Neutralizing Barry’s walks
The eternal battle over the strike zone waged between pitcher and batter is, at its heart, a question of trade-offs. The pitcher’s decision to lay it in there, or to keep it off the plate, is based on his judgment of whether the risk of allowing a hit, particularly an extra-base hit, is greater than the risk of allowing a walk. The batter’s decision to swing or take is based on his judgment of whether this pitch offers his best opportunity in the at-bat to get a hit, particularly an extra-base hit, against the risk of taking a called strike.
Back in the early 2000s, when Barry Bonds was hitting just about exactly like Babe Ruth, the league’s pitchers decided that the risk of walking him was vastly preferable to the risk of making him swing the bat. Bonds had demonstrated extraordinary plate discipline from a young age anyway, and this combined with pitchers’ extraordinary fear of him had Bonds, from 2001 through 2004, setting every record for walks—intentional, semi-intentional and otherwise.
The Bonds walkfest stimulated numerous discussions in the sabersphere regarding this collective decision by the National League’s pitching staffs. Yours truly was in the camp touting that the pitchers were doing their teams a net disservice: yes, by walking Bonds so regularly, they were limiting the damage he wrought with home runs and other run-producing hits, but every time they refused to pitch to him they guaranteed he would reach base. Force Bonds to swing the bat, some of us argued, and well over 60 per cent of the time he’ll make an out.
Certainly, every plate appearance isn’t the same as every other plate appearance. The base/out and score/inning situations significantly impact whether it would be sensible to pitch to Bonds, or anyone else. But on an overall “macro” basis, was my camp correct? Did pitchers walk Bonds too frequently? And if so, what precisely was the cost they paid?
Let’s see if we can figure it out.
Neutralizing the walks
The first thing is to decide what constitutes a “normal” walk rate.
Every batter walks sometimes; even the very weakest hitters draw a walk every once in a while. Clearly there’s a nominal rate of walks that’s simply a function of pitchers being human and missing the strike zone sometimes.
But above that baseline, walks are far from evenly distributed. Hitters demonstrate great differences in their intrinsic strike zone judgment, and beyond that, given two hitters with identical strike zone judgment, the one with the greater power will walk more often. Extra base hits (especially home runs) and walks are strongly correlated, for the obvious reason that it’s in the pitcher’s interest to be more careful with the hitter who can hurt him with the long ball. If the greatest damage a hitter is likely to inflict is hitting a single, there’s no reason not to pitch him right down the middle, but once the extra base hit enters the equation, the tradeoff value of the walk enters it as well.
Certainly in the case of a Bonds, this effect is extreme. But to some degree, all power hitters influence the pitcher’s willingness to allow a walk. So in assessing what might be a “normal” walk rate for a power hitter, we shouldn’t just take the league-average walk rate. We should use the walk rate exhibited by the average power hitter.
So let’s calculate the average walk rate presented by the top 100 hitters in career Slugging Percentage with at least 3,000 Plate Appearances, who’ve played most of their major league career since 1920. (Since 1920, because before then, home runs were so rare as to render the relationship between power hitting and walks different than it became with full “live ball” conditions in place.)
But wait a minute! When we do this, we discover something significant. The left-handed sluggers in this group draw, on average, a lot more walks than the right-handed sluggers with equivalent SLG.
Why would that be? Because, even though their production when swinging the bat is equivalent overall, these LHBs are more likely to be given a free pass since most pitchers are right-handed. Simply in seeking the platoon advantage, opposing teams are rationally more prone to walk (whether fully or semi-intentionally) the LHB than the RHB.
So we really need to determine two average walk rates: one from the top 50 lefty-swinging (and switch-hitting, since they always have the platoon advantage) sluggers, and another from the top 50 right-handed sluggers.
When we do so, we get two interesting lists, chock-full of bruising belters. The Bats Left/Bats Both group of 50 extends from Babe Ruth with his .690 career SLG, down to Ted Kluszewski at .498. The Bats Right bunch ranges from Albert Pujols at .624 to Orlando Cepeda and Bob Horner, tied at .499.
When Kluszewski, Cepeda, and Horner are the worst hitters around, you’re visiting an extremely heavy-hitting neighborhood. The BL/BB group hits .298 overall, with 31 homers per 162 games, while the righties hit .300 with 32. The overall Slugging Percentage for both groups is .536.
