• RBI In The Last Third Of The Game

    Posted by on August 22nd, 2008 · Comments (23)

    A post inspired by a recent comment made (to another entry here) by WasWatching.com reader “John ONeil.” To date this season, via Baseball-Reference.com, here are the Yankees “RBI” by player, during the 7th, 8th, and 9th innings of games:

    		7th	8th	9th	Total
    BAbreu		18	7	6	31
    RCano		9	11	3	23
    JDamon		8	8	4	20
    ARodriguez	15	1	1	17
    JGiambi		6	4	7	17
    DJeter		5	3	8	16
    HMatsui		8	1	2	11
    JPosada		1	4	3	8
    MCabrera		3	2	3	8
    XNady		6	2	0	8
    WBetemit		3	1	3	7
    RSexson		0	4	0	4
    CMoeller		2	0	1	3
    JMolina		1	2	0	3
    CRansom		2	0	0	2
    JChristian	1	1	0	2
    MEnsberg		1	1	0	2
    BGardner		1	0	0	1
    AGonzalez		0	0	0	0
    IRodriguez	0	0	0	0
    SDuncan		0	0	0	0

    Of course, this does not tell you how many chances a player had to drive in a run in these innings – nor does it tell you the score of the game at the time the RBI occurred. But, it is interesting to see that, through games of August 21st, A-Rod has just two RBI in the combined 8th and 9th innings of Yankees games this season. Two? How is that possible? But, it’s true.

    For the record, last season, A-Rod had 10 RBI in the 8th inning of games and 21 RBI in the 9th inning of games.

    And, I never would have guessed that Cano has the second most “late” RBI for the team. Considering how he’s been swinging the bat this season, what does that say about this Yankees offense?

    Comments on RBI In The Last Third Of The Game

    1. antone
      August 22nd, 2008 | 11:28 am

      Just for some comparison:

      Hamilton: (12,12,10) 34 RBI
      Morneau: (7,10,4) 21 RBI
      Cabrera: (11, 11, 8) 30 RBI
      Quentin: (11,16,6) 33 RBI
      Mora: (8, 11, 10) 29 RBI
      Youkilis: (9,17,5) 31 RBI
      Manny: (9,6,4) 19 RBI
      Howard: (6,8,6) 20 RBI
      Pujols: (11,7,3) 21 RBI
      Wright: (5,4,6) 15 RBI
      Berkman: (14,7,8) 29 RBI
      Braun: (7,7,12) 26 RBI
      Holliday: (9,9,6) 24 RBI

    2. antone
      August 22nd, 2008 | 11:28 am

      That should be an 8 not a smiley face for Cabrera..haha

    3. Misery Loves Company
      August 22nd, 2008 | 11:35 am

      FYI – Jeter is 1-10 this season from the 7th inning on with RISP when it’s critical (tied or down by 1 or 2 runs). The one hit didn’t drive in a run. (thru Sunday)

    4. ChrisS
      August 22nd, 2008 | 11:37 am

      And, I never would have guessed that Cano has the second most “late” RBI for the team. Considering how he’s been swinging the bat this season, what does that say about this Yankees offense?

      That RBI is an extremely poor metric for judging a player’s ability or skill level?

    5. August 22nd, 2008 | 12:07 pm

      ~~That RBI is an extremely poor metric for judging a player’s ability or skill level?~~

      So, when we see A-Rod’s 10 RBI in the 8th inning of games and 21 RBI in the 9th inning of games, last year, it also means squat?

    6. August 22nd, 2008 | 12:08 pm

      Misery Loves Company, FWIW, when I cite stats like those (for Jeter) here, usually, my readers kill me with the “small sample size” defense.

    7. ChrisS
      August 22nd, 2008 | 2:36 pm

      So, when we see A-Rod’s 10 RBI in the 8th inning of games and 21 RBI in the 9th inning of games, last year, it also means squat?

      As a predictive metric? Yep.

    8. OnceIWasAYankeeFan
      August 22nd, 2008 | 3:19 pm

      How about as a descriptive metric? That’s what Steve is using it for, to describe reality. And the reality is that a bunch of A-Rod’s RBIs last year came in the last two innings, and hardly any this year have come in the last two innings.

      I’d also argue that ANY late game RBIs should be regarded as more valuable. Yeah, a few of them will come during blow-outs, piling on against an over-taxed bullpen. But the rest – virtually every one of them will ice a game, get your team closer, or be tie or put you ahead. That’s what “clutch” hitting is all about, and A-Rod hasn’t done it. You can measure it by his BA with two outs and RISP or you can look at the appalling lack of late-inning RBIs. Different DESCRIPTIVE measures of the same fundamental fact.

      Are they predictive? Yeah, they’re predictive of an offense that struggles to score enough runs to win. They’re probably not predictive of A-Rod’s performance next season.

    9. ChrisS
      August 22nd, 2008 | 4:21 pm

      Descriptive of what, though? How un-clutch A-Rod is?

      The only thing those numbers describe is that RBI are dependent on so many variables in a small sample size, which results in huge swings. It doesn’t describe anything that can be solved. So what’s the point? That the Yankees should do everything in their power to get this year’s leaders in RBI after the 7th inning? What’s that solve for 2009? It’s a pointless stat.

