Sep
21

2009 Dominance Factors

By

Just before the season started I introduced you a cool new stat for minor league pitchers called Dominance Factor. Patrick Sullivan at Project Prospect developed the stat last May (then tweaked it this January) as a way to see which pitchers were really “dominating” the level they were playing at when you considered their age. It’s based on three factors, all of which the pitcher can control (to varying degrees): strikeouts, walks, and groundballs. The formula is very simple and straight forward:

Dominance Factor, DF = (K% + 0.72*GB% – BB%)+ (Age Level Standard – Actual age)*7

GB% is multiplied by 0.72 because generally speaking, 72% of groundballs turn into outs. The Age Level Standards are basically the average age at a given level, and are 20-yrs old for Low-A, 21 for High-A, 22.5 for Double-A, and 24 for Triple-A. If you want to see an example, click on the first link. The stat doesn’t have any real analysis purposes because of the assumptions used for GB% and age, so it’s best used for reference. It’s still fun to look at, though.

This year’s leader in DF was none other than RAB’s Minor League Pitcher of the Year Manny Banuelos, who checked in at 65.0 DF. That, however, doesn’t even come close to Phil Hughes‘ record of 85.98 DF with Triple-A Scranton in 2007. Mark Melancon was a close second (63.9 DF) because of his exceptional walk and groundball rates, and Zach McAllister (58.6) rounded out the top three. Andrew Brackman checks in at a disappointing 24.1 DF because he was three years old for his level and his walk rate was through the roof. You’ll see that the bottom of the leaderboard is mostly filled with journeyman and low level relievers, organizational arms like that.

The full table of Dominance Factors is huge, so it’s hidden after the jump. Players highlighted in yellow are no longer with the organization, and players highlighted in blue are eligible for the Rule 5 Draft this December unless added to the 40-man roster. I think I got everyone, but it’s tough to figure out when some of the international guys actually signed, so someone like Hector Noesi might actually be R5 eligible even though I didn’t highlight him. My bad if so.

All of the K, BB, and GB data come from the wonderful site First Inning, and I rounded the player’s age to whatever age they spent the majority of the year at. It’s also broken down by level, so you’ll see a DF for Pat Venditte with Low-A Charleston and then another for High-A Tampa. I could have combined them, but eh, I figured seeing them separate was more useful. I also made it a minimum of 25 IP at a level, so Ian Kennedy and his 22.2 IP at Triple-A Scranton didn’t make the cut. If you’re interested, here’s the DF’s for 2008, 2007, and 2006.

Remember to click the table for a larger view, but I don’t think you’ll have any trouble reading this one.

2009 Dominance Factors

Categories : Analysis, Minors

80 Comments»

  1. Bryan says:

    When are the Yanks going to take Venditte seriosly? And is it possible he sees AA sometime next year?

    • JohnnyC says:

      They do take him seriously. After all, they drafted him twice. But, that said, he’s still a fringe-y prospect with below-average stuff. Taking it a level at a time with him seems the prudent thing to do. He may ultimately be another AAAA guy.

    • MikeD says:

      He’s 23 and in A ball. If he can replicate at AA, then they’ll start to take him more seriously. Not sure why they haven’t promoted him yet, though. Nothing left to prove in A ball, but they probably see some things they want him to work on before advancing.

      He’s a good story. Remains to be seen if he’s anything more than a good story.

      • NC Saint says:

        Yeah, but they have to promote him for that to happen. If someone is way too old for his level *and* blowing away the competition, he’s not being taken very seriously.

  2. Accent Shallow says:

    This stat certainly matches my perception that there were no huge pitching performances from the farm this year.

    • Anther - Troy says:

      Depends on you view I guess or definition of the word huge. Banuelos 65 score would have been good for third place 2008 and 2007 and second 2006.

  3. Bob Stone says:

    Question – Why is (Standard Age – Actual Age) multiplied times seven?

