Rich Harden and Pitch Counts

Entering last night’s start against the Oakland Athletics, Rich Harden had been largely ineffective in his new role as the ace of the Texas Rangers pitching staff. Through his first five starts, Harden had only pitched 23.2 innings with a 4.56 ERA and an atrocious 6.97 FIP, thanks in large part to an 8.75 BB/9. As that suggests, Harden has been throwing a ton of pitches this season – he had thrown at least 87 pitches in every start so far despite only throwing more than 4.1 innings twice.

As Harden is a high-strikeout, high-walk pitcher, it’s unsurprising that he typically runs high pitch counts. Harden’s calling card with the Cubs last year were brilliant, 10 strikeout games in which he would be forced to leave after five innings and 110 pitches thrown. Also, given his injury history, pitch counts are often at the front of his manager and GM’s mind during each of his starts.

Harden’s pitch counts have been highly correlated with his success through his career so far. Below is a chart of average pitches per inning on a per-game level plotted against game WPA.

We see a strong correlation, and, in particular, a striking trend: if Harden can keep his pitches per inning below 16, he will almost certainly enjoy some level of success in his starts. He’s only run a negative WPA three times in 61 career starts in which he’s thrown 16 pitches per inning or fewer. The results are less conclusive as Harden’s average pitches per inning increases until his average pitches per inning eclipses 23. After that point, 75% (12 of 16) of his starts have resulted in negative WPAs.

Four of Harden’s first five starts resulted in an average pitches per inning above 17, with an average WPA of -.126. The other – his most recent start, prior to last night – saw Harden throw an average of 14.5 pitches per inning, right on the threshold where Harden is guaranteed to be above average. By WPA, Harden was +.056 in that start, despite 5 walks and only 2 strikeouts – not the most effective start, but he managed to record 8 ground ball outs, which were key in escaping with only three runs allowed in six innings.

Harden’s start on Monday was his best yet and by far. Harden went 7 innings and allowed no runs, striking out nine and walking none. He threw 108 pitches in the start, for an average of 15.4 per inning. That’s right in the range that Harden should be shooting for – as a strikeout pitcher who also gives up a lot of fly balls, Harden should look to avoid contact, and as such he will encounter many at bats of multiple pitches.

Of course, throwing fewer pitches generally means facing fewer hitters, leading to fewer runs. With Harden, however, his pitch counts will remain an issue for reasons other than his effectiveness. Due to his style of pitching, more pitches generally means more walks, which will kill any fly ball pitcher. Secondly, given his injury history, it is certainly in the Rangers best interest to keep Harden’s pitch count low. If he’s struggling to get through 6 innings in less than 100 pitches, keeping Harden’s arm out from under duress will mean that he’s not giving them enough innings to provide value in the first place.

Monday night’s start was a great sign on nearly every front for the Rangers and Rich Harden. If he can remain as efficient as he was in that start while maintaining the knockout stuff that allows for nine strikeouts in seven innings, he will be a key piece in a team that is emerging in a tight AL West.

Print This Post

Jack Moore's work can be seen at VICE Sports and anywhere else you're willing to pay him to write. Buy his e-book.

20 Responses to “Rich Harden and Pitch Counts”

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

    This is a really interesting concept, although it doesn’t really mean anything to me without context. What does the average pitcher’s WPA vs pitch count look like? I imagine it wouldn’t be all that different…

    Vote -1 Vote +1

    • AK707 says:

      Duh. The only way you are throwing 25 pitches per inning is if you are giving up baserunners. Baserunners generally = bad. If you are only throwing 15 pitches per inning, you probably aren’t giving up those runners. Maybe pitches/batters faced would show something more insightful? I doubt it will be, but it might.

      Vote -1 Vote +1

  2. ryco218 says:

    Another thing to take note of with regards to Harden (at least this year), is that his velocity has fallen off. Given his repertoire, this is disconcerting. Which brings me to my question…

    Is there a game-by-game log on this website of velocity for pitchers? I know that Harden’s velocity has been way down this year, but looking at his velocity chart, the last plotted point (I assume last night’s start) spikes to his career norms. I can only assume this is heavily correlated to his 9 K’s. I’d like to be able to view this in terms of a mph number instead of estimating based on the charts. Is this possible?

    Vote -1 Vote +1

  3. Red Sox Talk says:

    Isn’t this mostly selection bias? Any pitcher will use fewer pitches in a good inning, thus the correlation should always be pretty high. I think the key is compare Harden’s curve with other pitchers or the average starter.

