Saturday, April 12, 2014

Event score percent

Data from retrosheet 2010-2013:


Walks will score slightly less than singles because they're not as evenly distributed. A pitcher is more likely to surrender a base on balls in safer situations--- when the baserunner will be less likely to score. Intentional walks even more so. I imagine some fraction of HBP are intentional as well, that may explain why they score ever so slightly less than a base hit, which is about as close to random as we'll probably get in this table.

Reached on Errors are not limited to single-base ROE, so that number is including some extra bases. Doubles and triples are especially high, and home runs are 100 percent obviously.

This is different, of course, from the percentage an event scores to lead off an inning which looks drastically different:



Saturday, June 15, 2013

Home run rates by inning, score

Addendum to a post at Hardball Times:
http://www.hardballtimes.com/average-leverage-index-of-events/

HR per inning 2010-2012


Inning HR HR/PA HR/Batted Ball
1 1636 0.026 0.036
2 1576 0.026 0.036
3 1539 0.025 0.034
4 1876 0.030 0.041
5 1645 0.026 0.036
6 1768 0.028 0.039
7 1550 0.025 0.035
8 1411 0.023 0.033
9 1078 0.023 0.034
10 114 0.021 0.034
11 61 0.021 0.032
12 25 0.017 0.026

Home runs by score 2010-2012


Run_diff HR HR/PA HR/Batted Ball
-7 157 0.028 0.038
-6 247 0.027 0.036
-5 368 0.026 0.036
-4 560 0.025 0.036
-3 802 0.025 0.036
-2 1163 0.025 0.035
-1 1700 0.025 0.036
0 3761 0.025 0.036
+1 1743 0.025 0.036
+2 1231 0.026 0.037
+3 835 0.027 0.038
+4 556 0.026 0.037
+5 358 0.026 0.037
+6 239 0.027 0.038
+7 143 0.027 0.037

Friday, April 12, 2013

pitch_seq_tx


The pitches field of the play record

synopsis: play,inning,home/visitor,player id,count,pitches,event
The fifth field, pitches, is a string of variable length and contains all pitches to this batter in this plate appearance. Most Retrosheet games do not have pitch data and consequently this field is blank for such games.

    +  following pickoff throw by the catcher
    *  indicates the following pitch was blocked by the catcher
    .  marker for play not involving the batter
    1  pickoff throw to first
    2  pickoff throw to second
    3  pickoff throw to third
    >  Indicates a runner going on the pitch
    B  ball
    C  called strike
    F  foul
    H  hit batter
    I  intentional ball
    K  strike (unknown type)
    L  foul bunt
    M  missed bunt attempt
    N  no pitch (on balks and interference calls)
    O  foul tip on bunt
    P  pitchout
    Q  swinging on pitchout
    R  foul ball on pitchout
    S  swinging strike
    T  foul tip
    U  unknown or missed pitch
    V  called ball because pitcher went to his mouth
    X  ball put into play by batter
    Y  ball put into play on pitchout

Tuesday, September 4, 2012

GB% by Base State

An interesting question raised at Fangraph's Q&A: Does GB% go up with a runner on first?

I queried retrosheet event files and sorted their recorded batted ball types by Base State and found the answer is: Apparently not.


BASE_STATE GB% FB% LD% PU%*
0 44.9 36.9 18.1 7.8
1 43.7 37.5 18.8 7.8
2 46.3 35.8 17.8 7.7
3 44.7 37.2 18 8
4 45.3 36.9 17.8 8
5 44.9 37.2 17.9 7.9
6 46.1 36.6 17.3 7.9
7 44.2 37.9 17.9 8.5


2002-2011, includes post-season data. Pop-ups are included in FB%, all bunts are excluded. 

