Filter posts by...
Batting_Order
Thursday, July 22, 2010
I like this:
With as many moving parts and versatile defenders as the Rays have, their lineup is truly day-to-day for the rest of the team.
“We try not to let it affect us,” Zobrist said. “The effect that it could have if you let it is you could get frustrated thinking, ‘Well, maybe I could hit better if I stayed in the same spot.’ I think all those are excuses. If you’re in the lineup, you face the pitcher just about the same times as anybody else, so I think the best reaction to it is roll with it.”
...
“His job is to make the lineup. Our job is, whatever spot we’re in, to hit,” Zobrist said. “It’s simple. It really is, if you look at it that way. No matter what spot you’re in, if you help the team win from that spot, it’s a good spot for you.”
Thursday, April 08, 2010
Wonderful.
Tuesday, April 06, 2010
In my Markov tests in The Book, putting the pitcher 9th or 7th was breakeven, with 8th being the optimal spot, but on the order of 2-3 runs or so. (When MGL does his studies, he find the pitcher batting 9th to be ever-so-slightly preferable.) So, when Dave reports this:
The average pitcher-bats-9th team was 0.23 runs below optimal, while Pittsburgh and the AL teams averaged 0.07 runs below optimal. This would suggest flipping the pitcher and 8th hitter on the other NL teams would result in an improvement of about 0.16 runs. Over 162 games that is 25 runs or 2.5 wins, a surprisingly high number to me.
It should be surprising because it is impossible. In my tests, moving the pitcher from 8th to 4th caused a 0.10 runs per game difference. That’s the absolute limit for moving one player to the worst spot, while keeping the other 8 guys in the same order. I presume Pinto’s model used SLG, OBP and regression, which I think he based it on Cy’s work (all this was done a few years ago)? If that’s the case, then there is a limit to its usefulness.
Friday, March 19, 2010
Not sure if i ever linked to this. That article was from 2008. No idea how he’s doing the simulation, though I presume it’s similar to my Markov calculator.
Thursday, March 18, 2010
Sky follows the general recommendations in The Book, and comes up with the following against RHP:
1. Victorino (S)
2. Utley (L)
3. Werth (R)
4. Howard (L)
5. Ibanez (L)
6. Rollins (S)
7. Polanco (R)
8. Pitcher
9. Ruiz (R)
1. Nick Johnson (L)
2. Mark Teixeira (S)
3. Curtis Granderson (L)
4. Alex Rodriguez (R)
5. Robinson Cano (L)
6. Derek Jeter (R)
7. Nick Swisher (S)
8. Jorge Posada (S)
9. Brett Gardner (L)
It’s important to note that the main driver is the forecast of the players. And ZiPS has Jeter forecasted as, overall, a bit above average. That would put him pretty close to league average against RHP.
I think these lineups should be the standard ones, and then you can run simulations from this baseline to see where, and why, you can pick up more runs.
Monday, March 01, 2010
Berselius. For those not aware, I also have a Markov calculator, but it doesn’t do lineups. It uses recursion, the code is there for anyone to use. Adding in lineups simply means adding one more dimension to the array. I’ve done it (for The Book). The code is very small when you use recursion.
I encourage all you programmers to look at my code and expand it for lineups.
Friday, February 26, 2010
John Walsh and Steven Goldman.
Two points:
1. When you see a statistical significant to p < whatever-number observed split, that does NOT mean that the observed split it significant at that level. It means that the NON-ZERO difference is significant at that level. So, if let’s say you have a .400 wOBA in one split and .300 in the other, and it’s significant at p=.04, then it means that it could very well be .335 and .334 as the true splits. It’s non-zero.
2. Other than leadoff hitter, if you can find me a one-position switch of a .300 to .360 wOBA hitter that will cause more than a 10-run difference, I will be impressed.
Thursday, October 08, 2009
By , 03:49 AM
There is some controversy over who should catch when Burnett pitches for the Yankees. Girardi said that Molina will get the call. Molina is one of the worst hitters in baseball. Posada is one of the best hitting catchers in baseball (not named Mauer). There is a gigantic swing between the two. Having Molina in the lineup rather than Posada is equivalent to taking Mauer out of the MIN lineup, which would get you executed in Minnesota, right?
