Announcement

Collapse
No announcement yet.

Any new simple correlations better than OPS and wOBA?

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

  • wes_kahn
    replied
    Thanks for all these replies. After reading the one just above here by Stolensingle, my next question deals with ballpark effects.

    I have issues with how ballpark effects are calculated. If 28 ballparks are all bandboxes, and just two ballparks are fair, and none are pitcher's parks, I believe that ballpark effects should show this to be true rather than give pitcher friendly marks to parks 22-28, 7 bandboxes. Today, you have parks that aid batters more like Baker Bowl, but the park effects show them to aid pitchers by cutting down on home runs compared to other even easier parks. The number 30 park today is still a more hitter friendly park than most if not all the parks in 1935, except for Baker Bowl.

    I believe that ballpark effects don't change all that much from year to year, unless there is a change in the park itself (like moving the fences, changing foul territory, etc.) Today's Coors Field should be compared with Huntington Avenue Grounds of 1903, not just those two, but what I am getting at is that even the #25 offense-aiding park in today's Majors would rate as more hitter friendly than the most hitter friendly park of 1910.

    It's like having 28 Baker Bowls today. It probably greatly affects pitchers' arms today. Christy Mathewson admitted that he only beared down and threw his hardest to just a few batters per game and then just one or two pitches in that at bat. He actually believed in not showing the batter his best stuff until it was necessary. He could afford to give up a 380 foot jack to a batter, because unless it was pulled down the line, that was an easy out.

    It is all part of my belief that there needs to be a minimum area of fair territory in all Major League Parks, and most likely every park in existence today does not meet the minimum I am thinking about. It's fine to have a tiny distance to a foul line, but you better have a Death Valley somewhere else, like old Yankee Stadium or League Park.

    It is also my belief that opening the game to where running on the bases and not striking out or flying out would be the most advantageous way to win. It's just about impossible to watch a 3 1/2 hour game with 25 strikeouts, 3 to 4 home runs, and 10 pitchers. I do not believe in re-creating 1968, because that sucked pretty bad too. Something more like the 1930's would be much better when instead of having 20 batters top 40 homers, having one or two players hit 40 homers, but having 5 or 6 top .350 batting averages, with somebody threatening .400, and having a player with 20 triples, and maybe another with 55-65 doubles, and then having the best pitchers completing half their starts would make baseball much more popular and much more exciting.

    This will be the first Major League season since about 1954 where I have no plans to go to a game. I think I saw a handful of games at Ebbets Field in 1955, but I was 4. I know I saw multiple games in 1956. Instead of getting to Dodger Stadium or Progressive Field in 2020, I will be watching Kumar Rocker and Jack Leiter pitch for Vanderbilt. They will pitch complete games, and Hawkins Field is more pitcher friendly, where small ball must be used to win, even if you have the best power hitter in the nation like they had last year with J. J. Bleday.

    The Phillies Use Lifebuoy Soap, And They Still Stink.

    Leave a comment:


  • Stolensingle
    replied
    Originally posted by wes_kahn View Post
    I always thought that the calculation of total bases plus walks plus hit by pitch divided by at bats plus walks plus hit by pitch plus sacrifice flies would be an even better stat, and then others have considered adding stolen bases in the numerator and caught stealing with a constant in the denominator would be even more accurate.
    Yes, I call (TB + BB + HBP) cumulative bases, and have shown that it correlates > .90 with FG's batting runs.

    But more accurate measures (like batting runs) are obtained with linear weights. The old triples alley site published some linear weights formulas that included SB and CS, e.g.:

    .47*1B + .75*2B + 1.04*3B + 1.40*HR + .33*(BB + HBP) + .18*SB – .45*CS – .27* (AB – H)

    http://triplesalley.wordpress.com/20...inear-weights/

    You would divide the LW sum by PA to get a wOBA value that includes SB. Of course, newer base running metrics go beyond SB, to include advancing on the bases and avoiding DP. So let's see how we can incorporate them.

    FG's Offensive RAA includes all the LW data used in wRC+, plus base-running. It's not a rate stat like wRC+, wOBA or OPS+, though. To convert it into a rate stat, we divide it by PA.

    E.g., if you determine Off RAA/PA for last year, the top values are:

    Trout .1137
    Yelich .1124
    Bregman .0843
    Bellinger .0838
    Springer .0736
    Marte .0732
    Rendon .0720
    Moncada .0583
    Soto .0545
    Alonso .0534

    If you divide this by league average R/PA, .126, add 1.00, and multiply by 100, you get a value that is analogous to wRC+, but with base running taken into account:

    Trout 190
    Yelich 189
    Bregman 167
    Bellinger 166
    Springer 158
    Marte 158
    Rendon 157
    Moncada 146
    Soto 143
    Alonso 143

    These values are still not quite what we want, though, because I haven't adjusted them for park factors. They're actually like wOBA values, with baserunning included, except that they're relative to 100, rather than to league average R/PA, as real wOBA values are presented.

