Cliff Lee and Maxing Out His Innings

One of the benefits of never walking anyone is that you generally throw fewer pitches per batter and get to last deeper in games. With Cliff Lee running 1910-like walk rates his bullpen can figuratively safely go get smashed the night before any of his starts. Lee’s last start lasting just 6.1 innings was his shortest outing since equaling that total on May 21. Those two are the shortest he has worked all season long.

It’s not just that Lee has consistently gotten into the 7th inning, it’s that he’s consistently gotten into the 9th. 15 of his 20 starts so far have gone at least eight innings. Seven of those have gone a full nine. That sort of efficiency puts Cliff Lee in some rarefied air. With 161.1 innings pitched over those 20 starts, Lee is averaging 8.1 innings for every start the he makes.

I do not have historical numbers on how that exactly ranks, but I can use the slightly more general number of innings pitched per game to put Lee’s performance thus far into greater context. Since 1950 (a rather arbitrary year chosen to get past the early years and the World War II years), there have been just 49 pitchers to hurl at least 160 innings and average at least eight innings per appearance. Lee currently sits 40th on that list nestled between Tom Seaver’s 1973 season and Alex Kellner’s 1953 campaign.

Obviously, pitchers threw more innings the further back we go in history. Complete games used to be the norm instead of the exception. To get a sense of this one just needs to look at those 49 pitchers arranged in chronological order. Aside from Lee, the last pitcher to cross that eight-inning barrier was Greg Maddux in 1994. And before him, the last time it occurred was in 1983 when both Ron Guidry and Mario Soto did it.

Over the last 27 baseball seasons, only Greg Maddux has done what Cliff Lee is currently doing in terms of eating innings.




Print This Post



Matthew Carruth is a software engineer who has been fascinated with baseball statistics since age five. When not dissecting baseball, he is watching hockey or playing soccer.


32 Responses to “Cliff Lee and Maxing Out His Innings”

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

    I think his unwillingness to walk anybody has hurt his overall run prevention at the margins. Hitters are aware of it and have adjusted a bit from earlier in the season. Even the great Maddux would issue intentional “unintentional” walks once in a while to get around the best hitters in sticky situations.

    Vote -1 Vote +1

    • artie says:

      Run prevention doesn’t matter… if you don’t give up HR’s it doesn’t hurt FIP…. I think his WAR actually went up a good bit despite 4ER in 6.1IP because he had a lot of K’s, no walks and no HR’s allowed. You really can’t count runs that score via singles,doubles and triples as real runs… those can be impacted by defense and luck.

      You need to understand the new stats where giving up a double is better than walking someone or if you have a guy on 2nd, it’s worse to walk a guy then to give up an RBI single.

      Vote -1 Vote +1

      • DavidCEisen says:

        wow two offensive posts in a row.

        Vote -1 Vote +1

      • spindoctor says:

        Artie, giving up a double is not better than walking someone, and if you have a guy on second, giving up a single is not better than walking someone.

        As is mentioned below, striking a guy out OR getting the ball put in play is better than walking someone. We should not be talking about the results of the ball put in play.

        Vote -1 Vote +1

      • CircleChange11 says:

        You really can’t count runs that score via singles,doubles and triples as real runs… those can be impacted by defense and luck.

        They can also be affected by missing your location IN the strike zone.

        We need to get past the idea that “you can’t count those” as if a pitcher has no influence on hits and extra base hits.

        It’s not likely that a pitcher gives up a lot of extra base hits on good pitches where batters just get lucky and dump it “where they ain’t”.

        Extra base hits are often hard liners down the line or in the gaps off pitches that get far too much of the center of the plate.

        Vote -1 Vote +1

      • Kevin S. says:

        And batted ball types are something that’s accounted for in tERA. I’m a little surprised that doesn’t get used more, but we shouldn’t act like there aren’t appropriate stats for what you want.

        Vote -1 Vote +1

  2. Noodleshire says:

    I just noticed that Lee has more wins than walks and HR allowed this season…

    Christ is right.

