What is tRA?

As I’m sure many of you have noticed, tRA has come to FanGraphs. I’m going to try to describe it as concisely as possible, so here goes.

tRA can be seen as an extension of FIP including batted ball types, namely line drives, ground balls, and the different types of flyballs. The idea of using these is to separate defense from pitching while still incorporating some measurement of how ‘fieldable’ the contact a pitcher induces is. Line drives are a little trickier for defenders to handle than ground balls, and tend to lead to more runs scoring. This is reflected in the runs/outs data we have for batted ball types, which leads the way quite neatly to tRA.

If you aggregate the tRA outcomes (K, BB, HBP, HR, + batted balls), and apply run/out values, you end up with expected runs (xR) and expected outs (xO). We can easily convert this to runs per nine innings by taking xR/xO*27. That’s tRA. Note that it is not on the familiar ERA scale, as I believe a defensive neutral statistic should expect defenders to have a league average error rate. League average tRA is typically in the high 4s.

Why use tRA? Well, it’s an interesting tool to supplement FIP with if you want to look at how hard a pitcher is being hit. It’s not a FIP killer by any means, and the difference between StatCorner tRA (using MLB AM classifications) and FanGraphs tRA (using BIS classifications) should tell you why: batted ball types are pretty subjective. However, they’re not wildly inaccurate, and using tRA, especially alongside pitcher batted ball information will give a better understanding of what exactly a pitcher is doing.

I’m very pleased tRA has made its way to FanGraphs, and I can’t say enough thanks to David Appelman for making it possible. I hope everyone finds it useful.




Print This Post





28 Responses to “What is tRA?”

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

    Quick question mostly from a fantasy perspective of a guy who has enjoyed a lot of success relying on FIP. Should a large difference be expected? and what is considered dominate in terms of tRA? For example Beckett has been very good and his FIP and tRA are about a whole run different. Does this then mean Beckett has not been as dominant or is his 4+ tRA actually really good?

    Thanks for the help.

    Vote -1 Vote +1

    • Michael says:

      Part of it may be that FIP and tRA are in different scales, so they’re not comparable to each other. FIP/.92 and tRA are more comparable. Also, pitchers who give up line drives will get punished more I think in tRA, so high line drive totals can drive tRA up, like in the case of Beckett last season.

      Vote -1 Vote +1

    • Jack Moore says:

      2 quick easy translations from tRA to tERA (tRA on the ERA scale)

      tERA = .92*tRA
      or even easier
      tERA = tRA – .35, but this breaks down at the extremes.

      Basically, 92% of runs allowed are earned, and that tends to be around .35 runs in the case of most pitchers.

      Vote -1 Vote +1

  2. Jake says:

    And now Fangraphs OFFICIALLY is the only place you ever have to go for baseball stats ever.

    Vote -1 Vote +1

  3. marc w. says:

    Congrats to Graham and Matthew.
    And thanks again to David for putting it up here.

    Eyeballing some of the differences, it appears that the differences in park factors may be a lot higher than I’d have thought. Colorado pitchers like Ubaldo Jimenez and Aaron Cook have ~1.0 difference in their statcorner and fangraphs tRAs. How do you do the park factors here (I guess this is to David)?

    Vote -1 Vote +1

    • We have Colorado with a 109 park factor, but it’s my understanding that statcorner does component park factors on tRA and not just a single runs park factor? This could be a big difference.

      I also think that there may serious discrepancies in the IFFB stat that Gameday reports and BIS reports, by as much as 5%-10%/TBF. This however requires some further investigation, so don’t take the above sentence as gospel or anything.

      Vote -1 Vote +1

      • marc w. says:

        Right; I’d initially thought that the discrepancies between Gameday/BIS would account for much of the difference. But – and I’m just eyeballing these right now – the larger differences in places like Colorado and Arizona make me wonder. A smaller variance in both batted-ball classification AND some park factor weirdness might account for it.

        Vote -1 Vote +1

      • Nick says:

        You should really consider using component park factors. A lot of things like LD rate are affected heavily by ballpark for whatever reason. Just ask Brian Cartwright.

        Vote -1 Vote +1

  4. Ed Nelson says:

    What more can Fangraphs do to be crowned the absolute king of baseball stats and analysis?

    Bravo.

    Vote -1 Vote +1

    • Dan Novick says:

      Maybe have some crazy splits section and be able to select games from the gamelogs like you can on B-Ref. Splits and gamelogs are pretty much the only thing I go to B-Ref for.

      Vote -1 Vote +1

      • Nick says:

        Same.

        Vote -1 Vote +1

      • Dan, just curious which game logs you’re talking about, because we do have game logs.

        Vote -1 Vote +1

      • TCQ says:

        David,

        Don’t mean to speak for Dan, but I believe he’s talking about how B-R has the cumulative stats listed after each game. So if I look at, say, Yunel Escobar’s game log for June 30th, 2009(on Baseball-Reference) I can see that his OPS after that game was .777. If I then look at the game log for July 24th, I can see that Escobar upped his OPS to .856. So I can contrast those stats, which I like to do from time to time. You guys do have game logs(which include WPA and pLI, which is awesome), but they only list the events of that game; no overall stats.

        Or Dan might be talking about something else entirely. But that’s why look at B-R’s game logs, at least.

        Vote -1 Vote +1

  5. Steve says:

    Is there any reason that you guys don’t post averages for tRA and things like LD%, FB%, etc like you do for OBP/SLG/wOBA?

    Vote -1 Vote +1

  6. NadavT says:

    This is fantastic news!

    Has it been explained somewhere why Fangraphs presents FIP instead of xFIP? If so, I must have missed it.

    Vote -1 Vote +1

  7. Aaron B. says:

    “It’s not a FIP killer by any means…”

    THANK YOU so much Graham. This really needed to be said.

    Vote -1 Vote +1

  8. metty5 says:

    Marc,

    If tRA uses batted balled data, how accurate is the stat corner version for minor leaguers?

    Minor league batted ball data for LD can be very iffy.

    Vote -1 Vote +1

  9. rob m says:

    Where are the tRA stats located?

    Vote -1 Vote +1

  10. Ender says:

    I prefer xFIP to FIP in the first place, HR/FB% is just so darn random year to year that I can’t believe it is fully under the pitchers control and it seems to make xFIP much more consistent year to year than FIP.

    Have you guys ever thought of adding xFIP?

    Vote -1 Vote +1

    • Ender says:

      Braden Looper is an example of what I’m talking about. The past 4 years his basic stats have remained relatively steady. The big difference is that in 2006 he gave up a lot fewer HR than normal and so far in 2009 he has given up a lot more than normal.

      This has caused a huge gap in his 2006 vs 2009 FIP(3.46-5.84), tRA(4.04-6.33) and ERA(3.56-4.99) that probably doesn’t really pinpoint the underlying skill level very well.

      His xFIP over that period has been steady though with a 4.33-4.93 range. I just feel xFIP shows true talent a lot better and FIP, ERA, tRA are much more results oriented which has its uses of course.

      Vote -1 Vote +1

  11. Brooksy Boy says:

    Are you THE Ender? Like Ender Wiggin?

    By the way I also use B-R for getting stats on the entire league, especially their pitch data (Strike Swinging %, 1st Strike %, etc.)

    Vote -1 Vote +1

  12. Is there a difference between xFIP and FIP?

    Vote -1 Vote +1