But the lefties have a walk rate of .128, compared to a rate of .108 for the RHBs. Those are the figures we’ll use as the “normal” or “neutralized” walk rates for power hitters.
Applying the neutralized walks
So let’s see how this works. Let’s take a left-handed-hitting slugger who draws a ton of walks—oh, how about a switch-hitter this time. How about Mickey Mantle in 1961?
This was Mantle’s batting line:
PA AB R H 2B 3B HR RBI BB SO BA OBP SLG OPS RC
646 514 131 163 16 6 54 128 126 112 .317 .448 .687 1.135 174
His walk rate was a heady .195. But let’s assume that, in exactly the same number of Plate Appearances, pitchers had given a free pass to Mantle at the average rate for non-right-handed power hitters, that is, .128. Let’s assume that in all those freed-up PAs in which The Mick isn’t now drawing a walk, he produces at precisely his established rate of hits, doubles, triples, homers, and strikeouts.*
This is the resulting walk-neutralized 1961 Mickey Mantle batting line:
PA AB R H 2B 3B HR RBI BB SO BA OBP SLG OPS RC
646 557 127 177 17 7 59 139 83 121 .317 .402 .687 1.088 154
This Commerce Comet is allowed to hit five additional home runs (now he’s right on Maris’s heels!), and drive in nearly a dozen more runs than he actually did. (For the formulae to estimate the new rates of runs scored and RBI, as well as all the other stats, see the References and Resources section below.) But in exchange, Mantle is getting on base a whole lot less often, as his OBP plummets from .448 to .402. Thus his OPS goes down by that same margin, and instead of compiling 174 Runs Created, he produces “only” 154.
It becomes quite clear just how much Yankees’ opponents were hurt in 1961 by allowing Mantle to draw 43 more walks than the average lefty-swinging power hitter.
Let’s try it with a right-hander. How about Frank Thomas, 1995.
Actual:
PA AB R H 2B 3B HR RBI BB SO BA OBP SLG OPS RC
647 493 102 152 27 0 40 111 136 74 .308 .454 .606 1.061 144
Walk-neutralized:
PA AB R H 2B 3B HR RBI BB SO BA OBP SLG OPS RC
647 559 96 172 31 0 45 126 70 84 .308 .385 .606 .991 127
Just like ’61 Mantle, our pitched-to Big Hurt delivers five more big flies than he did in reality, and racks up a bunch of additional RBI. But the huge reduction in walks yields a 70-point drop in OPS, and 17 fewer Runs Created.
We can do this in the opposite direction as well. Let’s do it for an unusually walk-averse lefty slugger, such as, say, Hal Trosky in 1936, actual and then walk-neutralized:
PA AB R H 2B 3B HR RBI BB SO BA OBP SLG OPS RC
671 629 124 216 45 9 42 162 36 58 .343 .382 .644 1.026 152
PA AB R H 2B 3B HR RBI BB SO BA OBP SLG OPS RC
671 579 129 199 41 8 39 149 86 53 .343 .428 .644 1.072 158
Our average-patience Trosky produces slightly less-gaudy power numbers than his actual swing-happy counterpart, but his OBP, OPS, and RC are all distinctly improved.
For a free-swinging RHB, why don’t we look at 1987 Andre Dawson. First is actual, then walk-neutralized:
PA AB R H 2B 3B HR RBI BB SO BA OBP SLG OPS RC
662 621 90 178 24 2 49 137 32 103 .287 .328 .568 .896 111
PA AB R H 2B 3B HR RBI BB SO BA OBP SLG OPS RC
662 582 94 167 22 2 46 128 71 96 .287 .370 .568 .938 119
The walk-neutralized Dawson would still have led the league in HR and RBI, but by narrow margins in both categories. An interesting question to ponder: would this Dawson—a demonstrably more productive hitter than the actual Hawk, but with slightly less-imposing power counting stats—still have won that season’s MVP vote?
Just for kicks, how about we re-cast the 1998 home run record chasers in this light.