I expanded on this table in a full article at BTB here:

http://www.beyondtheboxscore.com/2012/9/6/3295298/do-pitchers-induce-more-ground-balls-in-double-play-situations

Saturday, August 11, 2012

Use of Southpaws since 1960

Percent of Plate Appearances from right-handed pitchers vs left-handed pitchers since 1960:

Year RHP% LHP%
1960 73.81 26.19
1961 71.59 28.41
1962 72.42 27.58
1963 70.26 29.74
1964 69.39 30.61
1965 68.58 31.42
1966 68.76 31.24
1967 70.06 29.94
1968 71.27 28.73
1969 71.28 28.72
1970 69.57 30.43
1971 69.14 30.86
1972 69.4 30.6
1973 66.78 33.22
1974 67.45 32.55
1975 67.92 32.08
1976 69.66 30.34
1977 69.24 30.76
1978 66.71 33.29
1979 69.05 30.95
1980 68.92 31.08
1981 70.58 29.42
1982 69.92 30.08
1983 68.77 31.23
1984 69.09 30.91
1985 68.56 31.44
1986 68.67 31.33
1987 67.59 32.41
1988 69.51 30.49
1989 68.14 31.86
1990 66.13 33.87
1991 67.98 32.02
1992 69.29 30.71
1993 70.89 29.11
1994 72.1 27.9
1995 72.86 27.14
1996 74.18 25.82
1997 73.61 26.39
1998 73.58 26.42
1999 74.95 25.05
2000 75.44 24.56
2001 75.81 24.19
2002 74.78 25.22
2003 72.93 27.07
2004 72.98 27.02
2005 73.28 26.72
2006 72.92 27.08
2007 72.56 27.44
2008 71.74 28.26
2009 72.44 27.56
2010 71.88 28.12
2011 73.27 26.73


 Baseball saw its most southpaws in 1990 with 33.8% of all PA's and its least in 2001 with just 24.2%. Lefties have seen less playing time since the mid-nineties, but have been slowly on the rise over the past decade.

Thursday, June 28, 2012

Run Expectancy 2010-2011

As a follow-up to my post on Beyond the Boxscore, here is an updated Run Expectancy table generated from the 2010-2011 seasons:


BASE STATE 0 Outs 1 Outs 2 Outs
“---” 0.497 0.264 0.105
“1--” 0.888 0.518 0.236
“-2-” 1.110 0.689 0.330
“--3” 1.385 0.936 0.349
“12-” 1.441 0.893 0.460
“1-3” 1.810 1.118 0.486
“-23” 1.995 1.402 0.605
“123” 2.397 1.536 0.799

Tuesday, June 12, 2012

wBABIP by COUNT

We are all aware that different counts offer different advantages and disadvantages in the pitcher-batter match-up. But just how much of an advantage is a 3-1 count compared to an 0-2 count? 

Obviously these two situations offer different odds of a walk or a strikeout occurring, but I wanted to concern myself exclusively with how the count affects the ball off the bat.

Adding linear weights to the traditional BABip formula is something I think could be useful for this. For instance, if we took the 'generic' wOBA weights for total bases used in The Book, and then borrowed the denominator from the traditional BABIP formula, but included homeruns:

(0.90*1B + 1.24*2B + 1.56*3B + 1.95*HR) / (AB - SO + SF)

This gives us a better estimate of not only if the ball was converted to an out or not, but perhaps how well the ball was hit, by adjusting for extra bases, as well. We can then apply the results to the average wBABIP in all counts for reference:

COUNT wBABIP +/- AVG % of AVG
“3-0” 0.466 0.106 1.29
“3-1” 0.414 0.054 1.15
“2-0” 0.405 0.045 1.13
“1-0” 0.378 0.018 1.05
“3-2” 0.378 0.018 1.05
“2-1” 0.376 0.016 1.04
“0-0” 0.375 0.015 1.04
AVG 0.360 0 1.00
“1-1” 0.356 -0.004 0.99
“0-1” 0.342 -0.018 0.95
“2-2” 0.341 -0.019 0.95
“1-2” 0.328 -0.032 0.91
“0-2” 0.318 -0.042 0.88
... from 2002 - 2011, includes post-season, excludes all pitcher's PA.