      The Yankees offense performed pitifully this year. The only descriptive statistic needed to describe reality is that they’re on pace to score ~200 runs less than 2007. And it doesn’t matter if those runs are score in the 6th or 2nd innings. Add 200 runs to the Yankees and they’re in first or the wild card lead. My guess is that, over the course of a season, teams score ~1/3 of their runs in the last third of a game:

      Team, % runs scored (1/2/3)
      2007 New York Yankees: 34%/36%/29%
      2007 Boston Red Sox: 31%/36%/32%
      2007 Baltimore Orioles: 35%/34%/28%
      2007 LAA Angels: 37%/36%/27%
      2008 NY Yankees: 31%/37%/32%
      2008 First place Rays: 33%/35%/32%
      Etc.

      There does seem to be a dropoff in the last third regardless of team (at least with these few), which indicates that because they’re either facing a very good starter, who is dominating, or the closer, who is typically an above average reliever, they’re not scoring as many runs. Wow. Now that we’ve got that out of the way we can go back to bashing Cashman for not getting clutchity-clutch hitters.

      And batting average sucks as a metric, too. Just throwing numbers out there from splits doesn’t mean squat. Splits are going to be weird because they involve small such sample sizes. As the number of PAs approach a statistically significant number, the metrics typically fall very close to the player’s mean.

    10. August 22nd, 2008 | 5:55 pm

      I hear what you’re saying…but…it still seems strange that your clean-up hitter, who many consider to be THE BEST hitter in the game, has only ONE, just one, RBI in the 9th inning, all season, and it’s now August 22nd. And, it seems worse when he also only has one in the 8th inning too.

      Again, I understand the nature of stats, etc., I’m just saying, despite that, it still looks bad…all things considered.

    11. hopbitters
      August 22nd, 2008 | 6:00 pm

      I’d also argue that ANY late game RBIs should be regarded as more valuable.

      -

      I guess Lee isn’t on duty today. Runs count the same no matter when you score them. The sixth run in a 6-5 victory isn’t any more meaningful than the third run you scored in the first. Conversely, the tying run that didn’t drive in during the rally in the 9th in a 6-5 loss doesn’t hurt you any more than the run you didn’t drive in during the second inning.

    12. August 22nd, 2008 | 10:40 pm

      [...] to see some RBI in the 8th and 9th inning of this one, [...]

    13. OnceIWasAYankeeFan
      August 22nd, 2008 | 11:19 pm

      I guess Lee isn’t on duty today. Runs count the same no matter when you score them. The sixth run in a 6-5 victory isn’t any more meaningful than the third run you scored in the first. Conversely, the tying run that didn’t drive in during the rally in the 9th in a 6-5 loss doesn’t hurt you any more than the run you didn’t drive in during the second inning

      ____________________________–

      Right. So, you can also say that the missed field goal in the 10th minute of the second quarter hurts no more or less than the missed field goal as time expires, right?

      Sorry, but when time is running out, whether time is measured in outs or in minutes left in the game, cashing in (or blowing) scoring opportunities have a much greater effect on the outcome. Or perhaps you should take a look at that site, what is it called, fan graphs or something? The one that charts the likelihood of victory based on the events of the game. I don’t pay a lot of attention to it, but I very seriously doubt that breaking a tie in the second inning has the same impact on likelihood of victory as breaking a tie in the eighth inning.

    14. hopbitters
      August 23rd, 2008 | 12:06 am

      So, you can also say that the missed field goal in the 10th minute of the second quarter hurts no more or less than the missed field goal as time expires, right?

      -

      Unless the field goal counts for four points as time expires, they’re exactly the same.

    15. John ONeil
      August 23rd, 2008 | 2:51 am

      Steve,

      Thanks for researching the real numbers. I couldn’t remember exactly what was said on YES. I knew the difference between ’07 and ’08 for A-Rod was enormous. From your numbers its clear they were talking 8th & 9th, not 7th, 8th and 9th.

      As for the value of runs scored early vs. late, or in blow out vs. tight games, I recommend looking at the stuff the guys at The Hardball Times and fangraphs.com have done to track Win Probability Added and Leverage Index. Basic concept is that a home run in the first inning that makes the score 1-0 increases the probability of winning less than a home run in the 8th inning that makes it 1-0. Not hard to understand that. WPA measures how a batter has increased/decreased his team’s probability of winning over a season.
      The numbers for A-Rod can be found here http://www.fangraphs.com/statss.aspx?playerid=1274&position=3B. In his two MVP Years, A-Rod had WPA of 5.52 and 6.85. So far this year, he is 0.75. Ouch. By comparison, Ortiz in his career years (’05 and ’06) had WPA of 8.21 and 7.76. This season’s leaders are Lance Berkman at 5.99 and Manny at 5.18.

      WPA is not a perfect stat by any stretch. A batter on a good team that a takes an early lead in most of its games will suffer compared to a great player on a team with less talent how drives in runs late. It is also not a predictive stat nor a measure of skill. A-Rod is obviously skilled. But the general perception that he is not getting runs in highly leveraged situations is born out by the numbers.