Here’s Sammy Sosa:
PA AB R H 2B 3B HR RBI BB SO BA OBP SLG OPS RC
722 643 134 198 20 0 66 158 73 171 .308 .377 .647 1.024 149
PA AB R H 2B 3B HR RBI BB SO BA OBP SLG OPS RC
722 638 134 196 20 0 65 157 78 170 .308 .381 .647 1.028 157
And here’s Mark McGwire:
PA AB R H 2B 3B HR RBI BB SO BA OBP SLG OPS RC
681 509 130 152 21 0 70 147 162 155 .299 .470 .752 1.222 193
PA AB R H 2B 3B HR RBI BB SO BA OBP SLG OPS RC
681 597 125 178 25 0 82 173 74 182 .299 .380 .752 1.133 166
We see no significant difference between Sosa’s lines. He’d been a notorious hacker earlier in his career, but by this point Sammy was walking at just about exactly the average rate for a right-handed power hitter (and I’ve always felt that it was Sosa’s mid-career achievement of strike zone judgment, far more than any taboo substances he may or may not have been ingesting, that was the key to his breakthrough).
But McGwire’s line is radically altered: 82 dingers and 173 ribbies! This makes it clear that it was the enormous difference in walk rates between these two sluggers—McGwire’s was more than twice Sosa’s—that kept their home run totals fairly even. Give them equal walk rates, and McGwire would have blown Sammy away in terms of homers—yet Big Mac would have been a less productive batter than he actually was overall.
But, all right, enough of this frivolity. Let’s get to the main event.
The most respected hitters of all time
If walks are a valid indicator of the respect pitchers accord to hitters—oh heck, let’s just go ahead and call it “fear”—then these last three are in a league of their own when it comes to fearsomeness. What sort of numbers might they have compiled if their walk rates had been equal to the average of left-handed-hitting sluggers?
Ted Williams, actual:
Year PA AB R H 2B 3B HR RBI BB SO BA OBP SLG OPS RC 1939 677 565 131 185 44 11 31 145 107 64 .327 .436 .609 1.045 149 1940 661 561 134 193 43 14 23 113 96 54 .344 .442 .594 1.036 145 1941 606 456 135 185 33 3 37 120 147 27 .406 .553 .735 1.287 183 1942 671 522 141 186 34 5 36 137 145 51 .356 .499 .648 1.147 168 1946 672 514 142 176 37 8 38 123 156 44 .342 .497 .667 1.164 170 1947 693 528 125 181 40 9 32 114 162 47 .343 .499 .634 1.133 166 1948 638 509 124 188 44 3 25 127 126 41 .369 .497 .615 1.112 156 1949 730 566 150 194 39 3 43 159 162 48 .343 .490 .650 1.141 180 1950 416 334 82 106 24 1 28 97 82 21 .317 .452 .647 1.099 98 1951 675 531 109 169 28 4 30 126 144 45 .318 .464 .556 1.019 137 1952 12 10 2 4 0 1 1 3 2 2 .400 .500 .900 1.400 5 1953 110 91 17 37 6 0 13 34 19 10 .407 .509 .901 1.410 41 1954 526 386 93 133 23 1 29 89 136 32 .345 .513 .635 1.148 126 1955 417 320 77 114 21 3 28 83 91 24 .356 .496 .703 1.200 118 1956 503 400 71 138 28 2 24 82 102 39 .345 .479 .605 1.084 121 1957 546 420 96 163 28 1 38 87 119 43 .388 .526 .731 1.257 167 1958 517 411 81 135 23 2 26 85 98 49 .328 .458 .584 1.042 112 1959 331 272 32 69 15 0 10 43 52 27 .254 .372 .419 .791 45 1960 390 310 56 98 15 0 29 72 75 41 .316 .451 .645 1.096 95 Total 9791 7706 1798 2654 525 71 521 1839 2021 709 .344 .482 .634 1.116 2382