    +8 Vote -1 Vote +1

  4. jfpbookworm says:

    Am I missing something? Of course fewer pitches per inning correlates with a higher WPA. Fewer pitches per inning correlates with fewer batters faced per inning, which means that the pitcher is getting more of those batters out. Saying “if Harden can keep his pitches per inning below 16, he will almost certainly enjoy some level of success in his starts” is saying “if Harden can get more batters out, he will be more successful.” Well, yeah.

    How does Harden’s WPA correlate with pitches per batter faced?

    Vote -1 Vote +1

    • Erik says:

      “How does Harden’s WPA correlate with pitches per batter faced?”


      Another interesting question: what is the difference between those 9 BB/9 IP outings and the 1BB/9 ones? Is he consistently running 3-0 and 3-1 counts instead of getting ahead of batters, or is he simply getting unlucky with pitches on 3-2 counts?

      Vote -1 Vote +1

      • DSMok1 says:

        This is what should be looked at. Pitchers per inning is pure selection bias.

        Vote -1 Vote +1

      • DSMok1 says:

        I just ran a quick regression of Harden’s WPA to his Pitches/batter faced.

        There is basically no correlation (R^2=.01)

        Vote -1 Vote +1

      • Erik says:

        A little digging on THT pulled up a few related articles:

        7/5/07: Should Scott Kazmir try to promote BIP over K to increase IP?

        8/27/07: What is an “efficient” pitcher? Are they preferable to “inefficient” pitchers? (part III of a series)

        Both of these articles conclude that pitch efficiency (i.e., pitches / PA) does not correlate with getting hitters out. This agrees with DSMok1’s analysis above.

        This is little more than rephrasing what I had written previously, but how about figuring why guys like Harden and Kazmir can be high K/IP guys with what seems to be unpredictable control that leads often leads to high BB/IP? Do they actually have bad control, or do they simply “nibble” too much leading to high BB rates? Do they throw an unusually high number of pitches within 2″ (or some other range) of the edge of the strike zone?

        Vote -1 Vote +1

  5. suicide squeeze says:

    It’s funny…I was thinking about this exact thing during the game. They showed a line score from a start he made against the rangers 3 years ago. He threw a two-hit shutout on a very Maddux-like 81 pitches. I miss that Rich Harden.

    Vote -1 Vote +1

  6. resolution says:

    What’s confusing to me is that while he’s never had the best control, his walk rates have really spiked the past two seasons. As he essentially only throws two pitches, this is pretty odd. If he can get back to a 3.6bb/9 rather than a 4.3, he should be fine. Has he been nibbling more?

    Vote -1 Vote +1

    • Zack says:

      Some starters can get away with just 2 (plus) pitches; AJ Burnett threw his FB or curve 94.1% of the time last year, been around that same ratio the last few years and he has success.

      But from watching him and Harden, if they dont have their top stuff, it’s usually a bad outing- where a pitcher who can throw 3-4 pitches can still put in a ‘quality start’ if one of his pitches is ineffective for a game.
      Think sinkerballers can be the same way, if they only have a sinker (like Wang) then they can be great when its on, but if its up they get hit around and cant make any adjustments because they have nowhere to go

      Vote -1 Vote +1

      • resolution says:

        I guess now that I think about it, his control should be better since I believe neither of them are true breaking balls… His changeup just does crazy stuff…

        Vote -1 Vote +1

  7. TheUnrepentantGunner says:

    This is poor again, Jack.

    As others have (rightfully!) pointed out, pretty much any pitcher that can get through innings quickly is going to do better. Correlation does not imply causation!

    In fact, the root cause is surely, that if you are throwing 30 pitches an inning, you have probably faced 6-7 batters, which probably means you have given up a run or three.

    Furthermore, even if you are pitching excellently, striking out person after person on a 3-2 count, you will probably only get through 5-6 innings at 20+ pitches an inning, which won’t add much to the wpa, versus a 9 inning gem.

    The next thing you are going to tell me is that throwing more balls than strikes isn’t a formula for success either.

    In short, at least Cameron has conviction, and is interesting, even when dreadfully wrong. Carson is entertaining, and geunine when he writes. 2/3 of your articles are dross.

    Vote -1 Vote +1

  8. Hank says:

    Is R^2 = .37 really a strong correlation?

    Vote -1 Vote +1

  9. Ritom says:

    Beyond the issues raised above, why is WPA being used here? Wouldn’t the amount of runs being scored by Harden’s team affect that number? So why not use a context-neutral stat (xFIP?) instead? Although it would probably end up being the same result given the selection bias raised previously.

    Vote -1 Vote +1

  10. eric says:

    Great correlation.

    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>