    Vote -1 Vote +1

  3. DonCoburleone says:

    Yet another stat that shows Cliff Lee’s greatness this year (and last year, and 2008)… But it also points out the greatness that was Greg Maddux – hands down my favorite baseball player ever. Just take this in: Over 7 seasons from 1992-1998 spanning 226 Starts and over 1,650 innings pitched, Mad Dog allowed only 270 Walks against 1,250k’s with an ERA under 2.25!! His worst FIP of those 7 seasons? 2.85! His lowest WAR in those 7 seasons? 7.3! It doesn’t get any better than that. LONG LIVE MAD DOG!!!

    Vote -1 Vote +1

  4. KPS says:

    any stat that says walking someone when there is a runner on second is WORSE than giving up a single is clearly flawed. Walks aren’t rallies (from the offensive side), and there is only one situation where a walk leads to an allowed run. Walks are the most overrated thing in baseball.

    Vote -1 Vote +1

    • ACWNS09 says:

      You’re looking at it wrong, or at least you’re looking at it differently than FIP does. FIP doesn’t know that he allowed a single, just that he didn’t strike him out or walk him. This ball in play could turn into an out, dependent on the defense behind the pitcher and luck. The fact that it didn’t turn into an out is irrelevant to it. Also, context is irrelevant in that situation.

      FIP basically treats all balls in play equally (and by that, I mean it ignores them entirely).

      Also, walks are not the most overrated thing in baseball. There’s wins, RBI, batting average, errors, Derek Jeter, etc.

      Vote -1 Vote +1

      • CircleChange11 says:

        A stat that ignores hits under the assumption that pitchers have no control over has a flaw. Period.

        That doesn;t mean it’s a useless stat, or that it should be ignored … but it is not an all-encompassing stat. I just fail to understand the overwhelming desire to strip the team aspect out of pitching stats, since the pitcher does not EVER pitch alone.

        Example: The pitches are being called on the outer part of the plate, so the defense shades to the opposite field. Pitcher misses his spot, inside half, and the batter pulls the ball hard to straight away center, and the ball rolls to the wall. FIP/etc calls it luck. I call it a BIG pitching mistake.

        Look at batting averages leaguewide based on the various locations of the strike zone.

        Clearly they’re are spots that are better to throw the ball than others.

        FIP gives you credit for not allowing a HR, but a double off the wall is chalked up to defense or luck.

        Vote -1 Vote +1

      • Wally says:

        Circle,

        “A stat that ignores hits under the assumption that pitchers have no control over has a flaw. Period.”

        After having this conversation with you(?) and/or other’s regarding how much pitcher’s control BABIP, GB%, HR/FB and their effect on run scoring, I’ve spent several days doing multiple regressions, coming up with models and using a Baysian approach, and what ever I could think of, and the short of it was: The effect of even modestly controlable factors like GB% has at best very weak effects on run scoring. I’m sure plenty of saber-types have done this and reported it in the past, but I wanted to see for myself. The problems are somewhat two fold, there is a lot of variation in GB% year-to-year, and while the slope of GB% to ERA is statistically significant, the R^2 is an aweful .025. So, this moderately controlable factor has roughly only a 2-3% effect on ERA. This means that it is such a small part of the equation that it is easly overwhelmed by other factors that are more controllable (Ks for example) and have higher total effects (~20%). And no matter what model I tried to get that 2-3% of the variation in there, it just wasn’t worth adding another variable.

        So to argue against pitchers having no control is to set up a strawman. Most people don’t actually think they have no control over the types of hits (and I mean GB/FB/LD), its just that they have less control over the types of hits and the types of hits have relatively little effects on run scoring when compared to K, BB, HR. And by the way, K, BB, and HRs make up about 60-65% of the variation in run scoring. Unfortunately the remaining 35-40% comes from a slew of factors that are either not controlled at all by the pitcher (like fielding or park) or are only slightly controlled (such as the GB% above).

        Vote -1 Vote +1

    • DavidCEisen says:

      There is only one situation where a walk leads to an allowed run, but many situations where a walk leads to potential runs.

      Also the stat doesn’t say walking someone is better than giving up a single. The stat says it is best to strike out the batter and better to allow the player to put the ball in play–as there is a roughly 70% chance that will lead to an out.