But, does it really make that much difference to the Yankees’ chances of winning the game? First of all, Molina is much the better defensive catcher I assume (would Molina even have an MLB job if he wasn’t excellent defensively?). Also, he is going to bat 9th and thus only get 2 or 3 PA before he is replaced with a pinch hitter if the Yankees are losing or if Burnett comes out of the game. In addition, Posada becomes available to pinch hit for anyone (although most of the Yankee starters are pinch-hit proof).
Anyway, I ran both players through my sim. Although my sim does pinch-hitting, I think it probably lets Molina bat for most, if not all, of the game. I’m not sure about that. Plus, it does not fully include catcher defense, like pitch blocking (which I think Posada is terrible at). Oh, and Posada is still playing hurt right now, which I included in the sim. I knocked him down a notch or two, although he is still way better offensively than Molina.
The results: With Posada (batting 7th), Yankees score 6.17, allow 4.18 rpg and win 72.15% of the time, in 100,000 games. With Molina (batting 9th), they score 5.85, allow 4.05 rpg and win 71.33% of the time.
Given the limitations of the sim which I just mentioned, and the possibility that Burnett actually pitches better with Molina behind the plate, I think it is a wash at worst. Nothing to get in a huff about if you are a Yankee fan.
Tuesday, September 01, 2009
By , 02:58 AM
If you were hoping for a nice study, you will be disappointed. While I’ve always wondered this, I have not yet studied the issue. However, I did run a couple of lineups through my sim to see what it would come up with.
Here is an exactly league average lineup (total lwts of exactly zero, according to my projections), which is comprised of two of the best hitters in the NL plus a bunch of below average hitters. I even put a bad one in the leadoff spot, so that you have a bad hitter, a very bad hitter, and the two really good ones in Braun and Fielder, followed by 5 more below average ones. Here is that lineup:
Alex Escobar
Craig Counsell
Ryan Braun
Prince Fielder
Casey McGehee
Matt Gamel
Jason Kendall (boy does he suck)
Frank Catalanotto
It generated 4.515 rpg versus a average RHP and 4.171 versus an average LHP for an average of 4.343 rpg.
Now, here is also a league average lineup (total lwts of zero) but much more balanced:
Dexter Fowler
Seth Smith
TroyTulowitzki
Chris Ianetta
Garrett Atkins
Ryan Spilborghs
Carlos Gonzalez
Ian Stewart
This lineup scored 4.331 versus a LHP and 4.562 versus a RHP, for an average of 4.446.
That is .1 rpg better than the previous lineup, which is a lot: 1.6 wins a year.
Now, my sim includes base running and of course batting order makes a difference, so don’t put too much weight in this little experiment. I thought the results were fairly interesting though.
Friday, August 28, 2009
By , 06:31 PM
I’ve always wanted to have a blog or a web site devoted only to manager lineups. Seriously. I’ve also always wondered why managers change lineups more often than they change their socks. One day a player (like Gwynn) can bat leadoff and the next day he bats 8th (a huge difference, needless to say). Or one day player A bats 8th and player B bats 9th and the next day, the manager switches them. Why would he do that even though it probably makes no difference at all?
Anyway, here is an odd lineup to say the least. What possesses a manager to make an order like this? Any ideas? The current batting averages gives you a clue I think as to why the manager has this order. If that is the case, that is one of the more egregious misuses of BA and batting order you will find. I don’t know much about the FLO manager, but this suggests that he doesn’t have a clue (as to how to determine which players have what value going forward).
Florida Marlins
Player AVG HR RBI
C COGHLAN LF .293 9 36
W HELMS 3B .274 2 26
H RAMIREZ SS .364 19 85
J CANTU 1B .276 13 71
J BAKER C .268 8 39
D UGGLA 2B .242 23 67
C ROSS CF .269 20 69
J HERMIDA RF .267 13 47
C VOLSTAD -R P .133 0 3
Tuesday, August 25, 2009
As one reader pointed out, there’s nothing necessarily logical about any of the baseball rules. You can have 10 guys on the field, a 3-2 ball-strike count, 5 bases, all DHs, etc. So, from that standpoint, you can change any rule, as the only mainstay is that you have a pitcher on the mound and a batter at plate. Everything else is negotiable. That said, this was my response to Greg’s wild idea (quoted below):
Basically, it’s like any sport, where the offense dictates who he can pass/throw/kick to, etc.
The reason I’d be against it is that it becomes very specialized, the kind of thing I don’t like in football (either you are a receiver or a running back, etc).
That said, just because I find one thing I don’t like with it doesn’t mean that overall it’s necessarily worse!
