Don't be shellfish...FacebookTwitterGoogle+

We all realize that a player’s lineup position will have an impact on his statistics. I’ve recently been pondering the fantasy impact of scenarios such as:

  1. What is the impact of Russ Martin and Matt Kemp’s fantasy value if they hit 2nd vs. in the 5th through 7th spots?
  2. If Kelly Johnson moves from 1st to 6th, will his fantasy value be impacted?
  3. How much does it hurt a Cameron Maybin or Alex Gonzalez to hit 8th in an NL lineup?

I’ve come across analysis on optimizing a lineup – such as in Tom Tango’s Inside The Book – but I haven’t come across direct guidance on how this impacts fantasy value.  So I thought I’d take a shot at it.

If I were to predict the impact of lineup position on a player’s 5×5 stats, this would be my best guesses in descending order of perceived impact:

  1. Runs – Hitting in the top 4 would have an advantage vs. hitting 5th-8th (with NL players hurt by pitchers’ reduced ability to drive in runs)
  2. RBI – Hitting in the middle of the lineup would have an advantage vs. hitting at the top or bottom of the lineup (with top of the lineup worse off in NL because of pitcher)
  3. SB – Hitting first or at bottom of lineup would have an advantage because managers are less likely to run players with middle of lineup hitters at the plate (exception hitting 8th in NL because pitchers will bunt them over)
  4. AVG – Maybe hitting at the top of the lineup gives you better protection so you’ll see better pitches?  (disadvantage hitting 8th in NL as you have no protection)
  5. HR – While it’s possible that top of lineup hitters are less likely to swing for HRs than middle of the lineup, I’m not convinced there is any real difference here except that top of lineup hitters get more plate appearances.

The big reason why I think Runs and RBIs are the most affected by lineup position is that they are largely out of the hitter’s control.  You can hit Pujols 8th in an NL lineup and he’ll hit over .300 and hit 30 HRs.  But there is little doubt that he’ll come up with less batters on base and be stranded more on base.  Because this is so lineup dependent vs. skill dependent,  Runs and RBIs prove to be the easiest ones to measure lineup impact because we can analyze aggregated totals for each position (vs. adjust for player skill levels).

Here is how I measured the Runs/RBI impact:

  1. Gathered the hitting statistics by lineup spot for each of the 30 teams for 2008 (courtesy of Baseball-Reference.com)
  2. Created a Plate Appearance index that accounts for the differences by lineup position. This is based directly on the aggregated results from B-R and, obviously, #1 hitters get the most plate appearances over the season, #2 hitters get the second most, etc.
  3. Created a Runs index as follows:  (R-HR)/((1B+BB+HBP+ROE)+(2B*1.7)+(3B*2.4)).  This formula divides Runs by Total Times on Base, removing HRs as this nets the same result for any lineup position.  ROE = Reached on Error.  I weighted doubles and triples up based on some information gathered on Baseball Analysts regarding run probability when on 1st base, 2nd base, or 3rd base.  The impact of weighting doubles and triples vs just counting all hits as equal was negligible.  I was not able to factor fielder’s choices into the denominator.
  4. Created a RBI index as follows:   (RBI-HR)/(H-HR).  I wasn’t sure how to value the RBI potential of doubles and triples over singles so I didn’t account for it.  RBIs potentially gained by sacrifice flies or ground outs weren’t accounted for in the denominator.
  5. Created a R/RBI index that averages the two indexes.

Below are the results for AL and NL.  The indexes are skill-independent.  So if an AL hitter moves from the leadoff spot to the 9th spot and hits exactly the same, they will have 22% less plate appearances (110/90 – 1), 41% less runs (127/90 – 1) and 14% more RBIs (1- 78/90).  Also, realize that this is based on opportunities to score or drive in runs.  Just because the NL 9th hitter has a higher index than 8th hitter does not mean they will score more runs in total.  It means that a NL 9th hitter who gets on base is more likely to score a run than an 8th hitter.

