good, than be good."

        Comment


        • #34
          Originally posted by drstrangelove View Post
          Here is Ichiro's batting average:

          2001-.350
          2002-.321
          2003-.312
          2004-.372
          2005-.303
          2006-.322
          2007-.351
          2008-.310
          2009-.352

          -29, -9, +60, -69, +19, +29, -41, +42

          His batting got worse in 2002, 2003, suddenly got better in 2004, but then suddenly got worse in 2005, 2006, but got better in 2007, but got worse in 2008, then got better in 2009.

          George Brett
          1979-.329
          1980-.390
          1981-.312

          Really high in 1980 +61 points, but then got really worse in 1981, -78 points.

          Tony Gwynn
          1986-.329
          1987-.370
          1988-.313

          +41, -57


          Ted Williams
          1940-.344
          1941-.406
          1942-.356

          +62, -50

          1956-.345
          1957-.388
          1958-.328

          +43, -60

          Hank Aaron
          1958-.326
          1959-.355
          1960-.292
          1961-.327

          +29, -63, +35

          1970-.298
          1971-.327
          1972-.265
          1973-.301

          +29, -62, +36

          Tris Speaker
          1915-.322
          1916-.386
          1917-.352
          1918-.318
          1919-.296
          1920-.388

          +64, -34, -34, -22, +92

          TY Cobb
          1919-.384
          1920-.334
          1921-.389
          1922-.401
          1923-.340
          1924-.338
          1925-.378

          -50, +55, +12, -61, -2, +40

          Rod Carew
          1975-.359
          1976-.331
          1977-.388
          1978-.333

          -28, +57, -55

          Helimann
          1920-.309
          1921-.394
          1922-.356
          1923-.403
          1924-.346
          1925-.393
          1926-.367
          1927-.398
          1928-.328

          +85, -38, +47, -57, +47, -26, +31, -70

          Dick Groat
          1959-.275
          1960-.325
          1961-.275

          +50, -50

          Joe Torre
          1969-.289
          1970-.325
          1971-.363
          1972-.289

          +36, +38, -74

          Ashburn
          1957-.297
          1958-.350
          1959-.266

          +53, -84

          Willie McGee
          1984-.291
          1985-.353
          1986-.256

          +62, -97




          My point isn't that BA is all luck, or even 25% luck. But that BA jumps around enough season to season for us to know that 20-40 points a year (about 10-12%) is likely luck. One player who is naturally a .260 hitter might hit .300, while another player who is naturally a .340 hitter, might also hit .300. Over the course of a season, some of this variability will get reduced, but it isn't zeroed out for everyone, because given enough players, it won't (isn't) hard to find players who still show huge variability. Yes, maybe, just maybe, over a long enough career it might average down to something small.

          Productive outs shows a far, far wider variability.
          Thats for sure, that luck factor with thousands of at bats has to be so small, very small.
          No use even considering it since no one hitter over thousands of at bats is not likely to have more luck than other hitters.
          Luck is overplayed on this board in "some" cases, time periods. I'll give a season, not a career, non factor.
          Last edited by SHOELESSJOE3; 02-19-2016, 07:26 AM.

          Comment


          • #35
            Originally posted by drstrangelove View Post
            According to BBRef (which comes from Elias)

            1) successful sacrifices by a pitcher with one out
            2) advancing any runner with none out
            3) driving in a base runner with the second out of the inning…
            I take it from your comments that like me you don’t give the productive out stat much weight.
            The pitcher who’s afraid to throw strikes, will soon be standing in the shower with the hitter who's afraid to swing.