Ted Williams, walk-neutralized:
Year PA AB R H 2B 3B HR RBI BB SO BA OBP SLG OPS RC 1939 677 585 129 192 46 11 32 150 87 66 .327 .414 .609 1.023 147 1940 661 572 133 197 44 14 23 115 85 55 .344 .431 .594 1.024 145 1941 606 525 130 213 38 3 43 138 78 31 .406 .485 .735 1.220 185 1942 671 581 135 207 38 6 40 153 86 57 .356 .443 .648 1.091 164 1946 672 584 135 200 42 9 43 140 86 50 .342 .429 .667 1.096 166 1947 693 601 118 206 46 10 36 130 89 54 .343 .429 .634 1.063 162 1948 638 553 119 204 48 3 27 138 82 45 .369 .453 .615 1.068 153 1949 730 635 143 217 44 3 48 178 93 54 .343 .429 .650 1.079 176 1950 416 363 79 115 26 1 30 105 53 23 .317 .405 .647 1.051 95 1951 675 589 103 187 31 4 33 140 86 50 .318 .406 .556 .961 133 1952 12 10 2 4 0 1 1 3 2 2 .400 .477 .900 1.377 4 1953 110 96 17 39 6 0 14 36 14 11 .407 .483 .901 1.384 42 1954 526 455 87 157 27 1 34 105 67 38 .345 .428 .635 1.063 123 1955 417 358 74 127 23 3 31 93 53 27 .356 .439 .703 1.142 109 1956 503 438 68 151 31 2 26 90 64 43 .345 .430 .605 1.035 113 1957 546 469 93 182 31 1 42 97 70 48 .388 .472 .731 1.203 158 1958 517 443 78 145 25 2 28 92 66 53 .328 .418 .584 1.002 106 1959 331 282 31 71 16 0 10 45 42 28 .254 .350 .419 .769 41 1960 390 335 54 106 16 0 31 78 50 44 .316 .408 .645 1.053 86 Total 9791 8474 1729 2922 577 77 576 2025 1253 777 .345 .426 .635 1.061 2295
In our imaginary universe, The Thumper delivers six seasons with 200 or more hits—in reality he delivered none—and approaches 600 career home runs despite all the military service. He reaches the 40-homer mark five times, while in reality he enjoyed only one such season.
But he also makes some 550 more outs than the actual Williams, and thus his career Runs Created total is reduced by nearly 100.
Babe Ruth, actual:
Year PA AB R H 2B 3B HR RBI BB SO BA OBP SLG OPS RC 1914 10 10 1 2 1 0 0 2 0 4 .200 .200 .300 .500 1 1915 103 92 16 29 10 1 4 21 9 23 .315 .376 .576 .952 20 1916 150 136 18 37 5 3 3 15 10 23 .272 .322 .419 .741 18 1917 142 123 14 40 6 3 2 12 12 18 .325 .385 .472 .857 22 1918 380 317 50 95 26 11 11 66 58 58 .300 .411 .555 .966 72 1919 542 432 103 139 34 12 29 114 101 58 .322 .456 .657 1.114 128 1920 616 458 158 172 36 9 54 137 150 80 .376 .532 .847 1.379 200 1921 693 540 177 204 44 16 59 171 145 81 .378 .512 .846 1.359 229 1922 495 406 94 128 24 8 35 99 84 80 .315 .434 .672 1.106 116 1923 699 522 151 205 45 13 41 131 170 93 .393 .545 .764 1.309 209 1924 681 529 143 200 39 7 46 121 142 81 .378 .513 .739 1.252 194 1925 426 359 61 104 12 2 25 66 59 68 .290 .393 .543 .936 75 1926 652 495 139 184 30 5 47 146 144 76 .372 .516 .737 1.253 185 1927 691 540 158 192 29 8 60 164 137 89 .356 .486 .772 1.258 201 1928 684 536 163 173 29 8 54 142 137 87 .323 .463 .709 1.172 173 1929 587 499 121 172 26 6 46 154 72 60 .345 .430 .697 1.128 148 1930 676 518 150 186 28 9 49 153 136 61 .359 .493 .732 1.225 183 1931 663 534 149 199 31 3 46 163 128 51 .373 .495 .700 1.195 184 1932 589 457 120 156 13 5 41 137 130 62 .341 .489 .661 1.150 147 1933 575 459 97 138 21 3 34 103 114 90 .301 .442 .582 1.023 116 1934 471 365 78 105 17 4 22 84 104 63 .288 .448 .537 .985 86 1935 92 72 13 13 0 0 6 12 20 24 .181 .359 .431 .789 11 Tot. 10617 8399 2174 2873 506 136 714 2213 2062 1330 .342 .474 .690 1.164 2718