    • alex gonzalez says:

      it is an arbitrary number for the sake of trying to say that older players shouldnt be in the lower levels. if they are it is more likely that they will pitch better against younger competition. i dont really know for sure.

      • Bob Stone says:

        It’s obviously a weighting factor that penalizes older players and rewards younger players. I just wondered why it was 7 instead of 3 or 6 or 8 or any other number.

      • The next step with this quality metric is to start tracking correlation of dominance factor to both reaching and succeeding in the majors.

        What level of DF tends to forecast ML success? 70 and above? What level forecasts the inability to get big league hitters out? 45 and below?

        Also, is DF the best forecaster of ML success, or does a different metric show higher correlation?

        Sullivan’s work is great, though. Quality stuff.

      • pete says:

        i get really confused when you say logical things like this, ag

    • Ed says:

      I think it’s just a made up number that seems to work ok. It was originally 10, then revised to 7 as that seemed more accurate.

      If there was real research put into it, it wasn’t explained.

    • Andy In Sunny Daytona says:

      Dog years?

  4. pat says:

    Incredible how Nova goes from given away for 25k to most likely a MLB contributor next year.

    • Rick in Boston says:

      I hate delving into the psychology of players, since it’s complete speculation and we have no clue, but Nova’s getting returned to the Yankees may have been the kick in the butt he needed to put it together. He’s had the stuff but never the results. So maybe something clicked.

      • pat says:

        Same with Tabata. He certainly looked like a train wreck in progress while still in our organization. Was he just suffering through some growing pains that would have fixed themselves or was getting traded to PIT the kick in the butt he needed? We’ll never know.

        • Rick in Boston says:

          I think getting away from his crazy, baby-stealing wife was what got Tabata hitting again. But really, some guys just need a change of scenery to get it together. I hope Nova keeps it up in AAA next year – some more ground balls would definitely be awesome.

          • pat says:

            More GB’s would be very nice. He held his own as a 22 yr old in AAA that’s not too shabby. I forget he’s that young.

            • King of Fruitless Hypotheticals says:

              that’s what i was going to say–damn he’s young.

              hope to see him compete for meat-tray’s 11th starter spot next year…

    • Ed says:

      That’s what’s supposed to happen with the Rule 5 draft. Being taken in the Rule 5 draft means you’ve got talent, but your team didn’t have room for you. It’s not a bad thing.

  5. Seeing Romulo Sanchez’s name in blue scares me.

  6. Tom Zig says:

    Kei Igawa is Rule 5 Eligible?

    Someone please take him!

  7. CB says:

    This dominance factor is a very good example of how not to develop a metric. It’s largely ad hoc, justifies none of it’s implicit assumptions, and has no clear conceptual foundation.

    This isn’t the kind of work that a statistician or methodologist in metric construction would do. This is the kind of work one sees from a bad management consultant who has mastered the art of microsoft excel or access.

    And unfortunately, this kind of shoddy work happens far too often in the quantitative analysis of baseball.

    • pat says:

      What would you do to improve it?

      • CB says:

        You’d have to start by rigorously and explicitly defining what the “dominance” is and doing so based on validated models for the concept.

        When you are trying to measure a construct that is inherently abstract or even subjective then the conceptual foundation of that construct is the critical step in creating anything valid.

        And that’s what seems to be rarely done in baseball analytics.

    • Do you have any examples of well-developed metrics? Which metrics do you like and think are good with clear conceptual foundations, etc.?

      • CB says:

        tRA is relatively well defined conceptually and analytically. They did a very good job of relating a concept and seeing it through to creating a metric.

        Conversely, many “traditional” baseball statistics e.g. batting average, on base percentage, etc. are also well defined and constructed.

        “Earned run average” on the flip side is a traditional baseball stat that isn’t rigorously constructed because the concept of “earned” wasn’t thought through or made explicit enough (though that doesn’t make ERA useless, just limits is).