      Vote -1 Vote +1

  5. John says:

    Lol on that last part ACWNS09.
    You people are too obsessed with FIP. I don’t see how a run scored on 2 hard hit doubles is any less legitimate then a home run. You act like the batter is going up there with his eyes closed. They intentionally try to hit the ball where the defense won’t get to it. When you do that you can’t just call it an unlucky break for the pitcher.

    Vote -1 Vote +1

    • Kevin S. says:

      The hitter who is capable of putting the ball to a spot on the field of his choice would be successful a heck of a lot more than 35% of the time. A hitter has some control over whether he pulls the ball, sends it up the middle or shoots it the other way, and he has some degree of control over how hard it’s hit. He does not have the ability to choose the optimal vector.

      Vote -1 Vote +1

      • CircleChange11 says:

        There’s a reason why LD% is an important stat. Harder hit balls go for hits more than “not hard hit balls”.

        I am of the belief that pitchers can influence the number of hard hit balls by their pitch [1] location, [2] movement, and [3] changing speeds.

        We act like pitchers are just helpeless observers about it, even though all teams have scouting reports regarding what zones the hitter likes the ball, and where they generally have success hitting it in the field (spray chart) … and they adjust the defense accordingly so that every advantage is to the defense.

        Strange that when I see pitches groove one and give up a double to the gap that scores 2 runs, they never seem to have the “Darn the bad luck” attitude, but are pissed that they missed their spot.

        Don’t they know that hits are just dumb luck or bad defense?

        Vote -1 Vote +1

      • Wally says:

        Circle,

        Doing a year over year correlation of LD% gives you a slop of basically 0 and similarly an R^2 of 0 (one for both would be a perfect correlation). So, LD% has no predictive value in the pitcher’s next year’s LD%. FB% and GB% are better however, and even come close to K/9. Here’s what I got for 2007-2009 (not a huge sample, but it should be enough):

        Type: slope/R^2
        FB%: .71/.53
        GB%: .72/.57
        LD%: .03/.01
        K/9: .8/.6
        BB/9: .64/.48
        HR/9: .39/.16

        So you can see the motivation for xFIP there. Then the issues with FB% and GB% expand as you try to tie them to scoring. Here’s their slopes and R^2 with ERA (I know I should just use RA, but it should be close enough):

        GB%: -.013/.025
        FB%: .007/.06

        By comparison:
        K/9: -.2/.19
        BB/9: .27/.09
        HR/9: 1.6/.35

        So, you can almost see FIP in there (-2K +3BB +13HR, which is how they got it, a multiple regression), and I suspect the numbers would come close to the true FIP equation with a larger sample size, percent of batters instead of per 9IP, and RA instead of ERA.

        Oh and line drives:

        LD%: .01/.06

        So, GB% and FB% actually do show a decent amount of control, between K/9 and BB/9, but on predicting scoring they just suck (slopes are atleast an order of magnitude lower and R^2 generally are too). And on the predictability side of HR/9 get a great deal of their correlation from pitchers remaining in the same home park year to year more often than not. This is why people care so much about K/9 and BB/9. Everything else is a pretty small factor (GB, FB), or it isn’t well controlled by the pitcher (HR), or both (LD).

        Vote -1 Vote +1

    • Travis L says:

      It seems like too many people are using FIP differently than I do. I pretty much only use FIP/xFIP predictively, rather than retrospectively.

      Basically FIP tries to remove luck out of the evaluation of a pitcher. When looking at a contribution that has been made, I think you have to look at the full results, including hits, etc. So when we talk about how valuable or good a performance has been, I think it’s more appropriate to use something like WAR.

      Now, when talking about how likely someone is to continue that performance, FIP is great. It’s better correlated that past results (with luck included). But for talking about what did happen, I tend to use WAR.

      Vote -1 Vote +1

      • Phantom Stranger says:

        Travis is exactly right. FIP is a great tool for projections of future performance, but should solely not be used to evaluate actual events that have already happened. It is too coarse of an analysis that throws out too much information when talking about games that we have complete data on that have already occurred.