      Basically, his numbers this year are much less meaningful than they were last year or in 2005. Giambi is up a little, Abreu and Damon a little more. Jeter, Cano, Melky and Matsui are down.And But none have dropped nearly as much as A-Rod. In fact his drop exceeds the combined total of all four of them. It was unrealistic to expect another year like last, but to fall as far as he has is a major reason why we are where we are and another measure of our anemic offense.

    16. OnceIWasAYankeeFan
      August 23rd, 2008 | 7:54 am

      Hopbitters,
      Read John Oneil’s post. Repeat as necessary.

    17. Raf
      August 23rd, 2008 | 8:47 am

      Hopbitters,
      Read John Oneil’s post. Repeat as necessary.
      ————
      “WPA is not a perfect stat by any stretch. A batter on a good team that a takes an early lead in most of its games will suffer compared to a great player on a team with less talent how drives in runs late. It is also not a predictive stat nor a measure of skill.”

      That part? :)

    18. OnceIWasAYankeeFan
      August 23rd, 2008 | 9:50 am

      :)

      Actually the part about the impact of runs scored at different points in a game. ;)

    19. hopbitters
      August 23rd, 2008 | 12:55 pm

      Basic concept is that a home run in the first inning that makes the score 1-0 increases the probability of winning less than a home run in the 8th inning that makes it 1-0.

      -

      Win probability tells you the likelihood that a team would win a game in a simulation based on historical data where the outcome was already determined and factored into the analysis. That’s why it has no predictive value.

      The win probability values are based on a series of events, not a single event. If you don’t factor in what happened in the rest of the game, they have no meaning. So the value of the run in the 8th is based on the value of not allowing any runs up to that point. If you score the run in the first inning and add the value of not allowing runs for 8 innings, they are the same.

    20. John ONeil
      August 23rd, 2008 | 1:10 pm

      Coming through when the team really needs you is expected of an MVP. If Mariano had an ERA equal to that of a great starter ~ 3.00, he would not be regarded as the greatest closer of all time. His job is highly leveraged and demands more. A placekicker who leads the league in distance and accuracy during the first 55 minutes of a game, but misses 12 of 13 attempts in the closing minutes would lose his job (maybe the team would go baseball-style and have “closer” kicker who just kicks the game enders)

      A field goal is a field goal. A home run is a home run. But the ability to hit it when there is more pressure on you is obviously valuable even if the field goal is still only three points. For most athletes, the difference in performance between the different stages of the game is small and the numbers revert to the mean as the sample size gets bigger. A-Rod has shown an almost all-or nothing pattern in the last four years with respect to hitting in important situations. Two excellent years. Two bad years.

      So the good news is that WPA is not a good predictor of future performance. We have nine more years of A-Rod and hopefully we will get at least 5 of his “deliver in the clutch” years. The bad news is that unless he (and a few others) start hitting when we need it (like with RISP), our slim playoff hopes will soon disappear.

    21. hopbitters
      August 23rd, 2008 | 1:27 pm

      Just to be clear, I agree absolutely that A-Rod has not been performing at the same level in “pressure” situations (though A-Rod not performing at his best is still better than a lot of people). I just don’t agree that the home run he hits in the first counts any more or less than the one he doesn’t hit in the ninth. The one he doesn’t hit with RISP versus the solo shot hurts without question.

    22. OnceIWasAYankeeFan
      August 23rd, 2008 | 3:49 pm

      The win probability values are based on a series of events, not a single event. If you don’t factor in what happened in the rest of the game, they have no meaning. So the value of the run in the 8th is based on the value of not allowing any runs up to that point. If you score the run in the first inning and add the value of not allowing runs for 8 innings, they are the same.

      _________________________________

      I don’t think you have a clue what you are talking about. The win probability increases with a home run in the eighth inning more than it does with a home run in the first inning. The win probability does not change afterwards because someone threw nine innings of shut out ball. If anything the win probability would take a jump when the pitcher works out of a two on, none out jam in the eighth and hands it off to his closer.

      And the value of the home run in the eighth inning is dependent not on having allowed no runs up to that point, but in how that home run changed the margin between the two teams. If it turned a ten run deficit into a nine run deficit, it didn’t have a lot of effect on win probability. Same thing if it turned a ten run lead into an eleven run lead. But if it turned a one run lead into a two run lead, you can bet that the Win Probability changed markedly. And you can also bet that the solo shot in the third that changed the score from 2-1 to 3-1 did not change the win probability as much as the same home run in the eighth.

    23. hopbitters
      August 23rd, 2008 | 5:19 pm

      The win probability increases with a home run in the eighth inning more than it does with a home run in the first inning.

      -

      The “win probability” is based on the historical outcome of games (starting sometime in the late 70s) in similar situations, based on the number of outs and the score differential (at least with WEF). Because there is more variation in possible outcomes in early innings, the “probability” of winning is less. Either run still contributes the same amount to the total runs scored.

    Leave a reply

    You must be logged in to post a comment.