BASE STATES

A look at how base-state effects the batter-pitcher match-up.
...
PITCHERS 
Base State PA FIP* BABIP K% BB% HR% GB% FB% HR/FB PPA IBB%
"---” 1045868 4.36 0.294 17.9 7.6 2.8 43 36.9 10.7 3.84 0.01
“1--” 342077 4.27 0.311 15.9 6.9 2.8 41.6 37.8 10.4 3.84 0.01
"-2-” 167165 4.53 0.293 16.9 10.3 2.2 44.4 35.9 9.5 4.03 3.71
“12-” 135831 4.21 0.290 17.5 7.7 2.6 42.7 37.5 10.2 3.77 0.05
“--3” 55109 4.67 0.301 17 10.7 2.2 43.7 36.8 9.1 3.98 3.58
“1-3” 58709 4.34 0.307 15.9 7.7 2.5 43 37.2 9.5 3.78 0.55
“-23” 42529 4.42 0.297 16.6 9.4 1.9 44.5 36.5 8.9 4 10.88
“123” 48959 4.06 0.296 17.9 6.7 2.6 42.8 38 9.5 3.69 0
*where FIP here is the 'generic' version of the metric, using 3.2 as the constant, and 'PPA' is pitches per plate appearance.
...


I was inspired to run this query after Tom Tango at The Book Blog mentioned that a pitcher is much more likely to walk a batter with a runner on 2nd rather than 1st. This is affirmed by the chart-- with the difference a little over 3%.  Similarly, pitchers attempt to avoid contact in the '--3' and '-23' base states when first base is also vacant and HR/FB drops as the pitcher is apparently living ouside-the-zone a touch more than usual.


This is naturally an advantage for the hitter as he sees a bump in his OBP and OPS therefore, but not so much with his SLG percentage:

BATTERS 
Base State PA wOBA* BA OBP SLG OPS
"---” 1045868 0.323 0.258 0.32 0.413 0.734
“1--” 342077 0.327 0.277 0.325 0.433 0.759
"-2-” 167165 0.311 0.252 0.36 0.394 0.754
“12-” 135831 0.312 0.254 0.314 0.405 0.718
“--3” 55109 0.323 0.277 0.373 0.429 0.802
“1-3” 58709 0.330 0.296 0.338 0.459 0.797
“-23” 42529 0.292 0.273 0.412 0.423 0.836
“123” 48959 0.319 0.281 0.316 0.445 0.761
...
*where wOBA is also the 'generic' version of the metric and does not include IBB or SB/CS.

It's also worth noting that BABIP is at it's highest when the first basemen is holding the runner. BABIP jumps about 17 points from bases empty to a state of "1--", but GB-BABip, in particular, sees a 20 point surge once right side of the infield is opened up:

BABIP
Base State GB_BABIP FB_BABIP LD_BABIP wBABIP*
"---” 0.216 0.195 0.707 0.356
“1--” 0.236 0.199 0.721 0.369
"-2-” 0.210 0.195 0.701 0.346
“12-” 0.207 0.196 0.710 0.350
“--3” 0.221 0.192 0.728 0.352
“1-3” 0.226 0.189 0.747 0.360
“-23” 0.220 0.189 0.725 0.348
“123” 0.219 0.187 0.730 0.353

*wBABIP is (0.90*1B + 1.24*2B + 1.56*3B + 1.95*HR) / (AB - SO + SF), 
GB excludes bunts, FB includes "pop ups"

I'm uncertain why GB-BAbip would be lowest of all the base states during "12-", but I'm open to suggestions. The 1B may be behind the runner in most of these situations, and the SS/2B may play deeper with the double play an option, but none of this should be an advantage over the bases empty state.

FB-BABip seems to fall slightly as the bases fill up, especially with a runner on third, which leads me to wonder if this is the effect of a slight shift in positioning as the outfielders feel less inclined to prevent the extra base hit rather than the bloop single.

I'm also going to speculate that LD-BABip reaches it's peak with both the 1B and the 3B holding the runners in a state of "1-3". Unlike ground balls, line drives are pulled more often than not (WAG), and with the majority of hitters being right-handed, the hot corner would see more LD-BIP than a first baseman.


Monday, January 23, 2012

More on Perfect FIP Games

David Wishinsky over at Athletics Nation proposed the idea of a 'defensive independent perfect game," a concept which he proposed would simply be a standard Complete Game with no walks or homeruns allowed. Tom Tango over at the Book blog then expounded on the idea, lobbying that the feat also require no hit batsmen and a FIP of 0.00 (requiring K*2/ IP  to be greater than 3.2).
In 9 innings of baseball that would require at least 15 strikeouts. Using the B-Ref play index, I found 12 performances that met this criteria:

FIP PERFECT GAMES