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
    Commentary for scientific statisticians
    If anyone is questioning the technicalities of how we round, we should technically round our final OPS+'s to just 2 places, as well as all of our final rates, so slugging would look like .59 not .592 or .587.

    This is because we are using league OPS+ numbers to remove pitchers and those OPS+ numbers tend to have 2 digits (ie 94 or 93 or 95). However, it is my opinion the concerning significant figures that when you are using an implicit 100 point scale, you should not lose a significant figure just because your final average is slightly less than 100 rather than slightly above 100. If pitcher outhit the league and league OPS+ scores were 101 versus 99, I don't believe that justifies one less degree of accuracy. We could just arbitrarily make OPS+ on a 200 point scale from 100 up and get another digit in all the answers without having any more real accuracy.

    Hits and at bats are discrete variables so we don't have to limit our significant figures ie 99 hits in 326 at bats is .30368098 as far as we want to go because you can't have half a hit or half an at bat, but I personally believe that when you have an implicit three place (or 4 place or 5 place) scale you should keep the same number of digits whether you are just below (99) or just above (102) the number of digits in the scale.

    Everyone else can start reading again.
    It's interesting how baseball-almanac & fangraphs round up vs. baseball reference.

    Example:
    Fangraphs lists 1947 AL slg avg at .365
    Baseball-almanac lists it precisely at .36445 rounded up to .365
    Baseball reference lists AL slugging that year at .364

    Apparently baseball reference rounds up 4 decimals while Fangraphs & BB almanac rounds up 5. .36445 rounds up to .3645 which round up to .365. BBRef's method seems to be .3644= .364.

    I have always used BBref averages and will continue to do so probably because I agree with their methodology. (rounding first 4 digits instead of first 5)
    Last edited by layson27; 05-27-2014, 10:15 AM.

    Comment


    • Originally posted by brett View Post
      The thing is that players rates don't necessarily stay the same if they walk less. And for gosh sakes a walk is worth more than just about any average non-walk at bat for anyone. Linear weights says that a walk is worth .35 runs.

      If you slugged .850 and batted .400 you are only creating about .23 runs per at bat.

      Ruth produced about 2500 net runs in about 10,000 plate appearances, but if someone walked every time they would produce about 3500. Now THAT might be something interesting but we would need to know the linear weights value for a walk for each year. How close a player came to being worth more than he would have been if they had just walked him every time. Using those numbers, Ruth for his career comes out being worth about .714 (ironic) of the value that someone would have if they walked every time in random situations.

      Let's see how close Ruth comes in his top rate seasons.

      In 1920 I have Ruth worth about 177 total runs, and a 100% walk rate to be worth about 216 for a rate of about .819
      In 1923 I have Ruth at about 192 and a 100% walk rate worth about 244 for a rate of .787

      I have always thought that a few players like Ruth, and also Williams and Hornsby and Bonds when their OB% starts to get real close to or above .500 they were really changing the entire approach to pitching to them, but we do see that no one QUITE reached the level where they outperformed 100% walk rate.
      Yeah I wasnt trying to diminish the value of a walk. Just thought it would be interesting to see how counting totals, namely the so-called benchmark stats would change.

      Anyway, a handful of final numbers are in the first post. See what you think. Should I list their actual OPS next to it? How much better is this one, than the one we already knew?

      Comment


      • Was also thinking about doing some comparison by those at your position, over your career.

        And then I thought about WAA and the baseline thing. Speaking only of WAA here, could we somehow set the baseline as the avg, of say, the top 50 all-time at that particular position? Or maybe a bit higher if it sounds more fair.

        Comment


        • I like it. Really, what I will keep coming back for is the expected OPS+. You put in the work and that should stand as a reference. BBRef has had the single season sOPS+ on the road, but never did it for career, nor did they factor out pitchers.

          I would like to see Piazza's next!

          Comment


          • Ok, he'll be the next guy I do in the main chart.

            Anyway.....

            Here's Roberto's best five and ten ADJ REL SA

            Any other guys anyone is curious to see?

            RobertoBest10.jpg
            Last edited by Sultan_1895-1948; 05-28-2014, 05:43 PM.

            Comment


            • We have the top 5 slugging seasons for Mays & Aaron. I'd like to see how close they are after 10 years. I imagine they still would be very, very close.

              Comment


              • Ok I can do that. Did you ever do Vladdy?

                Comment


                • Originally posted by Sultan_1895-1948 View Post
                  Ok I can do that. Did you ever do Vladdy?
                  I just have his relative slugging, 1.30.
                  What's your source for league OBP & OPS+? I want to be on the same page in case I do Guerrero's & Pujols' relative OBP & Expected OPS+. I've been using BBRef, but I want to make sure I'm using the same stats you are.

                  Comment


                  • Yes, BBref, league by league. I won't post the excel file since you said you don't have it.