Babe Ruth, walk-neutralized:
Year PA AB R H 2B 3B HR RBI BB SO BA OBP SLG OPS RC 1914 10 9 1 2 1 0 0 2 1 3 .200 .302 .300 .602 1 1915 103 88 16 28 10 1 4 20 13 22 .315 .397 .576 .973 20 1916 150 127 19 34 5 3 3 14 19 21 .272 .358 .419 .777 19 1917 142 117 15 38 6 3 2 11 18 17 .325 .396 .472 .867 22 1918 380 326 49 98 27 11 11 68 49 60 .300 .391 .555 .946 70 1919 542 464 100 149 36 13 31 122 69 62 .322 .415 .657 1.073 123 1920 616 529 155 199 42 10 62 158 79 92 .376 .456 .847 1.303 202 1921 693 596 175 225 49 18 65 189 89 89 .378 .459 .846 1.306 229 1922 495 427 92 135 25 8 37 104 63 84 .315 .402 .672 1.074 115 1923 699 603 146 237 52 15 47 151 89 107 .393 .473 .764 1.237 215 1924 681 584 139 221 43 8 51 134 87 89 .378 .459 .739 1.198 195 1925 426 363 61 105 12 2 25 67 55 69 .290 .380 .543 .923 74 1926 652 556 135 207 34 6 53 164 83 85 .372 .450 .737 1.187 182 1927 691 589 155 209 32 9 65 179 88 97 .356 .431 .772 1.203 196 1928 684 585 159 189 32 9 59 155 88 95 .323 .409 .709 1.118 168 1929 587 496 121 171 26 6 46 153 75 60 .345 .424 .697 1.122 145 1930 676 567 146 204 31 10 54 168 87 67 .359 .431 .732 1.163 178 1931 663 577 145 215 34 3 50 176 85 55 .373 .454 .700 1.154 183 1932 589 512 115 175 15 6 46 153 75 69 .341 .428 .661 1.089 144 1933 575 499 93 150 23 3 37 112 74 98 .301 .393 .582 .975 113 1934 471 409 73 118 19 4 25 94 60 71 .288 .382 .537 .919 83 1935 92 80 12 14 0 0 7 13 12 27 .181 .285 .431 .716 10 Tot. 10617 9102 2122 3121 550 148 779 2408 1359 1441 .343 .422 .693 1.115 2660
Remember that tremendous Mickey Mantle 1961 season we were looking at earlier, the year in which Mantle hit .317 with 54 homers? Mantle’s Slugging Percentage that season was a fat .687, which not only led the majors in 1961, it would be the highest SLG produced by any hitter in any season between 1957 and 1994.
Think about this: Ruth’s Slugging Percentage was higher than that.
For his career.
Check out what kind of counting stats the walk-neutralized Bambino would pile up. Seven seasons with 200 or more hits (with another one at 199). Eight seasons of 50 or more homers. Ten seasons of 150 or more RBI.
My personal favorite has to be 1921: 132 extra-base hits! 175 runs scored, and 189 RBI!
This Sultan of Swat would bow out with well over 3,000 hits, and no fewer than 779 career home runs, beyond Mr. Aaron’s tremendous reach. Yet for all that additional damage, taking more than 700 walks away from him renders this Ruth with more than 50 fewer career Runs Created than the actual Ruth.
And now, Mr. Bonds.