    However, there's a fairly simple way to take into account park factors.. Since park dimensions presumably don't come into play for baserunning--at least, not in any obvious, quantifiable way--we can just add baseruns AA/PA to the wRC+ values. That is, we divide baseruns by PA, divide the resulting value by .126 to get baseruns per PA AA, then multiply these values by 100 to get values that can simply be added to the wRC+ values. When we do this, we get what we might refer to woRC+ values, i.e., weighted offensive RAA. I have included the wRC+ values in (), so one can compare a player's total offense relative to just his hitting:

    Trout 189 (180)
    Yelich 186 (174)

    Bregman 166 (168)
    Bellinger 164 (162)
    Springer 157 (156)
    Marte 155 (150)
    Rendon 154 (154)
    Moncada 146 (141)
    Bogaerts 142 (141)

    Soto 141 (142)
    Betts 141 (135)
    Alonso 140 (143)
    Muncy 140 (134)
    Bryant 140 (135)
    McNeil 139 (143)
    Semien 139 (137)
    LeMahieu 135 (136)
    Acuna 135 (126)
    Devers 135 (132)
    Arenado 131 (128)

    Donaldson 128 (132)
    Story 128 (121)
    Chapman 126 (125)

    Suarez 126 (133)
    Harper 124 (125)

    Albies 122 (117)
    Kepler 119 (121)
    Baez 116 (114)
    Realmuto 115 (108)

    Grandal 115 (121)


    Players with positive base runs (in blue) have woRC+ values that are higher than their wrC+ values. Trout, Yelich and Acuna, who all finished in the top 5 in baseruns, get a substantial boost, as do some others. Players with negative baseruns (in red) have woRC+ values that are less than their wRC+ values. However, for most players, the difference between wRC+ and woRC+ is fairly minor. In fact, the correlation between this metric and wRC+ for the top 30 in WAR is .97. For someone like Billy Hamilton, of course--a poor hitter but a great baserunnner--it makes a big difference. His career wRC+ is a well below replacement level 67. His woRC+ is 83, which is at least above replacement, Another good example, is Jonathan Villar, who led MLB in total base runs last season. His wRC+ was 107, whereas his woRC+ was 119. Baserunning is not going to turn a poor hitter into a great offensive player, but it can make a significant difference in value for some players.

    My question to those that are far more expert with this stuff is: Has a correlation for any of the alternates ever been run to see if they are any more or less accurate in predicting runs than regular OPS or wOBA?
    The best correlation with actual runs scored is obtained with baseruns--not to be confused with baserunning runs that I've discussed above. This is discussed in the triplesalley link I posted above. In fact, the author uses baseruns to adjust the linear weights formulas, so they come closer to predicting actual runs scored.
    Last edited by Stolensingle; 01-25-2020, 10:00 PM.

    Leave a comment:


  • dominik
    replied
    Originally posted by Jar of Flies View Post
    wRC+ is a more refined version of OPS+ available at Fangraphs for a good starting hitting metric.
    It doesn't include baserunning though. Baserunning is only like 10% or so of offense so hitting is a good proxy but to fit 100% you have to include baserunning. Best correlation thus is probably WAR.

    Leave a comment:


  • Jar of Flies
    replied
    wRC+ is a more refined version of OPS+ available at Fangraphs for a good starting hitting metric.

    Leave a comment:


  • Any new simple correlations better than OPS and wOBA?

    Of the simple correlations of the 21st century, OPS and wOBA have been considered among the best and most accurate correlations to runs and wins.

    However, so many people have discussed that both OBA and SLG count the hits in both parts of the calculation. There have been others that posited that on base plus isolated power should be a better calculation. I always thought that the calculation of total bases plus walks plus hit by pitch divided by at bats plus walks plus hit by pitch plus sacrifice flies would be an even better stat, and then others have considered adding stolen bases in the numerator and caught stealing with a constant in the denominator would be even more accurate.

    My question to those that are far more expert with this stuff is: Has a correlation for any of the alternates ever been run to see if they are any more or less accurate in predicting runs than regular OPS or wOBA?

    I am not convinced that stolen bases and caught stealing get enough worth in calculations. The eye test to me has me believing that players like Maury Wills, Lou Brock, and others would not have scored runs like they did without stealing tons of bases. Wills never would have come close to 130 runs if he had chosen not to attempt a steal in 1962 or 100+ in 1961. I believe those steals were worth a lot of runs and the 89% success rate in 1962 meant a lot more than just the stolen bases themselves. It forced pitchers to throw more fastballs to players hitting behind Wills. It forced pitchers to divert some attention to batters. Numerous times, pitchers lost their effectiveness when Wills got on base and stole second and sometimes third.

    Jim Gilliam and Tommy Davis both commented in interviews that they never saw as many fastballs in their career as they did when Wills was on base. It's no coincidence that Davis, who was at the tops in connecting against fastballs had his career year when Wills had his career year. Gilliam's 1962 and 1963 seasons were major Phoenix rising from the ashes seasons for him after coming close to being replaced after 1960 and 1961.

Ad Widget

Collapse
Working...
X