            Comment


            • #36
              Originally posted by SHOELESSJOE3 View Post
              Thats for sure, that luck factor with thousands of at bats has to be so small, very small.
              No use even considering it since no one hitter over thousands of at bats is not likely to have more luck than other hitters.
              Luck is overplayed on this board in "some" cases, time periods. I'll give a season, not a career, non factor.
              We are absolutely on the same page. I was speaking just to one season. I would not today nor have I ever thought that a career .344 hitter was a lucky .325 hitter or that a .310 hitter was an unlucky .330 hitter. Naturally that assumes that they play out nearly full careers and we respect the effects of parks, but if actual career batting averages were off by more than 5 points from luck (up or down) for someone with 5,000+ at bats, I'd be surprised. Indeed if it were less than 2 points, I wouldn't be surprised.
              Last edited by drstrangelove; 02-19-2016, 01:57 PM.
              "It's better to look good, than be good."

              Comment


              • #37
                Originally posted by scorekeeper View Post
                I take it from your comments that like me you don’t give the productive out stat much weight.
                Yes. If we were to massage the context a lot, which people could do, but isn't done to adjust for opportunities, we still are left with the issue of the skill level of the runner on base which imo can be huge and can't, with today's data, be corrected. I don't doubt that some players do have the smarts and the ability to pick and choose where to move a runner; I simply doubt that we have the methods or data to discern it with any accuracy.
                "It's better to look good, than be good."

                Comment


                • #38
                  Originally posted by drstrangelove View Post
                  Yes. If we were to massage the context a lot, which people could do, but isn't done to adjust for opportunities, we still are left with the issue of the skill level of the runner on base which imo can be huge and can't, with today's data, be corrected. I don't doubt that some players do have the smarts and the ability to pick and choose where to move a runner; I simply doubt that we have the methods or data to discern it with any accuracy.
                  I agree. However, I don’t mind at all when the performance is shown as a rate rather than a total. FI, say if the number of PAs with a runner on 1st is shown along with the number of times that runner has moved up following a PA and that number is expressed as a percentage, I’m fine with it. Ex, player #1 had 100 PAs with a runner on 1st and 60 times that runner moved up for a percentage of 60%.
                  The pitcher who’s afraid to throw strikes, will soon be standing in the shower with the hitter who's afraid to swing.

                  Comment


                  • #39
                    Originally posted by drstrangelove View Post
                    We are absolutely on the same page. I was speaking just to one season. I would not today nor have I ever thought that a career .344 hitter was a lucky .325 hitter or that a .310 hitter was an unlucky .330 hitter. Naturally that assumes that they play out nearly full careers and we respect the effects of parks, but if actual career batting averages were off by more than 5 points from luck (up or down) for someone with 5,000+ at bats, I'd be surprised. Indeed if it were less than 2 points, I wouldn't be surprised.
                    Got that DR.

                    Comment


                    • #40
                      Originally posted by drstrangelove View Post

                      According to BBRef (which comes from Elias)

                      1) successful sacrifices by a pitcher with one out
                      2) advancing any runner with none out
                      3) driving in a base runner with the second out of the inning.
                      Hi Drstrangelove. Love your work and thought processes. Always have admired them. Particularly..your work on WAR and OPS+ re integration was a revelation, and I think it changed a lot of minds here.

                      Ostensibly, Elias' definition of "productive outs" is terribly incomplete.

                      (Just putting this out there to test the waters on productive outs. A case study, if you will, to illustrate a larger point and framework, perhaps)...

                      One might calll it "Total Bases Advanced". The total number of base runners advanced while hitting, plus those extra bases advanced while running.

                      Example: Juan Pierre vs. Adam Dunn:

                      1. ​Juan Pierre and Adam Dunn played almost exactly the same number of career games (1,994 vs 2,001, respectively)...and played almost exactly the same years, also.

                      2. Juan Pierre took 111 more extra bases on passed balls, wild throws, balks, defensive indifference than Dunn.

                      3. Pierre took 239 more extra bases than Dunn on the basepaths (1st-3rd on single, 2nd-home on a double, 2nd-home on a single).

                      4. Pierre had 166 more sacrifice hits than Dunn.

                      5. Pierre advanced runners while batting (independent of sacrifice hits) 64% of the time. (E.g., runners moving up when he put the ball in play). Dunn, 43% of the time.