Read More
Wednesday, June 24, 2009
I reply to a thread at Primer:
Batting the pitcher 8th, or moving everyone down a spot in the order and putting the guy you originally decided to bat 9th (when you moved the pitcher to 8th), into the 1 spot?
If this is the choice, it’s a no-brainer, and it’s the former. In no way can you put one of the worst hitters on the team at the top of the order.
***
As for the general issue, my research in The Book (see it for free on Amazon’s Look Inside) using Markov chains is that moving the pitcher from 9 to 8 will add roughly 2 runs in a 162 game season. MGL’s research via his simulator is that it’s a break-even or a slight advantage to keeping the pitcher in the 9th slot.
***
The most egregious thing you can do is move the pitcher to the cleanup slot. This would cost you 0.1 runs per game (about 16 runs in a season). Basically, moving the pitcher up the order costs you around 4 runs per slot. Move pitcher from 8 to 7 to 6 to 5 to 4 and remove 4 runs each rung. That’s the impact of a batting order. And remember, this is by far the worst hitter on the team. That’s the impact here.
I presume most people would think that moving the pitcher to the cleanup slot would cost you half a run a game.
Monday, April 27, 2009
Devil Fingers goes in-depth to come up with a Pads lineup against LHP and RHP. I like how he has it all laid out. Tough call when facing RHP, because their 3 best hitters are all LHH. It depends how much you want to split your lefties.
Sunday, April 12, 2009
By , 12:49 PM
Similar to what we did in The Book, I looked at how batter performance varied with times through the order. Specifically, I looked at starting pitchers only, the NL and AL only, for 2006-2008.
For each PA, I put it into one of 4 buckets: First time at bat, second time, third, and fourth or later. If, for example, a pinch hitter came in while the starting pitcher was facing the lineup for the third time, this PA did NOT go into the “3rd time through the order” bucket, since the batter had not faced this pitcher before in the game. In fact, pinch hitters were excluded completely. Pitcher batting was excluded as well.
Anyway, I compiled the stats in each bucket and compared them. Since the pools of pitchers and batters are obviously not equally represented in all 4 buckets (for example, better pitchers AND batters tend to dominate the “4th time” bucket), I adjusted the stats in each bucket by the average pitcher and average batter in that bucket. To do the adjustments, I used each pitcher’s and batter’s full year stats for that year.
The percentage of all the PA looked at that fell into each bucket were:
1 .354
2 .364
3 .251
4 .032
So only 3.2% of all PA by starting pitchers were against batters who had faced them at least 3 times prior in that game. I hope I did everything right, as that seems low.
Anyway, here are the individual components for each bucket, normalized to 1.00 where “1.00” represents an average component for all the buckets combined (basically, all PA by starting pitchers excluding pitcher hitting and pinch hitters). Each component is per PA, where PA excludes IBB and SH.
Foul Out
1st time: 1.009
2nd time: .996
3rd time: 1.041
4th+ time .940
Single
1st time: .996
2nd time: 1.012
3rd time: 1.03
4th+ time .994
Dbl+Trpl
1st time: .963
2nd time: 1.024
3rd time: 1.050
4th+ time 1.065
Dbl/(Dbl+Trpl)
1st time: 1.005
2nd time: 1.003
3rd time: .989
4th+ time .991
Trpl/(Dbl+Trpl)
1st time: .945
2nd time: .967
3rd time: 1.105
4th+ time 1.087
HR
1st time: .947
2nd time: 1.015
3rd time: 1.089
4th+ time 1.049
BB (No IBB)
1st time: 1.029
2nd time: 1.004
3rd time: .990
4th+ time 1.019
HBP
1st time: 1.036
2nd time: .959
3rd time: 1.044
4th+ time 1.04
SO
1st time: 1.106
2nd time: .980
3rd time: .928
4th+ time .923
ROE
1st time: .927
2nd time: 1.030
3rd time: 1.017
4th+ time .989
SF
1st time: .958
2nd time: 1.067
3rd time: 1.003
4th+ time 1.001
GDP
1st time: .975
2nd time: 1.035
3rd time: 1.024
4th+ time 1.006
IFS (infield single)
1st time: .972
2nd time: 1.025
3rd time: 1.035
4th+ time 1.059
Bunt Single
1st time: 1.024
2nd time: 1.014
3rd time: 1.009
4th+ time .871
WP
1st time: .894
2nd time: 1.029
3rd time: 1.068
4th+ time 1.505
PB
1st time: 1.077
2nd time: .979
3rd time: .933
4th+ time .901
wOBA
1st time: .338
2nd time: .347
3rd time: .355
4th+ time .350
I am not sure why the 4th+ time through the order, the offensive output actually goes down to 2nd-3rd time levels. Maybe the weather is a factor. Maybe the strategic nature of the PA’s at the end of the game. Maybe the selective sampling aspect that if starting pitchers are still in the game the 4th+ time through the order, they are pitching well that day or the run scoring environment is low.