American League (2008)

Lineup Position PA Index Runs Index RBI Index R/RBI Index
1 110 127 78 104
2 108 121 92 108
3 105 103 123 115
4 103 98 126 114
5 100 95 114 106
6 98 89 98 95
7 95 93 96 96
8 92 85 95 91
9 90 90 90 87

National League (2008)

Lineup Position PA Index Runs Index RBI Index R/RBI Index
1 110 125 75 100
2 108 120 82 102
3 105 103 107 105
4 103 98 133 115
5 100 87 119 102
6 98 81 100 90
7 95 79 93 86
8 93 77 86 82
9 90 84 82 84

For Runs, the indexes mostly confirmed expectations.  I found it a bit surprising that AL 6th hitters are at a slight run disadvantage vs. AL 7th hitters but it’s understandable as 7th hitters benefit more from being driven in by top of the lineup hitters.  I’m also surprised how big of a run advantage leadoff hitters have vs. 3rd/4th hitters as their actual run totals tend to equal out.  I guess the biggest driver here is the larger OBP that 3rd/4th hitters generally have vs leadoff hitters (oh for the days of Rickey and Boggs’s .400+ OBP).  A last point is that the Runs Index for AL/NL are near identical for the first 4 lineup spots but then the NL slides more precipitously for 5th-8th hitters.  This may be a sign of both the NL’s lack of a DH (which reduces lineup depth) and the weak 8th/9th hitters in the NL.

For RBIs,  the AL RBI index distribution is as expected.  The 3rd/4th/5th slots have the biggest advantages with 3rd/4th being similarly advantaged over the 5th slot.  The NL indexes, however, show a much bigger RBI advantage for 4th hitters and (to a lesser expent) 5th hitters vs. 3rd hitters.  NL 2nd hitters also see a big drop in RBI advantage vs. their AL counterparts.  The explanation for both is likely reduced RBI opportunities because of low NL 8th and (especially) 9th hitter OBPs.

The Runs/RBI indexes indicate that any move from the top 5 lineup spots will lead to a measurable decline in total Runs/RBIs – at least partially driven by the increased plate appearances.

So to answer the earlier examples:

  1. The R/RBI impact for Kemp and Martin if they hit #2 vs #5 is basically nothing – you’re just trading off Runs when hitting 2nd to RBIs when hitting 5th.  But moving down to 6th or 7th does have a 13-20% impact on R/RBI.
  2. If Kelly Johnson is moved to the 6th slot, he’d see an estimated 33% increase in RBIs (100/75 – 1) but an estimated 35% drop in Runs.  It nets out to a net loss in combined Runs/RBI but it might not be a bad tradeoff if you’re team is weaker in RBIs than Runs.
  3. Hitting 8th in an NL lineup is the death knell for Runs but is actually better for RBIs than either hitting 1st or 2nd.

Hope this analysis comes in handy when re-assessing players’ values based on their lineup slot.  Next step for me is trying to figure out how I can neutralize the skill-dependency to measure SBs and AVG impact of lineup position.

32 Responses

  1. Scott says:
    (link)

    Great article, definitely makes me reassess making a big play for a Kemp, despite how much I like him.

  2. Elijah says:
    (link)

    I guess the biggest driver here is the larger OBP that 3rd/4th hitters generally have vs leadoff hitters (oh for the days of Rickey and Boggs’s .400+ OBP).

    Shoot, lotta teams could use a Brett Butler let alone those guys. (Curse you White Sox leadoff hitters hitting in front of Quentin)

  3. Homer at the Bat says:
    (link)

    @Scott: But remember, Kemp is generally hitting 7th right now, not 5th. Thus, according to Rudy’s tables, a move from the 7 hole to the 2nd spot (or 5th spot) would result in a 18.6% increase in Runs/RBIs ((102-86)/86).

    The bigger problem is trying to figure out what the hell Torre’s going to do on any given night. Hudson seems pretty entrenched in the 2nd spot, circling the bases for Manny. However, those darn 4th-7th spots seem sorta up in the air. Kemp appears to be moving up to the 5th spot when Manny or Ethier take a day off… but that doesn’t happen incredibly often, or at least often enough to make an appreciable difference in Kemp’s value right now.

    If you’re targeting Kemp and figuring on a Dodgers lineup shift, I wouldn’t hold your breath, despite Kemp’s potential.

  4. NoonTime says:
    (link)

    Nat’s closer slop… Taverez was brought in to pitch the 7th with a 3 run lead… he is sucking. Now it’s Mock’s turn. Hanrahan has pitched well his last 3 outings.

  5. Quintero says:
    (link)

    Great tackle by Mr. Rudy Gamble.
    I think hitting 8th in an NL lineup is the most interesting case.
    I would certainly consider the possibilities of walking Pujol every time if he hits in front a pitcher. Also early this season, when Chris Synder wasn’t hitting well and showing little aggressiveness, the opponent just walked him and tried to get the next batter/pitcher in early-game/2-out situation.
    Is it possible for you to share the BB/IBB stats from players who hitting 8th in an NL lineup? It might be interesting. Thx!

  6. Bailey says:
    (link)

    Rudy, maybe I missed it but what I don’t see here is players ability factored into the calculations or being held constant. You can’t just compare the product of a leadoff hitter to the number 7 guy, there’s a reason that guy is batting leadoff – he is better at something, and thus should have better numbers. Or did you compared the same player when he batted 1st to when he batted 7th for example?