        Vote -1 Vote +1

      • Graham says:

        Amen. There is a colossal and frequently misunderstood distinction between statistics being prescriptive and descriptive. When people pile on about how stupid a statistic like batting average is, they are missing a sort of fundamental, obvious fact: those hits happened in those at-bats. Is batting average useful for predicting how a player will do next year, or even next week? No. But it is a valid measure of how well (or how lucky — it doesn’t matter in an analysis of past events) a player hit during a given period. Irrespective of slugging percentage, a .330 average with a .425 BABIP at the end of the year isn’t any more or less valuable than a .330 average with a .240 BABIP, even though it’s clear that batted ball data would have a glaring impact on any predictions for the future.

        With regard to FIP, I think there’s a common misconception that goes something like this. ERA bad –> FIP new –> FIP better. I perfectly understand the flaws with stats like BA and ERA (and I’m intentionally using them here because, while they are inherently flawed, they’re by no means useless in reviewing a player’s performance), but during a season, teams win or lose based on what happens on the field — not what we expect would be the most likely thing to happen.

        So if you really want to evaluate a player’s year to date, you need to look to descriptive diagnostics, and nothing else. Even if a pitcher puts up a .150 OPP BABIP over three months, I don’t believe it should matter with regard to evaluating how well he pitched during those three months. After all, baseball is a game of both skill and luck. You can try to take the luck out of the equation when trying to get a sense of what will happen, or of what a player’s baseline level of skill is; but when analyzing a pitcher’s season to date, you can’t simply pretend that a screaming line drive got caught, or that a routine grounder to first hit the bag and bounced away from the first baseman. You have to fold these incidences into your analysis.

        Anyway, in the case of Cliff Lee — he is a badass. He’s having a great season so far, and thanks to his skill set, it’s likely that he’ll continue to have a great year. He’s a ton of fun to watch, and I can’t wait to see him the postseason again.

        Vote -1 Vote +1

      • artie says:

        Dead on – the problem is people do mistakenly use it retroactively… look at Cy Young talk…

        Folks forget these are predictive empirical, one size fits all models – they are not measuring actual results…. if you want to go back and say a guy got lucky on his defense – actually measure it, don’t just guess at it or model it…. review the game tapes and quantify how many ‘lucky’ outs or unlucky ‘hits’ he gave up. It’s the same ridiculousness with run support… look at it on a game by game basis and determine if the run support is actually making a difference – a 10-1 win doesn’t make the pitcher lucky with run support. Similarly a pitcher who loses 8-0 isn’t victimized by poor run support – yet you typically only hear how many runs a team scored in someone’s losses (and not how many runs that pitcher gave up in those games)

        Cliff Lee is a great pitcher but when he has an outing where he gave up 4ER in 6.1innings his FIP should not be going down if you are going to use FIP as an actual measurement of performance (as opposed to a predictor of performance).

        The problem came with the need to make FIP look like an ERA with the constant that is added…. had it just been an arbitrary # and it was understood that lower is better and indicative of future performance, folks would not be so cavalier about substituting ERA with FIP when talking about past results.

        Vote -1 Vote +1

      • Wally says:

        Graham,

        “a .330 average with a .240 BABIP”

        How would one accomplish such a thing? I guess a TON of sacrifices and no strike outs.

        But I don’t understand the flaming against “retroactive” use of FIP. FIP measures true talent better than say ERA or WPA, that’s the point of using for past performances, and why its more predictive. With the Cy Young example, why do I want to reward a pitcher who just got lucky?

        You also just knock down a strawman regarding the “old” stats. I don’t think anyone suggest that BA or ERA is useless, they are just not particularly useful in determining true talent or value. BA for example, does help to tell us what kind of hitter a player is. Say one guy’s line is .330/.380/.480 and another is .260/.380/.480. There is a lot of information in the “type” of hitter when using those 3 numbers in combination, one is similar to Ichiro and the other is closer to Jack Cust. Even the dreaded BA, HR, RBI gives you a certain type of information, though even less useful. And I would guess very few people that write for this site or even just read it would claim these stats are useless. But just showing me a guy’s BA doesn’t really tell me much until you get to the extreme fringes of what’s typically seen (like >.350 and <.240). Its a similar case with ERA. A guy can be just OK, but have some very good fielders in a pitcher friendly park, catch a little luck, and put up some really nice ERAs (say Trevor Cahill this year), and the opposite is also true.