                    This is what I have so far for a data base. Still working on the SLGeff though.

                    OPS+-OBP-SA-SLGeff.jpg

                    Comment


                    • Oh, and here's Mays/Aaron best five and ten. Couldn't be more tight but Aaron eeks out a 10 year victory thanks in part to a lower league slugging. For what it's worth, their league OPS+ is the exact same (to the decimal), so removing pitchers wouldn't change anything.

                      Actually, best 10 might be a better measure of true greatness, than seeing career numbers. If I had a choice, I think I'd value best ten over overall when ranking player. Maybe a 70/30 split, if I could figure out a way to make that work mathematically. Taking away early years and decline years, but not excluding them if they happened to be great. Sounds good to me. What do you think?

                      MaysAaronBest10.jpg
                      Last edited by Sultan_1895-1948; 05-28-2014, 10:34 PM.

                      Comment


                      • Originally posted by Sultan_1895-1948 View Post
                        Oh, and here's Mays/Aaron best five and ten. Couldn't be more tight but Aaron eeks out a 10 year victory thanks in part to a lower league slugging. For what it's worth, their league OPS+ is the exact same (to the decimal), so removing pitchers wouldn't change anything.

                        Actually, best 10 might be a better measure of true greatness, than seeing career numbers. If I had a choice, I think I'd value best ten over overall when ranking player. Maybe a 70/30 split, if I could figure out a way to make that work mathematically. Taking away early years and decline years, but not excluding them if they happened to be great. Sounds good to me. What do you think?

                        [ATTACH]139170[/ATTACH]
                        That's close enough to make a Norelco razor commercial.

                        Comment


                        • Originally posted by brett View Post
                          I like it. Really, what I will keep coming back for is the expected OPS+. You put in the work and that should stand as a reference. BBRef has had the single season sOPS+ on the road, but never did it for career, nor did they factor out pitchers.

                          I would like to see Piazza's next!
                          Piazza and a few others are finished. He gained 8 OPS+ points and finished just a hair behind Foxx.

                          You were right about Boggs. Overall he loses 15 OPS+ points, down to 116 when adjusted.

                          Comment


                          • Originally posted by Sultan_1895-1948 View Post
                            Actually, best 10 might be a better measure of true greatness, than seeing career numbers. If I had a choice, I think I'd value best ten over overall when ranking player. Maybe a 70/30 split, if I could figure out a way to make that work mathematically. Taking away early years and decline years, but not excluding them if they happened to be great. Sounds good to me. What do you think?
                            In general, I agree with that approach
                            I do prefer 10 best consecutive years, since it rewards consistency over many seasons. But that runs into problems with players who missed time for military duty.

                            My overall formula for ranking hitters is:
                            Best 10 consecutive years- weighed twice (with adjustments for players missing time due to war)
                            Best 14-15 consecutive years- weighed once (adjusted for those who missed time due to war)
                            Best 4 hitting seasons- weighed once
                            Career value/rate stats- weighed once

                            Granted, if I do 10 best overall it eliminates some problems, but really only a few all-time great hitters missed multiple seasons during their prime.

                            But now that I think about it, using 10 best seasons for those few players affected would make things much easier, and it makes sense too.

                            Comment


                            • Interesting stuff. Milwaukee was a bad offensive park, but even worse for Mathews, but notice how guys who lose slugging at home (or on the road, Boggs) often DON'T suffer much in the OB% department. Wonder if there is a reputation thing going on.

                              Also unexpected, Cobb's park tend to reduce offense. He had higher rates on the road, but a little lower relative rates. In fact, Cobb has a 102 road "t"-ops+ but his adjusted s-OPS+ as you calculated turns out to be 164-165 which means he park adjusted home road splits were 106/94 (versus the normal 104/96). It means his park was a solid 6+% below average.

                              I think Ott tends to get rated a little higher than Mathews, but I'd flip flop them now.

                              In fact Mathews, given 2100+ games at third base may exceed Foxx and Lajoie. How about Frank Robinson v Mathews?
                              Last edited by brett; 05-29-2014, 09:12 AM.

                              Comment


                              • I think players deserve credit for lasting, but of course it cuts rates. If you take a players margin OPS+ above some level like 60 or 80 and multiply it by games or plate appearances it avoids giving too much for hanging on.

                                For example, if a guy has a 140 OPS+ for 12,000 PAs and another guy is 170 for 8000 PAs , and you use a margin baseline of 80, you would get:

                                Player 1: 140-80=+60 x 12,000 PAs (and maybe divide by 650) to get +1108 (that is OPS+ above margin x seasons)

                                Player 2: 170-80=+90 x 8000 PAs is also +1108

                                Or you can set whatever baseline you like, though I wouldn't go above 100. 80 would be about half way between "replacement" and average.

                                Comment

                                Ad Widget

                                Collapse
                                Working...
                                X