  7. @NoonTime: As a Garrett Mock prospector in my NL-only, I am PISSED at Acta. Mock is a young pitcher who just threw 2 innings (24 pitches) of scoreless ball last night. He should not have been used again today – let alone brought in with 1st/3rd no out. Acta is threatening Torre for worst bullpen manager.

  8. @Bailey: The calculations are based on the number of opportunities they created. So it doesn’t matter, let’s say, if a leadoff hitter has an OBP of .300 or .400. What matters is, when leadoff hitters get on base, how often do they score. A leadoff hitter who gets on base is much more likely to be driven in than an 8th hitter – independent of their skills. Runs/RBIs, therefore, are largely team-dependent unless you hit a HR.

    This approach cannot work for SB/AVG/HR as these are largely skill-dependent vs team-dependent. You would have to use players who hit at different lineup spots and that, in itself, is going to prove problematic because of small sample sizes and circumstantial differences (maybe a lefty is moved from leadoff to 7th when facing a lefty so the average decline has nothing to do with lineup slot). So not sure how successful I’ll be in that analysis…

  9. Scott says:
    (link)

    NL Only Keeper League: I trade Johan and C. Lee for Kemp and Gallardo. This is more of a flexibility move as both owners need to move talent to keep other players. Am I giving up too much? P.S., I’m not too worried about this year. So, am I giving up too much for 2010 and beyond?

  10. @Quintero: Great question. Here’s a comparison of BB rate / IBB rate for AL 8th hitters vs. NL 8th hitters

    AL: 7.5% Non-Intentional BB/Plate Appearance; 0.64% IBB/PA, 8.1% BB/PA

    NL: 7.7% BB/Plate Appearance; 1.5% IBB/PA, 9.2% BB/PA

    In both cases, the IBBs are counted within the BBs.

    So, basically, 8th hitters walk at about the same rate in both leagues if you factor out IBBs but an NL 8th hitter is almost 2.5x more likely to get intentionally walked than an AL 8th hitter.

  11. Josh says:
    (link)

    This is amazing work. Some of the best fantasy baseball research I’ve seen.

    One question that is left for me: weren’t the brewers and cardinals hitting their pitchers eighth occasionally last year? I don’t know if they are still doing it, but I would imagine that would sort of throw a wrench in these numbers for 5, 6, or 7 hitters.

  12. @Josh: Thanks. You are correct that hitting the pitcher 8th would impact the assumptions slightly. For a team that does that, their 5th-7th hitters would have reduced run rates while their 9th hitter would have a reduced RBI rate. I don’t think it was prevalent enough to have an effect on the data but I could conceivably back out those two teams to see if there are differences…

  13. GasTheObese says:
    (link)

    Glad someone mentioned the IBB/BB rate of 8th spot hitters in the NL. Jordan Schafer’s .389% obp is a prime example of that.

  14. jsp2014 says:
    (link)

    this site is the shit (obv).

  15. Billy says:
    (link)

    Quick trade question…would you all do Kazmir for Bedard/Dukes?

  16. Dingo says:
    (link)

    GAAAAAAA, JOBAAAAAAAAAAAA!!!

    Fantastic post, but I had to get that out of my system.

  17. Paulie Allnuts

    Paulie Allnuts says:
    (link)

    @Rudy Gamble: Excellent post. I took advanced calculus years ago, but this is almost as hard to follow (then again, I have lost many brain cells in the last several decades.) Billy Beane should hire you. I will have to * this as a favorite, and refer to it when necessary.

  18. So, you’re saying that it’s better for your RBI production to have the best possible hitters in front of you, and better for your run production to have the best possible hitters in front of you? And that you will get more “counting” stats if you bat earlier in the order, and therefore get more at-bats (and have better batters around you?).

    Amazing, but I don’t know … these sounds like real leaps. I’m going to need much more research and evidence before I am convinced that all of the above is true.

    J/K. Good work.

  19. @Billy: I’d take Bedard/Dukes. I think Bedard and Kazmir are about equal.

    @jsp2014: Thanks!

    @Paulie Allnuts: Ha. I never really excelled in math after Algebra 2 and Prob/Stats. Trig and Calc weren’t my thing. Guess you can just take my word for it on the math and use the indexes to upgrade/downgrade Run and RBI projections…

  20. AL KOHOLIC says:
    (link)

    @Rudy Gamble: awesome research,what we all individually do with it can make a large difference in finishing 3rd or 1st,thanks,i know it raddled the brain doing this

  21. Prince says:
    (link)

    Rudy, thanks for your work! Helpful and impressive.