        "Even if a pitcher puts up a .150 OPP BABIP over three months, I don’t believe it should matter with regard to evaluating how well he pitched during those three months. After all, baseball is a game of both skill and luck….You can’t simply pretend that a screaming line drive got caught, or that a routine grounder to first hit the bag and bounced away from the first baseman. You have to fold these incidences into your analysis."

        No you don't. That pitcher that put up a .150 BABIP, probably has a lot of thanks to be giving to his fielders, coaches for positioning fielders, and luck. So, why reward him for things he can't control? Yes, it happened, that's great. But it didn't happen because of him.

        Now, all that said, I understand your point of view, I just don't agree with this "you have to see it my way" thing you got going on here. It depends on which question you're trying to answer, and there is a very reasonable justification for why either question is more important. If you just want to know how valuable the events that were at least set in motion by the pitcher are, use WPA. But understand that isn't "talent." If you want to reward a guy for the most WPA regardless of talent, fine. I'd prefer not to, but I understand the logic. To me however, it makes more sense to reward the guy that displayed the most talent over the past year (or decade or career) and ignore as much luck or other players' contributions as we can.

        For example, lets go back to Cahill. His ERA is .01 lower than Lee's. But his FIP is nearly 2 full points higher than Lee's. Would you actually argue Cahill has been as valuable to his team as Lee has been to his 2 teams? Seems to me there is some middle ground where we should be ignoring these extreme outliers, such as Cahill's season thus far and those on the opposite end (really low FIPs but high ERAs), from our Cy Young conversations.

        Vote -1 Vote +1

      • CircleChange11 says:

        review the game tapes and quantify how many ‘lucky’ outs or unlucky ‘hits’ he gave up.

        Damn right.

        IMO, this is the difference between [1] the guys that create the stats and formulas, and do the research and [2] the guys that just parrot or spit out stat numbers.

        Stats are often used in ways that diverge from the intended usage.

        The perfect example last year was Ricky Nolasco whose ERA was 2 full runs above his FIP.

        Rather than just say “he had bad or poor defense”, show me. Because it looks like to me that Nolasco either can’t pitch out of the stretch as well as he can from the windup, or that he doesn’t locate the ball as well with men or, or rushes his motion to prevent stolen bases, or something that he CAN control.

        I accept luck as part of baseball, I just don’t defer to it as a default.

        Vote -1 Vote +1

      • CircleChange11 says:

        Not sure that I can do the math, but don;t something like 70% of line drives go for hits?

        So, couldn’t we look at LD%, for a pitcher and factor in the 70% for hit number and come up with an estimation of how many “extra hits” an average pitcher would allow if his LD% increased every 10th of a percent? (given the large numbers of batters faced)

        Perhaps factor in a “hitter quality” number based on division or whatever. But there HAS to be a better way of “isolating” a pitcher’s contributions without just disregarding non-HR hits as a whole.

        Is FIP adjusted for park factor? Dpeending on situations, the park you’re pitching in could affect home runs allowed to a significant degree.

        I ask these (perhaps dumb) questions, because FIP seems like such an easy to caluclate stat, and perhaps the 3.20 number is what is adjusted for league/park/opponent/etc.

        Vote -1 Vote +1

  6. DavidCEisen says:

    Not that it really needs to be said, but if Cliff and Roy were pitching together, the Phillies bullpen would be the most rested in the league.

    Vote -1 Vote +1

    • spindoctor says:

      And if they were pitching together, the Phillies would likely be in first place, not chasing in both their division and for the wildcard! :)

      Vote -1 Vote +1

      • CircleChange11 says:

        Actually, the way the Phillies season is going, either Lee or Halladay would be hurt by now.

        Whatever affected the Mets last year has moved west to Philly this year.

        Vote -1 Vote +1

  7. CircleChange11 says:

    Cliff Lee also racks up innings, and has done so in Texas, because they keep him in the game even when the team is losing.

    It’s an AL thing, for one aspect. NL teams pinch hit for their pitchers in late innings when they need to score 2 or 3 runs.

    Similar to CC in MIL. Lee is a rental. They are going to get every drop out of him.

    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>