Announcement

Collapse
No announcement yet.

PECOTA projections puts White Sox in 3rd place

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

  • blurry
    replied
    https://twitter.com/MLB/status/13617...%2Fbaseball%2F

    Fangraphs has the White Sox at 60.5% chance of making the playoffs. They are 2nd in the AL Central behind the Twins whom have a 65.5% chance according to their projections. I think that's fair; the Twins are the team to beat in the division until proven otherwise.

    So for the AL playoff projections:

    91.6% Yankees
    71.4% Astros
    65.5% Twins
    60.5 % White Sox
    53.6% Blue Jays

    Sounds fair to me. The AL West is a mess. There's a huge drop off in talent after the White Sox in this division. Also lol at the Baltimore projection at 0.0%.

    Just need to get into the playoffs in baseball and things basically start over.

    Leave a comment:


  • TDog
    replied
    Originally posted by Mohoney View Post

    By this logic, baseball message boards are also pointless, because none of us (as far as I know) are credentialed employees in MLB front offices.
    It baffles me that such an inference can be drawn from from my posts.

    Seriously. I've made my point. You are welcome to disagree with my actual point.

    Leave a comment:


  • Mohoney
    replied
    Originally posted by TDog View Post

    It sounds like the statistical analysts are living in an alternate reality. Maybe I'm in the alternate reality where it's about winning games as opposed to a geometric-style proof that that shows the games will be, and later, have been won.

    Issues notwithstanding, predictions are pointless.
    By this logic, baseball message boards are also pointless, because none of us (as far as I know) are credentialed employees in MLB front offices.

    Leave a comment:


  • vegandork
    replied
    Originally posted by Dumpjerry View Post

    What is most impressive about this is the fact that is was before they got Tom Brady.

    Leave a comment:


  • Dumpjerry
    replied
    Originally posted by Mohoney View Post

    I would imagine that Tampa Bay's massive jump from 66 wins in 2007 to 97 wins in 2008 factored heavily into this.
    What is most impressive about this is the fact that is was before they got Tom Brady.

    Leave a comment:


  • vegandork
    replied
    Originally posted by rdivaldi View Post

    No. Here is a more simple analysis from 2017, PECOTA again lost.

    https://www.beyondtheboxscore.com/20...systems-graded
    I'll take a look at their yearlies. But before that I might take a look at Marcel.

    Leave a comment:


  • vegandork
    replied
    Originally posted by rdivaldi View Post

    Frankly, I would rather hear your opinion on their rotation anyway. I love statistics and reading into the numbers, but in the end I take it with a grain of salt. If predicting players via math was easy, there would be little fun in playing.
    Agreed. There are uses for both, but generally stats are better for past performance.

    For Quantrill - he's got a starters mix of pitches, and has solid command. The big issue in the minors and even in his debut season was his breaking stuff wasn't developing. His fastball has nice sink and he can mix it nicely with the change. That two pitch mix is why a lot of people thought he was a better relief option. But when 2020 began, the curve started to click. He's commanding it well and it's probably his best pitch now. I was really surprised San Diego traded him. He still has work to do on the slider, but he doesn't need it. The mix he has now, he could be a very effective back end starter that can pitch at a high level. And there's room for a lot more.

    McKenzie is just the type of pitcher I like. Old school rising fastball, and he uses it to challenge guys. That and some really wicked curveballs. His changeup is designed perfectly to complement his fastball, as it breaks in the other direction (down and in to lefties, while his fastball rises up and in to righties). He has good control with all three pitches, all of which are plus, but his command of the zone with them is something that even veterans are not as good at. It isn't always the guy with the best stuff that has the most success. McKenzie is the guy with very good stuff but the ability to do whatever he wants with it. He could be scary good.
    Last edited by vegandork; 02-12-2021, 02:46 PM.

    Leave a comment:


  • rdivaldi
    replied
    Originally posted by vegandork View Post

    One year. I suggest reading Nate Silvers prediction roundups. It's consistently PECOTA.
    No. Here is a more simple analysis from 2017, PECOTA again lost.

    https://www.beyondtheboxscore.com/20...systems-graded

    Leave a comment:


  • rdivaldi
    replied
    Originally posted by vegandork View Post

    You should always post as much evidence as there is.

    Personally, I trust PECOTA because their models, unlike others are not solely based on previous performance. They look at past production and skills of the player to build comparables, look at those comparables and narrow it down further to where they are in their development. Which is why PECOTA is so much better at predicting young players' rise (and older players' regressions).

    FWIW, I did not make my assumption on the Indian's rotation based on PECOTA. But obviously PECOTA sees what I see.
    Frankly, I would rather hear your opinion on their rotation anyway. I love statistics and reading into the numbers, but in the end I take it with a grain of salt. If predicting players via math was easy, there would be little fun in playing.

    Leave a comment:


  • vegandork
    replied
    Originally posted by rdivaldi View Post

    So, should everyone just start posting what other prediction models are saying about the Indians rotation?