    Question: Drop any of Kershaw, Jurrjens, or Wandy for Galarraga?

  22. @AL KOHOLIC: Thanks!

    @Prince: You’re welcome. And I wouldn’t drop any of those 3 for Galarraga unless you’re talking about somehow resurrecting Andres Galarraga when he was mashing in Colorado…

  23. Drev says:
    (link)

    @Rudy Gamble: I’d like your thoughts on Billingsley, Hamels, and Kazmir for the rest of the season; I’ve been offered Hamels and Kazmir for Billingsley. Thanks.

  24. @Drev: I like C-Bills best of the three but that’s only b/c Hamels seems snake-bitten. If you need SP, I would definitely make this deal. You’re getting 2 top 25 pitchers for the price of one…

  25. Griff says:
    (link)

    @Drev: Even if Billingsley wins the CY Young, I gotta believe Hamels and Kazmir will be worth it in the end. Hamels’s injuries have been as much bad luck as physical breakdown, IMO, and Kazmir can’t stay this bad.

  26. Dingo says:
    (link)

    I also wanted to add that I really appreciated how you clearly explained the complicated analysis that you conducted. Too many baseball statheads have a hard time translating their statistical wizardry into plain English for the rest of us to understand, but I found your explanation to be surprisingly easy to follow.

    Unfortunately, your findings are making me feel like my newly-acquired Curtis Granderson is not going to be the savior for my Runs deficit, now that Leyland is talking about keeping him in the #5 spot for a while.

  27. @Dingo: Thanks. I figure it’s a lot easier to spell out the math and the assumptions behind an analysis than have people ask questions afterwards. And explaining it in plain English is a good way of testing whether the analysis was correct in the first place.

    As for Granderson, you’re completely right. Hitting 5th will knock his Runs down considerably (25%). And Granderson is an average OBP guy (~.360) so he won’t outperform at that lineup position (his speed should help a little). On the positive side, it’s always easier to trade HR/RBI than it is to trade R/SB. So if Granderson keeps up his crazy HR pace, you’ll have no shortage of buyers….

  28. Well aware I’m late to this party, but this is very interesting stuff, Rudy. A quick question, though: you say that this is all skill-independent, but could you explain how that’s the case for RBI? I see you explained Runs in #6 (though there is some selection bias in that fast runners bat lead-off more often and thus are the guys on base). For RBI, though, I’m assuming you’re looking at RBI opportunities (i.e. someone is on base), but wouldn’t the skills of the hitter come into play when determining whether the runner is driven in? It’s late, so please excuse me if I’m missing the obvious answer here.

  29. @Derek Carty: Good point on the fact that player speed isn’t taken into account for Runs or RBIs. I imagine that this would slightly boost the Runs for spots with typically fast runners and the RBIs for those hitting in middle of lineup spots.

    As for RBI, one way would be to look at RBI opportunities. I’ve never seen that data in aggregated form by lineup position, though. So I went with (RBI-HR)/(H-HR) which assumes that getting a hit during non-RBI opportunities is the same skill as getting a hit during RBI opportunities. AL cleanup hitters have a ratio of .53 RBIs/Non-HR Hits while leadoff hitters have a ratio of .31. While speed of baserunners and double/triple-rates play small roles that I was unable to factor, I think this disparity is near exclusively driven by lineup position.

  30. Gotcha, Rudy. My main concern is a selection bias issue. The player who an MLB manager decides to put in the clean-up spot is not a random exercise. While managers don’t perfectly optimize lineups, they aren’t stupid either. You would never find Jose Molina batting clean-up for the Yankees or Fernando Tatis for the Mets. The guys who hit clean-up are big boppers, so I don’t think it’s fair to say that moving *anyone* to the cleanup spot would lead to a 33% increase in RBI. That 33% is being driven by the likes of Ryan Howard and Adam Dunn, and moving Alex Gonzalez into the spot wouldn’t produce the same result.

    However, this won’t always be of any concern to fantasy owners as the reason for the selection bias also means that we won’t have to worry about it much; these non-boppers won’t be moving into the cleanup spot. Still, if a borderline player gets moved into the spot ever, or if we look at murkier situations like the #2, #5, or #6 spots, we might run into some trouble.

    Still, very interesting stuff, Rudy.

  31. kangaroo hops says:
    (link)

    Is there any real statistical proof to the idea of protection of a big hitter? It sounds great in theory, but I have read sources (e.g., Paul Dickson in his Baseball Dictionary) who claim that there is no convincing evidence that protection makes any difference. Hey, not THAT type of protection! You know what I mean.

Comments are closed.