SABR XXIV

ARLINGTON, TEXAS

JUNE 18, 1994


SPEED, STRIKEOUTS, (BASE ADVANCEMENT), AND SCORING

MARK D. PANKIN

INTRODUCTION


INVESTIGATE EFFECTS ON SCORING OF

FAST RUNNERS

TEAM STRIKEOUTS

ADVANCING RUNNER FROM 2nd TO 3rd


PRIMARY RESEARCH TOOL: MARKOV PROCESS MODEL

BASED ON TRANSITION PROBABILITIES BETWEEN THE RUNNERS & OUTS SITUATIONS

USES MATRIX ALGEBRA TO COMPUTE ALL POSSIBILITIES

FAST ALTERNATIVE TO SIMULATION

CALCULATES LONG-TERM AVERAGE RUNS PER GAME FOR A SPECIFIED LINEUP

APPLICATIONS:

* EFFECTS OF PLAYER PERFORMANCE ON SCORING

* STRATEGY ANALYSIS

* FINDING THE BEST BATTING ORDER

KEY ASSUMPTIONS


PLAYERS HIT THE SAME IN ALL SITUATIONS EXCEPT WALK FREQUENCY ADJUSTED FOR RUNNERS & OUTS SITUATIONS, PITCHER BATTING NEXT

ADVANCEMENT ON HITS AND OUTS BASED ON BATTER HAND AND RUNNER SPEED (1984-92 DATA)

DOUBLE PLAYS, REACHING ON ERRORS BASED ON BATTER HAND, BATTER AND RUNNER SPEEDS (84-92)

WP, PB, BALKS ACCORDING TO 1984-92 ML AVERAGES

SB ATTEMPTS WITH RUNNER ON FIRST ONLY

SACRIFICE BUNT TRIES BY PITCHERS ONLY

SPEED: RICKEY HENDERSON IN 1990


MVP IN 1990

1990 PERFORMANCE AND ALTERNATIVES
BA
OBP
SLG
HR
SB
CS
SB %
1990
.325
.439
.577
28
65
10
.87
RICKEY+
.325
.439
.577
28
108
19
.85
"CECIL"
.325
.439
.577
28
0
0
"VINCE"
.292
.340
.400
6
65
10
.87


STUDY METHOD: USE MODEL TO GET 1990 OAKLAND SCORING WITH EACH OF THE 4 ALTERNATIVES AS LEADOFF BATTER

USE 1990 AS THE BASELINE

COMPARE SCORING RESULTING FROM THE OTHER 3

LINEUP:

LEADOFF (4 ALTERNATIVES)
LANSFORD
CANSECO
McGWIRE
D. HENDERSON
STEINBACH
JOSE
WEISS
GALLEGO

SPEED AND SCORING ANALYSIS RESULTS


SCORING COMPARISON TO 1990 BASELINE
RUNS/GAME
RUNS/162 GAMES
RICKEY+
0.018
2.9
"CECIL"
-0.080
-12.9
"VINCE"
-0.464
-75.2


EXTRA 43 SB, 9 CS HAS LITTLE EFFECT (RICKEY+)

SLOW RUNNER, NO SB TRIES WOULD COST 1 GAME PER SEASON ("CECIL") BASED ON 10 RUNS PER WIN

RICKEY'S BATTING PERFORMANCE WORTH 7-8 WINS COMPARED TO "VINCE"

STRIKEOUTS: 1991 TIGERS


SET AL RECORD FOR TEAM Ks IN A SEASON: 1185

METHOD OF ANALYSIS: COMPARE MODEL ESTIMATES OF TEAM SCORING AT VARIOUS LEVELS OF STRIKEOUTS

USE 1991 FULL SEASON STATS (PHILLIPS, WHITAKER, TRAMMELL, FIELDER, TETTLETON, FRYMAN, DEER, INCAVIGLIA, CUYLER) AS THE BASELINE

SCALE BACK Ks TO 747 (TWINS Ks, LOWEST IN ML)

NO STRIKEOUTS AT ALL

ALSO ESTIMATE TWINS SCORING WITH 1185 Ks


SCORING COMPARISONS TO BASELINE (ASSUMES ONLY CHANGE IS IN NUMBER OF STRIKEOUTS)
RUNS/GAME
RUNS/162 GAMES
TIGERS: 747 Ks
0.106
17.2
TIGERS: 0 Ks
0.294
47.6
TWINS: 1185 Ks
-0.123
-19.9


BASE ADVANCEMENT: RUNNER ON 2nd, 0 OUTS (2,0)


USE TEAM CONSISTING OF ALL THE SAME BATTER

AL 1993 AVERAGE (LEAGUE TOTALS/14/9)

RICKEY HENDERSON IN 1990
BA
OBP
SLG
HR
K
SB
CS
AL 93 AVG
.267
.337
.408
16
103
12
7
RICKEY 1990
.325
.439
.577
28
60
65
10