      • Jerkface says:

        RBI kind of owns.

        • Tom Zig says:

          So do wins

        • Jerkface says:

          I guess RBI is less of a metric, more of just a stat.

          So I have just now created xRBI, which removes any RBIs not gained in a pennant race so that we can accurately nominate the MVP. It also doubles important RBIs.

          Jeter now leads Teixeira 241 xRBI to 235

          Mauer has -10 xRBI

          xRBI owns.

          • Tom Zig says:

            How many does A-rod have?

          • CB says:

            RBI is very clearly constructed but it’s problem is that people too often don’t acknowledge it’s limitations (or conversely throw it away as meaningless because it’s “only” a team stat… which would imply the guy up at bat isn’t part of the team).”

            “Traditionalists” have taken an RBI to conceptually represent something closer to a “run created” which it is not. Nor does it’s foundation imply that it is.

            A similar thing is starting to happen to FIP. It’s no longer representing the narrow concept that it is supposed to measure and is given import it doesn’t have.

            • TheLastClown says:

              Would you mind expounding on that last bit about FIP?

              Do you mean that the construction of the metric itself has somehow changed, or that the evolving uses & interpretations over-reach the original intended measurement?

              • CB says:

                It’s an issue of it’s evolution and related scope creep.

                Forget the metric – that’s secondary. If you go back to the fundamental underlying conceptual construct that FIP is built on that construct is an extremely narrow one.

                But FIP is now used in a much broader way in many common parlances than what the original construct is supposed to represent.

                FIP is commonly used and discussed as a counter point to EqA or wOBA, only for pitching. As such that’s why you can base pitching WAR on FIP.

                So in essense, FIP is commonly used to represent something that is much closer to what tRA is actually doing.

                Part of the problem is that the enormous implicit assumptions in the underlying model for FIP aren’t made explicit enough.

                Do you believe that a batted ball struck off of Roy Halladay has the same probability of being a base hit as a batter bat hit off of Sergio Mitre?

                FIP makes that assumption.

                • TheLastClown says:

                  Does tRA adjust the event probability relative to individual expectations?

                • CB says:

                  Somewhat but not fully. What it does is at least take into consideration the distribution of batted balls in play.

                  If there’s one unifying theme of “advanced metrics” I’d say it’s the attempt to strip away a particular type of context in order to reveal true “talent.” But there always compromises one makes in such an endeavor and often one arrives at something that is quantitatively pure but conceptually or substantively flawed.

                  So tRA assumes that a line drive hit of Roy Halladay has the same probability of being a base hit as a line drive hit off of Sergio Mitre.

                  There’s always assumptions you have to make like that when you try to take away some kind of context.

                  And that’s made worse if the models one deals with are limited ones.

  8. kenthadley says:

    CB….you present your arguements very well, and make an interesting and convincing case….if I had an idea of what all those letters meant, I’d probably be even more impressed…..I need a degree in statistics to follow RAB today…..

  9. [...] feeds in nicely with the discussion in the Dominance Factors thread. The purpose of WPA is not to evaluate a player’s performance. It is to show the ebbs and [...]

  10. NC Saint says:

    Venditte had to go and appear on this chart twice. That guy is a living Doublemint gum commercial.

  11. [...] week I posted the 2009 Pitching Dominance Factors for the Yankees’ farm system, and yesterday Greg Fertel at Pending Pinstripes took a stab at [...]

  12. [...] River Avenue Blues took a look at all of the minor league pitchers through the lens of a statistic called Dominance Factor. Brett Sullivan developed this statistic over at Project Prospect. The formula is [...]

  13. [...] are the 2009, 2008, 2007, and 2006 DF’s. The full 2010 chart is after the jump, but there’s a few [...]

Leave a Reply

You may use <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong> in your comment.

If this is your first time commenting on River Ave. Blues, please review the RAB Commenter Guidelines. Login for commenting features. Register for RAB.