Monday, April 06, 2009
There some chatter here about the #2 hitter, the role of bunting, and the difference in run environment.
I should note that MGL wrote the sac bunt chapter, not me. I did the batting order chapter. I think he used 2000-2005 as his data source, whereas I used 1999-2002.
In any case, I’ll answer any question in general-speak that anyone has. I invited the readers there to this thread.
Thursday, March 19, 2009
Responding to the primer thread:
So, the mantra “batting order doesn’t matter” is dead now? I’m having trouble keeping up.
The quantitative description is that an optimal batting order can get you 5-15 more runs in a season. (This has been the case since I started looking at this several years ago, and has not changed since.) How you want to qualitatively say that “a little” or “a lot” is up to you. The numeric description is more accurate in any case, and will not change. That’s the only thing you have to keep up with.
Also: remember that moving pitchers from the 8th slot (best) to 4th slot (worst) will cost your team 0.1 runs per game. That’s the breadth of impact in making the single worst decision that no one would ever make.
***
Your best hitter should *generally* bat 2nd or 4th. It could happen that he will bat 1st (Rickey, Raines), it could happen that he will bat 3rd. You’d have to look at your entire set of players.
However, it will be very very difficult to find a situation where your five best hitters will not bat somewhere in the top 5. This may seem obvious to everyone here, but I think it’s easy to find teams that put one of their 2 or 3 worst hitters in the 2-hole.
***
Also, since star hitters have huge egos, you have to defer to their inner child, as their petulance will easily evaporate the carefully crafted 5-15 run gain you can otherwise hope to get.
Tuesday, March 03, 2009
A good preview discussion to batting order in The Book.
Friday, April 04, 2008
Inspired by a post Don Malcolm on Retrolist, I queried all the Retro event files, and counted the % of times that a particular position batted 3rd, 4th, 5th and subtracted from that the times they were batting 7th, 8th or 9th:
-1.000 1
-0.449 2
-0.384 6
-0.207 4
0.116 5
0.156 8
0.417 7
0.423 9
0.552 10
0.617 3
This means that 1B batted in the heart of the order 61.7% more times than they batted at the bottom of the order (70.4% minus 8.7%).
The order resembles the Defense Spectrum.
Justin will give it a go. Look for an update soon…
Tuesday, March 04, 2008
Alfonso Soriano has big splits. Per 600 PA, comparing bases empty and men on base:
BE: 100 singles, 33 HR, 40 2b+3b, 27 NIBB
MO: 95 singles, 26 HR, 36 2b+3b, 29 NIBB
His wOBA are: .379 with bases empty and .344 with men on base. IIRC, the difference for the average player is a 5 point drop or so. I’m sure someone can correct me. But, he’s got a 35 point difference here (based on almost 3000 PA with bases empty and 2000 with men on base). One standard deviation is roughly a 15 point difference, so we see here a difference of around 2 standard deviations.
While that doesn’t necessarily mean that Soriano definitely prefers to bat with bases empty, it points very strongly toward that. Soriano would be a good case study for PITCHf/x: how often does he take/swing at each count? How often does he do so when a pitch is at the fringes of the strike zone (+/- 15 inches from the middle, excluding +/- 5 inches in the middle)? If he actually approaches the PA differently, then this would point to his sample data being more indicative of something real.
While I can generally agree with Joe “BP” Sheehan’s article, I have concerns with Soriano.
As for the pitcher batting 8th causing strategic headaches: if this forces a manager to pinch hit for him earlier a couple of times a year, this is a good thing. A reliever is usually a better bet than a starter his third time through the order. (One day, some MLB team will wakeup, and go with a 5-day rotation, where you have 3 regular starters for the 1st, 3rd, 4th day, and an all-relief rotation for the 2nd and 5th day. At the very least, do this in September, when you’ve got plenty of called up young guns who can easily go 1 or 2 innings.)
from Amazon
Recent comments
Older comments
Page 1 of 191 pages 1 2 3 > Last »Complete Archive – By Category
Complete Archive – By Date