Pitch Type Linear Weights

You may have noticed a new section in the player pages called “Pitch Type Linear Weights”.

First of all a huge thanks goes out to Dave Allen and his tremendous work over at baseballanalysts.com. With his help, we’ve implemented his linear weights by pitch framework using the Baseball Info Solutions pitch type data.

What this section does is it uses linear weights by count and by event and breaks it down by each pitch type so you can see in runs the actual effectiveness of each pitch.

There are two stats for each pitch type. The first one is preceded by a “w” and this is the cumulative runs above average for a particular pitch type. Then there are the statistics trailed by “/C” which denotes runs above average per 100 pitches of that type.

This is kind of a first stab at the data and the weights we’re using are per an actual event, so singles, doubles, triples, etc… are weighted properly using data from 2005-2008. Behind the scenes we’re also calculating the data by batted ball type, so there’s always the possibility of displaying a “defense neutral” runs above average by pitch type as well.

Right now we’re only showing this in the pitcher’s stats pages and for the Baseball Info Solutions data. I expect over the next month that we’ll roll this out in the PitchFx sections and the leaderboards and maybe even for Plate Discipline stats.




Print This Post



David Appelman is the creator of FanGraphs.

19 Responses to “Pitch Type Linear Weights”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. Hero 3 says:

    So, when Felix Hernandez has a 5.8 wFB, does that mean he prevented, or allowed, 5.8 runs compared to average with his fastball?
    In other words, are higher or lower numbers better?

    Vote -1 Vote +1

  2. NadavT says:

    This looks amazing, but could you explain further what these stats mean? Do they merely break down a pitchers’ results by pitch type, or do they give the expected value of each pitch type based on its Pitch f/x data? In other words, does Felix’s value mean that we would expect him to prevent 5.8 runs for every 100 fastballs based on his movement, location, etc., or that he has been effective with his fastball so far this season?

    Thanks!

    Vote -1 Vote +1

    • These give the actual value of each pitch based on the count or the final result of the plate appearance.

      So here’s a quick example at-bat.

      Ball, FB – Starts at -.002, Ends at .032
      Strike, FB – Starts at .032, Ends at -.016
      Strike, SL – Starts at -.016, Ends at -.078
      Ball, SL – Starts at -.078, Ends at -.037
      Ball, SL – Starts at -.037, Ends at .057
      Fly Out, FB – Starts at .057, Ends at -.28

      If you take the difference between the start and the end values and then attribute them to the pitch type and then add them all up, that is what we’re calculating. (These are rounded values).

      Vote -1 Vote +1

      • NadavT says:

        I think I get it. So it takes the result of each pitch (ball/strike/in-play/etc.) and then assigns run values to it based on historical run values of similar situations, with “situations” meaning # of outs, # of men on, and count?

        If so, how would a “defense-neutral” version of this be different? Would it just be that in your example, “Fly Out, FB” would change to “Fly Ball, FB” with an appropriate adjustment to the change in run value based on the expected outcome of fly balls?

        Thanks for adding this awesome feature and for answering my questions!

        Vote -1 Vote +1

      • Mooooooooooooooose says:

        “These give the actual value of each pitch based on the count or the final result of the plate appearance.

        So here’s a quick example at-bat.

        Ball, FB – Starts at -.002, Ends at .032
        Strike, FB – Starts at .032, Ends at -.016
        Strike, SL – Starts at -.016, Ends at -.078
        Ball, SL – Starts at -.078, Ends at -.037
        Ball, SL – Starts at -.037, Ends at .057
        Fly Out, FB – Starts at .057, Ends at -.28

        If you take the difference between the start and the end values and then attribute them to the pitch type and then add them all up, that is what we’re calculating. (These are rounded values).”

        I didn’t understand this before and now I’m twice as confused! Somebody asked which numbers were better and you said “higher is always better for these stats.” Yet in this example, the strikes and fly out result in lower numbers and balls result in higher. Help me out here!!

        Vote -1 Vote +1

  3. Jake S. says:

    I’m blown away. I don’t know what I’d do without FanGraphs.

    Thank you Mr. Appelman, your work is very much appreciated.

    Vote -1 Vote +1

  4. radiosurgery says:

    Can you give us an idea of what terrible, average, and really good numbers might be for a pitch?

    Vote -1 Vote +1

  5. Matt Harms says:

    You guys are amazing. This is more evidence that BR and FanGraphs are probably the best two places for the stat-leaning and sabermetric community.

    Vote -1 Vote +1

  6. Rory says:

    This is great. I was wondering if FanGraphs could have a xBABIP column added to the players’ profiles. I’m referring to the formula put forth by Hardball Times http://www.hardballtimes.com/main/article/batters-and-babip/

    Vote -1 Vote +1

  7. Jake says:

    Fangraphs is not far away from officially being the only baseball statistics source ever needed for anything.

    Vote -1 Vote +1

    • Eric Cioe says:

      Totally agreed.

      This new feature is really, really cool. I’m still not entirely sure how it works, but I like it. One way to make it just a little cooler: make it a sortable category. We can see who throws the most fastballs, or who throws the hardest fastball, and now we’ll be able to see who saves the most runs with his fastball.

      Vote -1 Vote +1

  8. Justin says:

    Thanks for the stats.
    Does that mean we can expect more from D-train?
    I am just a beginner and wants to learn more about baseball.
    Thank you in advance.

    Vote -1 Vote +1

  9. Chris says:

    I’ve got a question, how much would a really poor (or good) defense effect these run values?

    Vote -1 Vote +1

  10. Xeifrank says:

    So if I add together the 2009 runs above average for each pitch type from a pitcher, how will that compare to the “Value” group of stats below that shows the pitchers RAR. For example, if I take Clayton Kershaw’s 2009 pitching stats, and add his pitch types up I get a total of 6 runs above average, and his RAR is listed as 8 runs above replacement. How to get these two stats to talk apples to apples, or is that impossible? Thanks.
    vr, Xei

    Vote -1 Vote +1

    • Well, there are certain things excluded in the pitch type stats such as IBB’s and there are always pitches that aren’t defined so they’re not going to match up exactly. Also, singles, doubles, and triples are counted in the pitch type stats so that is going to throw off the totals a bit.

      These are also above average and not above replacement, but they should match up fairly close if you remove whatever we give a pitcher as far as replacement goes.

      I think these stats are more to diagnose problems with a particular pitch or see how good a pitcher’s fastball or slider or whatever actually is than they are to measure a pitcher’s overall value.

      Vote -1 Vote +1

  11. drew says:

    Beyond awesome. Thanks David A!

    Check out Randy Johnsons slider. You won’t believe it if I told you.

    Vote -1 Vote +1

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>

Current day month ye@r *