Steamer Projections

Steamer Pitchers 2012 v 1.0

The initial run of the 2012 Steamer Pitcher Projections is now available for download. We’ll update our playing time projections more than once in the coming weeks and update our projections based on trades and signings. We’ll announce new downloads and new posts on our new twitter feed: @steamerpro.

We’re constantly working to perfect our system and find stats that are left unprojected and overlooked by the rest of the baseball world and the other projection systems so let us know what you’d like to see. Hopefully we can incorporate your ideas to further sculpt the system.

P.S. Hitter projections should be done within the week

Comments (39)

  1. Byron

    Wow, Steamer LOVES Strasburg for this year. Thanks for getting these out so early!

    Reply
  2. JRM

    Very fine. If you’re willing to talk about your methodology in some (not complete, but some) detail, I am reasonably sure it would be fascinating to those of us who have made it here to The Little Projections That Could. Jered Weaver (3.11 PECOTA, 3.16 ZiPS, 3.85 Steamer) is an awfully strong piece of evidence that you’re doing what Matt Swartz cited you as doing with velocity as a factor. (I think you are right to do so.)

    I realize that you don’t have the brand recognition that others do, but I (for one) would pay for these projections in 2013.

    Reply
  3. J. Cross (Post author)

    We definitely want to pick out the projections that differ the most from PECOTA, ZiPS and others and inspect where our systems diverge. I noticed that while both Oliver and PECOTA love Mr. Beachy, Steamer seems to love him the most. After we get the hitters out we’ll take some time to reflect on our results and our methods.

    Reply
  4. Jeremiah

    Great News! Thanks

    Reply
    1. Jeremiah

      Just a quick glance through and would love to see a Darvish projection!

      Reply
      1. Jeremiah

        One other quick note, Daniel Hudson seems to differ greatly from both Zips and Oliver. Very interesting!

        Reply
      2. J. Cross (Post author)

        I’m certainly curious about Darvish. We’ll dig up some stats and see what we get. Gotta get those hitters out first and thoroughly check for (more) bugs in the algorithm.

        Reply
  5. Mike

    Matt Moore is another one with some big differences vs other projections. You have him projected for nearly 6 BB/9! I imagine you must be going back 3 years and thus including his Single A wildness? I would think for young guys it makes sense to give them more credit for recent walk rates as control seems to be one thing that pitchers are able to make real improvements upon as they mature.

    Reply
    1. Brent Crossman

      To add on to this point. I know Clay Davenport has been working on different regression rates for different components (higher weights on the most recent data for strikeouts vs. walks, things like that).Have you guys looked into doing that? Pecota has also gone to a more flat weight between 3 year old data and 1 year old data. Just curious your guys thoughts on the benefits of these changes and whether you’re also making similar adjustments.

      Reply
  6. Joe P

    This is really interesting — Steamer does not like the Angels’ starting pitching staff of Haren, Weaver, and Wilson compared to any other system. The higher ERA projections are not due to an assumed higher run environment either as I look at the total projection of runs allowed, league-wide. Great stuff.

    Reply
  7. Sean

    What does SGP stand for in the projections? eg. Wins SGP

    Reply
    1. Dash

      SGP is shorthand for Standing Gains Points which is a way to convert player’s stats (i.e. wins, ks, era, etc.) into a quantifiable fantasy baseball specific output. So SGPs is basically a way to compare different players against each other that weighs their different stats on an equal footing, i.e. weighing saves against wins, ERA against Ks etc.

      A good explanation can be found here: http://www.hardballtimes.com/main/fantasy/article/my-approach-to-player-valuation/

      Reply
  8. J. Cross (Post author)

    Mike, thanks for the catch! There’s a bug and Matt Moore’s 2010 stats are being overweighted. He should have 81 BB instead of 104. We’ll put out a new version (1.01?) tomorrow.

    Brent, like Clay we have different weights for different components with some components with some things (like K-rate) depending more heavily on recent stats.

    Reply
    1. Brian

      Did the bug affect any other players besides Moore? I’m just curious if it’s a fault of the entire system that was only fixed for one player because it was so out of whack, or if it was an isolated incident.

      And thanks again for doing these, I’ve looked forward to them for the past few seasons. Keep up the good work!

      Reply
      1. J. Cross (Post author)

        Thanks!

        It did affect other players (really almost all players to some small extent) but we fixed it for everyone.

        Reply
        1. Brian

          Good to know, thank you! Something else I’m not sure was considered are the changes to Citi Field. The fences were moved in dramatically and the height was shortened.

          “footings have been poured and metal framing is going up by the day — framing that will bring in the fences of Citi Field some 13 feet in leftcenter, and as much as 15 to 17 feet in rightcenter – several feet more than had been originally planned … But perhaps an even more momentous change at Citi will be that the new wall will be eight feet high … 16-foot-high wall is history”

          Just another factor that could give you a big leg up over the other systems.

          Reply
  9. J Nichols

    First off, thanks for pushing the projection envelope and making them accessible. Your success is what we brought me here and I hope it continues.