Barry Bonds, actual:
Year PA AB R H 2B 3B HR RBI BB SO BA OBP SLG OPS RC 1986 484 413 72 92 26 3 16 48 65 102 .223 .330 .416 .746 64 1987 611 551 99 144 34 9 25 59 54 88 .261 .329 .492 .821 93 1988 614 538 97 152 30 5 24 58 72 82 .283 .368 .491 .859 100 1989 679 580 96 144 34 6 19 58 93 93 .248 .351 .426 .777 92 1990 621 519 104 156 32 3 33 114 93 83 .301 .406 .565 .970 128 1991 634 510 95 149 28 5 25 116 107 73 .292 .410 .514 .924 118 1992 612 473 109 147 36 5 34 103 127 69 .311 .456 .624 1.080 148 1993 674 539 129 181 38 4 46 123 126 79 .336 .458 .677 1.136 172 1994 474 391 89 122 18 1 37 81 74 43 .312 .426 .647 1.073 115 1995 635 506 109 149 30 7 33 104 120 83 .294 .431 .577 1.009 134 1996 675 517 122 159 27 3 42 129 151 76 .308 .461 .615 1.076 162 1997 690 532 123 155 26 5 40 101 145 87 .291 .446 .585 1.031 151 1998 697 552 120 167 44 7 37 122 130 92 .303 .438 .609 1.047 153 1999 434 355 91 93 20 2 34 83 73 62 .262 .389 .617 1.006 91 2000 607 480 129 147 28 4 49 106 117 77 .306 .440 .688 1.127 155 2001 664 476 129 156 32 2 73 137 177 93 .328 .515 .863 1.379 230 2002 612 403 117 149 31 2 46 110 198 47 .370 .582 .799 1.381 208 2003 550 390 111 133 22 1 45 90 148 58 .341 .529 .749 1.278 166 2004 617 373 129 135 27 3 45 101 232 41 .362 .609 .812 1.422 203 2005 52 42 8 12 1 0 5 10 9 6 .286 .404 .667 1.071 12 2006 493 367 74 99 23 0 26 77 115 51 .270 .454 .545 .999 98 2007 477 340 75 94 14 0 28 66 132 54 .276 .480 .565 1.045 99 Tot. 12606 9847 2227 2935 601 77 762 1996 2558 1539 .298 .444 .607 1.051 2892
Barry Bonds, walk-neutralized:
Year PA AB R H 2B 3B HR RBI BB SO BA OBP SLG OPS RC 1986 484 416 71 93 26 3 16 48 62 103 .223 .324 .416 .740 55 1987 611 527 103 138 33 9 24 56 78 84 .261 .358 .492 .850 92 1988 614 531 98 150 30 5 24 57 79 81 .283 .376 .491 .866 97 1989 679 586 95 146 34 6 19 59 87 94 .248 .344 .426 .770 85 1990 621 533 102 160 33 3 34 117 79 85 .301 .391 .565 .955 116 1991 634 536 92 157 29 5 26 122 81 77 .292 .382 .514 .895 103 1992 612 522 104 162 40 6 37 114 78 76 .311 .402 .624 1.026 128 1993 674 579 126 194 41 4 49 132 86 85 .336 .420 .677 1.097 163 1994 474 404 88 126 19 1 38 84 61 44 .312 .407 .647 1.054 103 1995 635 545 105 160 32 8 36 112 81 89 .294 .389 .577 .966 120 1996 675 582 116 179 30 3 47 145 86 85 .308 .395 .615 1.010 141 1997 690 589 117 172 29 6 44 112 88 96 .291 .389 .585 .974 130 1998 697 593 116 179 47 8 40 131 89 99 .303 .398 .609 1.006 139 1999 434 372 89 98 21 2 36 87 56 65 .262 .360 .617 .977 81 2000 607 519 126 159 30 4 53 115 78 83 .306 .395 .688 1.083 139 2001 664 568 127 186 38 2 87 163 85 111 .328 .425 .863 1.288 200 2002 612 523 112 193 40 3 60 143 78 61 .370 .463 .799 1.262 185 2003 550 468 107 159 26 1 54 108 70 70 .341 .440 .749 1.188 146 2004 617 526 123 190 38 4 63 142 79 58 .362 .457 .812 1.269 187 2005 52 44 8 13 1 0 5 11 7 6 .286 .372 .667 1.038 11 2006 493 419 69 113 26 0 30 88 63 58 .270 .380 .545 .925 82 2007 477 411 68 114 17 0 34 80 61 65 .276 .374 .565 .938 85 Tot.12606 10791 2161 3240 661 83 857 2225 1614 1676 .300 .395 .615 1.010 2556
Walk Bonds at the average rate of lefty sluggers, and he tosses aside the career home run marks of the mere Ruths and Aarons like Godzilla laying waste to a Tokyo commuter train. This Barry reaches the 500 career homer mark in 2000, and surpasses 600 in ’01, 700 in ’03, and 800 in ’06 on his way to a staggering final total of 857.