    ZiPS?
    Bieber 14- 9, 1.08, 3.51
    Plesac 9- 9, 1.28, 4.48
    Civale 9- 9, 1.30, 4.75
    Quantril 7- 8, 1.41, 4.98
    McKenzie 3- 4, 1.34, 4.93

    Giolito 16- 7, 1.03, 3.00
    Lynn 12- 8, 1.25, 3.97
    Keuchel 12- 8, 1.26, 3.99
    Cease 10- 9, 1.44, 4.48
    Lopez 10- 9, 1.36, 4.68

    Steamer?

    Bieber 14- 9, 1.08, 3.26
    Plesac 9- 10, 1.33, 4.71
    Civale 8- 9, 1.33, 4.67
    Quantril 6- 7, 1.38, 4.88
    McKenzie 8- 9, 1.32, 4.65

    Giolito 11- 10, 1.25, 4.08
    Lynn 12- 11, 1.27, 4.26
    Keuchel 11- 11, 1.44, 4.51
    Cease 8- 9, 1.35, 4.72
    Kopech 6- 6, 1.35, 4.57

    In short, basing your entire argument around a pre-season prediction model is foolish.
    You should always post as much evidence as there is.

    Personally, I trust PECOTA because their models, unlike others are not solely based on previous performance. They look at past production and skills of the player to build comparables, look at those comparables and narrow it down further to where they are in their development. Which is why PECOTA is so much better at predicting young players' rise (and older players' regressions).

    FWIW, I did not make my assumption on the Indian's rotation based on PECOTA. But obviously PECOTA sees what I see.

    EDIT 2: Why does Steamer hate Giolito?

    Leave a comment:


  • vegandork
    replied
    Originally posted by rdivaldi View Post

    What? Not even close, PECOTA has never shown to be more accurate than the other projection models. Read this analysis from 2019, PECOTA was actually the worst of the prediction models:

    https://www.getbigboard.com/harper-w...19-projections
    One year. I suggest reading Nate Silvers prediction roundups. It's consistently PECOTA.

    Leave a comment:


  • rdivaldi
    replied
    Originally posted by vegandork View Post
    PECOTA has been shown time and again to be the best projection system on offense by a long margin, and equal to the best projection models for pitchers.
    What? Not even close, PECOTA has never shown to be more accurate than the other projection models. Read this analysis from 2019, PECOTA was actually the worst of the prediction models:

    https://www.getbigboard.com/harper-w...19-projections

    Leave a comment:


  • rdivaldi
    replied
    Originally posted by vegandork View Post

    I don't understand how a projection model mimics my point, but I'm the obsessed one.

    You might think Keuchel and Lynn are great for 2020, and a lot of people would agree. I'd probably agree, at least on Lynn. But your argument that guys like Civale, McKenzie, Pleasac, and even Quantrill aren't right there with them (and actually predicted to be better) is NOT supported by PECOTA for 2020. Which is probably the last year it'll be close given the ages of the former.
    So, should everyone just start posting what other prediction models are saying about the Indians rotation?

    ZiPS?
    Bieber 14- 9, 1.08, 3.51
    Plesac 9- 9, 1.28, 4.48
    Civale 9- 9, 1.30, 4.75
    Quantril 7- 8, 1.41, 4.98
    McKenzie 3- 4, 1.34, 4.93

    Giolito 16- 7, 1.03, 3.00
    Lynn 12- 8, 1.25, 3.97
    Keuchel 12- 8, 1.26, 3.99
    Cease 10- 9, 1.44, 4.48
    Lopez 10- 9, 1.36, 4.68

    Steamer?

    Bieber 14- 9, 1.08, 3.26
    Plesac 9- 10, 1.33, 4.71
    Civale 8- 9, 1.33, 4.67
    Quantril 6- 7, 1.38, 4.88
    McKenzie 8- 9, 1.32, 4.65

    Giolito 11- 10, 1.25, 4.08
    Lynn 12- 11, 1.27, 4.26
    Keuchel 11- 11, 1.44, 4.51
    Cease 8- 9, 1.35, 4.72
    Kopech 6- 6, 1.35, 4.57

    In short, basing your entire argument around a pre-season prediction model is foolish.

    Leave a comment:


  • vegandork
    replied
    Originally posted by FielderJones View Post

    https://en.wikipedia.org/wiki/PECOTA#Accuracy

    I haven't been able to find a statistical analysis of PECOTA team wins average error for anything beyond 2008, but geez. 8.5 wins is the difference between mediocrity and division champion.
    With this kind of model, it's less about overall error and more about how many teams are within a standard deviation. I don't know how well it does in keeping most teams within that range, but I suspect not all that well. Team projection is fun, but it's not really what they do.

    PECOTA has been shown time and again to be the best projection system on offense by a long margin, and equal to the best projection models for pitchers. The model itself is not meant for team performance projection. They actually just take their individual projections, assign playing time, come up with total runs allowed and runs scored for each time, and then apply Bill James Pythagorean wins calculation.

    Leave a comment:


  • Mohoney
    replied
    Originally posted by FielderJones View Post

    https://en.wikipedia.org/wiki/PECOTA#Accuracy

    I haven't been able to find a statistical analysis of PECOTA team wins average error for anything beyond 2008, but geez. 8.5 wins is the difference between mediocrity and division champion.
    I would imagine that Tampa Bay's massive jump from 66 wins in 2007 to 97 wins in 2008 factored heavily into this.

    Leave a comment:

Working...
X