                      6. Pierre had a 48% career productive out percentage (per Elias' definition). Dunn, 23%.

                      7. Pierre, despite never striking out and being an extreme ground ball hitter, hit into 8 fewer double plays than Dunn.

                      That's about ~700 extra bases Pierre advanced as a runner, and runners he advanced while batting...in essentially the exact same number of games. This is independent of stolen bases and CS, which are already subsumed in other omni metrics.

                      To my knowledge, those 700 extra bases Pierre took (juxtaposed with Dunn) go either almost entirely or completely uncredited, statistically. They are only unaccounted for incidentally (and nobody pays any attention to them).

                      I think, perhaps, the entire way we credit overall offensive value is still extremely incomplete. And, often, totally misleading.

                      Comment


                      • #41
                        Originally posted by Floyd Gondolli View Post

                        Hi Drstrangelove. Love your work and thought processes. Always have admired them. Particularly..your work on WAR and OPS+ re integration was a revelation, and I think it changed a lot of minds here.

                        Ostensibly, Elias' definition of "productive outs" is terribly incomplete.

                        (Just putting this out there to test the waters on productive outs. A case study, if you will, to illustrate a larger point and framework, perhaps)...

                        One might calll it "Total Bases Advanced". The total number of base runners advanced while hitting, plus those extra bases advanced while running.

                        Example: Juan Pierre vs. Adam Dunn:

                        1. ​Juan Pierre and Adam Dunn played almost exactly the same number of career games (1,994 vs 2,001, respectively)...and played almost exactly the same years, also.

                        2. Juan Pierre took 111 more extra bases on passed balls, wild throws, balks, defensive indifference than Dunn.

                        3. Pierre took 239 more extra bases than Dunn on the basepaths (1st-3rd on single, 2nd-home on a double, 2nd-home on a single).

                        4. Pierre had 166 more sacrifice hits than Dunn.

                        5. Pierre advanced runners while batting (independent of sacrifice hits) 64% of the time. (E.g., runners moving up when he put the ball in play). Dunn, 43% of the time.

                        6. Pierre had a 48% career productive out percentage (per Elias' definition). Dunn, 23%.

                        7. Pierre, despite never striking out and being an extreme ground ball hitter, hit into 8 fewer double plays than Dunn.

                        That's about ~700 extra bases Pierre advanced as a runner, and runners he advanced while batting...in essentially the exact same number of games. This is independent of stolen bases and CS, which are already subsumed in other omni metrics.

                        To my knowledge, those 700 extra bases Pierre took (juxtaposed with Dunn) go either almost entirely or completely uncredited, statistically. They are only unaccounted for incidentally (and nobody pays any attention to them).

                        I think, perhaps, the entire way we credit overall offensive value is still extremely incomplete. And, often, totally misleading.
                        Tom Thress attempts to separate baserunning components into multiple levels, using Juan Pierre:
                        http://baseball.tomthress.com/StatTa...x=&id=pierj002

                        Component 1
                        Basestealing (stolen bases, caught stealing, pickoffs, balks). Component 1 decisions are allocated to baserunners, pitchers, and catchers.

                        Component 2
                        Wild pitches and passed balls. Component 2 decisions are allocated to baserunners, pitchers, and catchers.

                        Component 7
                        Double Plays. Component 7 decisions are allocated to batters, baserunners, pitchers, and fielders.

                        Component 8
                        Baserunner Outs. Component 8 decisions are allocated to batters, baserunners, and fielders.

                        Component 9
                        Baserunner Advancement. Component 9 decisions are assessed to batters, baserunners, and fielders.

                        Tom has Pierre worth ~ 3.5 wins on the bases, Dunn at negative ~1.25 wins.
                        Jacquelyn Eva Marchand (1983-2017)
                        http://www.tezakfuneralhome.com/noti...uelyn-Marchand

                        Comment

                        Ad Widget

                        Collapse
                        Working...
                        X