Oh, the Dreaded Bias!

Two days ago, Dave Cameron wrote about the players Steamer and Zips disagree on the most, showing that Steamer is a little down on Jurickson Profar, Bryce Harper and a few others. In the comments, mattybobo pointed out that Steamer has atypically low BABIP projections for players with no MLB experience and that this looked like a mistake.

Well, I did some checking and, um, he’s right.

More specifically, Steamer was adjusting down the rate at which hitters get singles too aggressively when forming Major League Equivalencies (using what should be the adjustment factor for *all hits* rather than the adjustment factor for singles). Anyway, after making some adjustments, I checked our year to year MLE BABIPs for batters who moved up a level, batters who repeated a level and batters who made the jump the majors and can now feel comfortable standing behind our projections.

Players whose projections were based largely on minor league stats, of course, see the most benefit. Profar jumps up from .232/.301/.370 to .247/.315/.386. Bryce Harper’s move is less drastic, from .257/.329/.450 to .263/.335/.457. Both projections are still notably pessimistic compared to the other forecasters on Fangraphs (ZiPS, Oliver, James and the Fans), and this is always cause to be wary, but the gap has narrowed somewhat.

One last note, our adjusted system will still, to some extent, under-project players who go from the minors to a full MLB season. Historically, players who had more than 300 MLB PA following a season with 300 minor league PA’s over performed their Steamer BABIP MLEs. I’m okay with that. A Jurickson Profar who gets close to a full season’s worth of PA is more likely to be a Profar who has exceeded expectations than one who has fallen short.  This gets tricky though and it’s something we’ll need to revisit in the future with a little more data on hand.

24 thoughts on “Oh, the Dreaded Bias!”

  1. Jared, just got your e-mail. I’m trying not to be all giddy that I made a small contribution to the internet baseball geek-o-sphere.

    Oh hell, I’m gonna just go ahead and be giddy.

    Thanks for the hat tip!

  2. Jared, all of us over at VivaElBirdos appreciate the correction and all of the work you put into this. Thank you

    1. Though, it is a little dangerous… the adjustment in Steamer make me kinda sorta wish the Cardinals would just roll with a Greg Garcia Kolten Wong MIF and see what happens. That’s probably not a realistic hope.

  3. Thanks very much, Jared.

    Random question – it looks to me like PAs have changed since I last looked at the projections (on Fangraphs). Is that a result of this change or something else?

    What I am seeing is that most players have had their PAs decreased, although some at the very low end of the spectrum have been bumped up.

    Also – what is the difference between the PA totals here (the Razzball #s) and on Fangraphs?

    Much appreciated!

    1. I think I can answer this one: the PAs are different between here and Fangraphs because here uses Razzball playing-time projections, and Fangraphs uses the Fans’ playing-time projections. Rearrange those words and you can answer your second question too!

      1. Aha – that makes sense.

        Just to make sure I get it – in neither case is Steamer producing its OWN projection of playing time – Razzball and Fangraphs are taking Steamer ratios and using their own PA estimates – correct?

        Is there specific rationale as to why you prefer Razzball’s PA projections (as opposed to any others)?

        Awesome – thanks!

    2. Byron (below) is right and there’s another change as well. I examined last year’s Fan playing time projections and found that they were generally optimistic so, on Fangraphs, I started scaling down the PA projections to adjust for this optimism. I also made a bare minimum of 100 PA for hitters, 50 IP for starters and 30 IP for relievers so that there’s a comprehensible stat line for a much larger pool of players.

      1. Hi Jared,

        Are there any links to historical fan projections (either Fangraphs or Tango’s Community forecasts). I’d like to include them in an analysis I’m doing on projecting playing time? I’ve looked over fangraphs (and the book blog) and can’t find them.

        1. The 2012 results of Tango’s community forecasts are at http://www.tangotiger.net/survey/ . He probably has the rest. You should e-mail him. Or I’m pretty sure he has google alerts on his name, so maybe Tangotiger Tom Tango will be enough to bring his Sauron-like eye to here.

  4. hey J Cross…

    Any chance you can provide an updated downloadable spreadsheet to use with The Price Guide (like you did last year)?

    With all of the changes you’ve made over the last several weeks the projections on The Price Guide are all out dated.

    1. I adapted the CSVs at the top of the page to be used in the price guide and uploaded them. Click my name to access them. Note that the CSVs do not include projections in:
      For Hitters: Strikeouts, KAB, GIDP, E, A
      For Pitchers: Home Runs Allowed, HR/9, QS, Holds, Saves+Holds

    1. Right now just past year’s ERAs. I would like to pay attention to spring training ERA drops but right now we’re not set up to do.

  5. J. Cross — any view on whether modified Fans’ playing time projections (at Fangraphs) are likely to be better/worse than Razzball playing time projections?

  6. I didn’t see a “contact us” link or anything like that so I will ask my question here. Are Steamer projections updated during the season or is it strictly a pre-season projection? If they are updated, is it just updates to expected playing time or is performance (i.e. rate stats) updated as well? Thanks.

Leave a Reply

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>