COMPARE TO BASELINE SCORING LEVELS:

RUNNER ADVANCES ON EVERY NON-K OUT (1984-92 AVERAGE IS ABOUT 50%)

REDUCE BATTING AVERAGE (WALKS UNCHANGED) IN (2,0) ONLY UNTIL SCORING IS REDUCED TO BASELINE LEVEL

REPEAT PROCESS ASSUMING NEVER ADVANCE ON NON-K OUTS, AND INCREASE BATTING AVERAGE IN (2,0) ONLY UNTIL SCORING INCREASES TO BASELINE

ADVANCEMENT IN (2,0) RESULTS


BASELINE SCORING

ALL 1993 AL AVG TEAM: 5.032 RUNS PER GAME

ALL 1990 RICKEY TEAM: 10.357 RUNS PER GAME

SCORING WHEN ALWAYS ADVANCE (2,0) --> (3,1) ON NON­K OUTS

ALL AL AVERAGE TEAM

* WITH NO CHANGE IN BATTING AVG: 5.071 R/G (INCREASE OF 6 RUNS PER 162 GAMES)

* A DECREASE IN BATTING AVERAGE OF 22% IN (2,0) WILL REDUCE SCORING TO THE BASELINE LEVEL

ALL 1990 RICKEY HENDERSON TEAM

* WITH NO CHANGE IN BATTING AVG: 10.432 R/G (INCREASE OF 12 RUNS PER 162 GAMES)

* A DECREASE IN BATTING AVERAGE OF 12% IN (2,0) WILL REDUCE SCORING TO THE BASELINE LEVEL

ADVANCEMENT IN (2,0) RESULTS (CONTINUED)


SCORING WHEN NEVER ADVANCE (2,0) --> (3,1) ON NON­K OUTS

ALL AL AVERAGE TEAM

* WITH NO CHANGE IN BATTING AVG: 4.987 R/G (DECREASE OF 7 RUNS PER 162 GAMES)

* A INCREASE IN BATTING AVERAGE OF 20% IN (2,0) WILL RAISE SCORING TO THE BASELINE LEVEL

ALL 1990 RICKEY HENDERSON TEAM

* WITH NO CHANGE IN BATTING AVG: 10.265 R/G (DECREASE OF 15 RUNS PER 162 GAMES)

* A INCREASE IN BATTING AVERAGE OF 13% IN (2,0) WILL RAISE SCORING TO THE BASELINE LEVEL


BA
OBP
SLG
HR
K
SB
CS
AL 93 AVG
.267
.337
.408
16
103
12
7
" -22%
.208
.288
.318
12
103
12
7
" +20%
.320
.389
.490
19
103
12
7
RICKEY 1990
.325
.439
.577
28
60
65
10
" -12%
.286
.408
.508
25
60
65
10
" +13%
.367
.476
.652
32
60
65
10

SUMMARY AND CONCLUSIONS


HIGH SB/FAST RUNNERS DO NOT INCREASE SCORING BY VERY MUCH

A RICKEY HENDERSON ADDS ABOUT 1 WIN A SEASON COMPARED TO A NO SB/SLOW RUNNER

A LOW OBP/HIGH SB PLAYER DOES NOT CONTRIBUTE MUCH TO THE OFFENSE, AND HURTS SCORING WHEN BATTING LEADOFF

HIGH STRIKEOUT TEAMS SCORE LESS OTHER THINGS BEING EQUAL

DIFFERENCE OF ABOUT 2 WINS PER SEASON BETWEEN 1991 TIGERS AND 1991 TWINS FROM DIFFERENCE IN TEAM STRIKEOUTS

BUT TIGERS OFFENSE DEPENDS ON HR (LED 1991 AL WITH 209, 32 MORE THAN #2) AND HIGH OBP DUE TO WALKS (LED WITH 699, 52 MORE), BOTH OF WHICH LEAD TO HIGHER NUMBER OF STRIKEOUTS

ADVANCING RUNNER IN (2,0) IS WORTH SOME DECREASE IN BATTING AVERAGE

ROUGH ESTIMATE: FOR AN AVERAGE PLAYER, INCREASING THE ADVANCEMENT PROBABILITY BY 0.5 OVERCOMES 20% DECREASE IN BA IN (2,0) SITUATION

FOR BETTER HITTERS, THE CORRESPONDING DECREASE IN BA IS SMALLER; FOR POORER HITTERS, THAT DECREASE IS LARGER

ANALYSIS COULD BE REPEATED FOR SPECIFIC PLAYERS IN A REAL LINEUP

OBJECTIVE DURING GAME MAY BE TO SCORE ONE RUN RATHER THAN MAXIMIZE SCORING

Return to the Markov Page

Return to the Baseball Page

Return to the Home Page