    After reviewing the initial release, the name that really jumped out for me was the projected success for Bartolo Colon and his incredibly low 1.06 BB/9 (best in the league and half of his 2011 total of 2.19 and a third of his career 3.01). Is this simply the result of a small sample size or the Coliseum park effect? I also note that Steamer loves all A’s pitchers.

    Reply
  10. Aaron

    Daniel Bard and Holds.

    I love the fact that you guys try to project the much-maligned (and rightfully so) HOLD statistic. I play in a 6×6 and have a hell of a time finding holds projections. The bad news is that it doesn’t look like you’re doing a great job at it. I doubt you care because we’re talking about holds (not skill, not skill, holds), but you project Bard to lead with 26 HDs even though last yr there were 9 players with more than 26 HDs, and 25 players with over 20.

    The other thing is you are projecting Bard to pitch in 80 games and start 14 games…dealing with RPs making the conversion to SP sounds difficult but it looks like a glitch in the algorithm to me.

    Reply
    1. Dash

      That is very strange as I see the original version of our system has him projected to pitch in 43 games with 14 starts. That 80 must have come in somehow due to a manual error. Thanks for the catch!

      Reply
  11. Eric

    Jared, Thanks so much! What are the coefficients you used for SGP? Where did you get them from? Thanks!

    Reply
    1. J. Cross (Post author)

      Dash determined the denominators for hitters and we used the denominators published on lastplayerpicked.com for pitchers.

      Reply
  12. Hizouse

    Ah, I didn’t realize you had released these until just now! Thanks! A few quick comments:

    Josh Spence (SD) appears to be missing.
    Steamer expects great things from Padres prospects Erlin, Keyvius Sampson, and Bradley Boxberger. I know there’s a heavy park effect, but could there possibly be a bug in translating their minor league stats to Petco?
    Brandon Beachy’s last name is misspelled (no “e” before the “y”)
    I think you can safely delete Billy Wagner, but it is interesting to see what the system projects for him.
    Scott Linebrink should be SLN, not ATL.

    Thanks!

    Reply
    1. J. Cross (Post author)

      Thank you! I’ll make those changes including deleting Wagner since his projected line at this point isn’t based on many innings.

      We do seem to like Padres pitching. Maybe I’ll go back and look at whether we were too enthusiastic about Padres pitchers last year or whether we have that park factor figured about right.

      Reply
  13. chuckiT

    I’m trying to use your projections with The Last Player Picked tool at lastplayerpicked.com but in order to use Steamer with the tool (Roto Price Guide), the Roto Price Guide needs a field for “earned runs”. Any chance you can provide a pitchers projected earned run total so that I can get customized dollar values for my league? Or any advice on computer what a pitchers run total would be with the columns you’ve already provided?

    Reply
  14. chuckiT

    sorry for bad grammar: any chance I can compute a pitcher’s projected earned run total with the information already given in the downloadable spreadsheet?

    Reply
    1. Brian

      ER=ERA*IP/9. Just create a new column. LPP is also very finicky with having proper column headings and is case sensitive.

      Reply
  15. chuckiT

    Thanks brian

    Reply
  16. chuckiT

    i noticed there was no addison reed (young future closer) for the CWS. some believe he’s competing with thornton for CL job as soon as this year.

    Reply
  17. buckoneil

    Any theories on why you guys are so low on Matt Cain compared to other systems?

    Reply
    1. J. Cross (Post author)

      Looks to be a combinations of things but the big one is the difference in projected HR rate. Only 3.7% of fly balls hit against Cain last year went over the wall. The next lowest rate was 5.1% against Halladay. Steamer thinks his ballpark will help him but that we’ll still see big time regression here. Steamer is pessimistic about Daniel Hudson for pretty much the same reason.

      Reply
      1. Brian

        He’s 6.5% for his career so it’s not much of an outlier. Averages 17 HRA for his career. Career high was 22, and only had 9 last year. 25 seems a bit high, but for guys like him that buck the trend it’s impossible to fit it to a formula.

        Reply
  18. Matthew

    Looks like you guys have Brett Myers projected out as a starter. Any idea how his new role as a closer will affect his value?

    Thanks for your hard work in doing and releasing these projections!

    Reply
    1. J. Cross (Post author)

      Thanks. I’m going to get on to this over the weekend, I think. Pretty soon we should have updated playing time projections as well as a post about projecting playing time.

      Reply
  19. B. North

    Thanks for sharing. Just found these projections thanks to Fangraphs giving you some street cred. I look forward to taking a deeper look at them. Is there a page that describes your methodology in developing these pitcher projections?

    Reply
    1. J. Cross (Post author)

      Not yet, but that’s definitely something we should do. I know that Dash and Peter are thinking about writing up their work for a possible presentation in which case I’ll post their write up on the site.

      Reply
  20. Sean
    1. J. Cross (Post author)

      Yes. Updating playing time coming soon along with a handful of new players.

      Reply
      1. Bill

        will you be done this week? my draft is happening so I was hoping to get your greatest/ latest update

        Reply
        1. J. Cross (Post author)

          Should be Tuesday night or sooner.

          Reply

Leave a Comment

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>