On SABR-L, Tom Hanrahan suggested taking a simulator and running it using the current SF Giant's statistics. Then, run the simulator with an intentional walk given to Barry Bonds whenever there is a base open. Reducing Bonds' "no base open" rate by an appropriate amount will create the same team totals, but will it affect expected runs? He also suggested doing this with a control group.
I don't have a simulator, but I do have some Markov matrices left
over from class.
I realized that I could modify the Markov matrices
to work out the expected values.
A few caveats: I've been unable to get Retrosheet data properly
analyzed on my upgraded computer, so I estimated the various advancement
parameters for groundouts and flyouts. I also left out errors and steals.
For singles, I used John Jarvis' advancement data at
http://knology.net/~johnfjarvis/stats.html
to estimate advancement on singles and doubles. (This should be refined
so that there is more advancement with 2 outs, and for left-handed batters)
I assumed that DP's occurred on 30% of the groundballs in DP situations,
which seemed to give a few to many DPs, but I guessed that the missing
stolen bases could account for that. I had SFs on half the fly/pops
that occurred in SF situations. This gave totals that seem a bit low,
but not terrible.
So, what I did was put in the Giants lineup using their current lineup.
For "pitchers" I used the average 2001 pitcher.('cuz that's what I had)
This then gave 162 game totals for each player, listed below.
The second set of stats is for players when Bonds is always walked
with runners on and first base open(RO1O).
As Tom suggested, I reduced his "no base open" walk rate.
RRF is "Runs Responsible For", rather than RBI - so this includes
runs that were sent in by a DP.
batter PA BB AB 1b 2b 3b HR R RRF SO SF GDP Lofton 792 90 700 125 39 12 12 106 54 86 2.4 11 Aurilia 774 49 720 114 47 3 20 109 76 120 4.6 17 Kent 754 58 692 127 47 3 41 133 118 112 4.8 15 Bonds 737 242 491 86 38 2 56 152 142 57 4.3 7 Santiago 725 38 681 125 34 7 23 89 125 103 6.4 26 Sanders 707 59 641 94 29 8 29 87 112 153 6.6 19 Snow 688 83 599 99 37 3 37 87 116 127 5.8 12 Bell 672 59 608 91 32 2 22 61 85 88 5.3 14 Pitchers 653 22 627 73 14 1 3 37 34 235 4.6 18 Total 6503 701 5757 933 316 41 243 861.7 1081 45 140Here, Bonds' had 76.498 expected occurances of RO1O.
XS
batter PA BB AB 1b 2b 3b HR R RRF SO SF GDP
Lofton 793 90 700 126 39 12 12 105 55 86 2.4 11
Aurilia 774 49 721 114 47 3 20 107 77 120 4.6 17
Kent 755 58 692 127 47 3 41 129 118 112 4.8 15
Bonds2 737 242 491 86 38 2 56 152 125 57 4.7 8
Santiago 726 38 681 125 34 7 23 88 130 104 6.6 25
Sanders 707 59 641 94 29 8 29 87 115 153 6.9 19
Snow 688 83 599 99 37 3 37 87 116 127 5.8 12
Bell 672 59 608 91 32 2 22 61 85 88 5.2 14
Pitchers 654 22 627 73 14 1 3 37 34 235 4.6 19
6508 701 5761 933 316 41 243 853.51 1082 46 140
The expected number of runs dropped by about 1%. Suggesting that
picking your spots to walk a batter makes a small difference. batter PA BB AB 1b 2b 3b HR R RRF SO SF GDP Lofton 787 89 695 124 39 12 12 105 54 85 2.4 11 Aurilia 769 49 715 114 46 3 20 108 76 119 4.6 17 Kent 749 57 687 126 46 3 41 131 117 112 4.7 15 Bonds2 731 183 544 95 42 3 62 149 156 64 4.7 8 Santiago 719 38 675 124 34 7 22 87 115 103 5.8 22 Sanders 701 59 636 93 29 8 29 86 105 152 5.9 18 Snow 683 83 595 98 36 3 36 86 111 126 5.4 12 Bell 667 59 603 90 32 2 22 61 83 87 5.1 14 Pitchers 648 22 622 72 14 1 3 37 33 233 4.5 18 Total 6454 639 5772 935 318 41 247 849.34 1080 43 135This Bonds has more RBI (and more HR due to more AB), but the team scores 4 fewer runs than if he were intentionally walked.
Using the average of the "regular stats" to generate a uniform lineup,
The run totals for hitters 1-9 were:
109,106,99, 97,96,93, 90,88,86
The RRF were
93,96,100, 104,101,96, 93,92,90
for a total of 864.26 runs.
Had this represented the totals with the cleanup hitter getting walked
with 1b open, the run/RRF totals were:
R : 109,104,96, 96,95,93, 91,88,86
RRF: 93,96,100, 88,106,100, 94,91,89
for a total of 856.32 runs.
Again, shifting walks to 1b-open drops about 1% off
"runs expected from these totals"
Had the clean-up hitter, with the orignal walk rate
been walked when 1b was open, his walks leap from
80 to 143,and the R/RRF are:
R : 111,107,100, 106,97,94, 92,89,87
RRF: 94,97,101, 82,118,109, 99,94,91
for a total of 883.09 runs.
So, walking Bonds appears to add about 4 runs per season. walking that "average" hitter would add about 19 runs/yr.
oddity: In 2001, Bonds and Snow had the same number of singles(49). They repeated this in 2002(70 singles).