Walk-neutralized Mark McGwire above had belted 82 in 1998, but walk-neutralized Bonds would best that by five in 2001. The fewest homers this Bonds would produce in the five-season span of 2000 through 2004 would be 53; his average home run output over that half-decade period would be 63.
But this devastating longball carnage inflicted on their opponents would come at a dramatic cost to the Giants. In that five-year span, walk-neutralized Bonds would make over 300 additional outs in the same number of plate appearances as the actual Bonds. Thus despite the tremendous tally of hits, homers, and runs batted in, this Bonds would create over 100 fewer runs for his team in those seasons than the actual Bonds. Over his full career, walk-neutralized Bonds would compile almost 350 fewer Runs Created than the actual Bonds.
It sure might seem counterintuitive, but there it is. Through the heart of Bonds’s awesome offensive peak, the fear he generated in opposing pitchers caused them to present the Giants with the gift of about 20 extra runs per year—in other words, about two wins per year—simply by walking him so vastly more frequently than a left-handed-hitting slugger typically walks.
* For sure, one can take issue with the assumption that a hitter’s production of all non-walk outcomes would continue at exactly the same rate if he walked less frequently. To the extent that a hitter would be swinging at pitches outside the strike zone in these walk-erased at-bats, it’s almost certain that his production rate of hits would decline, probably dramatically, and his strikeouts increase.
But in our thought experiment we aren’t assuming the hitter is swinging at ball four. We’re assuming the pitchers in these PAs are providing him with the same manner of pitches they did in his actual PAs when he didn’t walk. We really aren’t imagining the hitter changing his approach so much as we’re imagining the pitchers changing theirs. So for our purposes, the assumption of constant rate of all non-walk production is reasonable.
References & Resources
Formulae:
The first step is to neutralize the batter’s walk total: if Bats Left or Bats Both, it’s (Plate Appearances * .128); if Bats Right, then it’s (Plate Appearances * .108).
Everything else then keys off of this:
Neutralized AB = ((Actual AB) + (Neutralized BB – Actual BB))
Neutralized R = (Actual R * ((Neutralized BB + Neutralized TB + Neutralized HBP) / (Actual BB + Actual TB + Actual HBP)))
Neutralized H = (((Actual H / Actual AB) * (Neutralized AB))
Neutralized 2B = (((Actual 2B / Actual AB) * (Neutralized AB))
Neutralized 3B = (((Actual 3B / Actual AB) * (Neutralized AB))
Neutralized HR = (((Actual HR / Actual AB) * (Neutralized AB))
Neutralized RBI = (Actual RBI * (Neutralized TB / Actual TB))
Neutralized SO = (((Actual SO / Actual AB) * (Neutralized AB))
Neutralized HBP = (((Actual HBP / Actual AB) * (Neutralized AB))
Does RC account for the quality of other players in the lineup? Isn’t drawing walks in front of Jeff Kent going to produce a different number of runs than drawing it in front of Benito Santiago?
One of the things an analysis like this overlooks is the walk situation. I think pitchers are more likely to walk batters in high-leverage situations, where a base hit/home run would do a lot more damage.
Tango used Win Expectancy tables to create a chart of situations in which it made sense to walk Barry Bonds or not. This is the one for Giant home games:
http://www.tangotiger.net/walkbondschart2.html
There might be some reason to think that fewer walks would mean swinging at pitches outside the zone, but it might also mean that the batter is getting more strikes from weaker pitchers who were in real life told not to throw strikes to the slugger, and the batter might do better than his average marks on them. This would especially be true for LHB’s Ruth and Williams who were in the pre-relief specialist days and might have been walked in late innings by a RH starter they had already victimized.
“Does RC account for the quality of other players in the lineup?”
No, it’s an overall average.
“Isn’t drawing walks in front of Jeff Kent going to produce a different number of runs than drawing it in front of Benito Santiago?”
Yes, but that’s really just another way of saying that better-hitting lineups score more runs than poorer-hitting lineups.
All of these points are entirely well-taken. As my caveat says, there are plenty of specific situations in which the IBB (or semi-IBB) make perfect sense for Bonds or nearly all other hitters.
And the calculations I’m offerring are by all means nothing more than crude estimates. But I find the differences in OPS and RC so great in the cases of extremely-high walks totals that they strongly suggest that, on an overall macro basis, we shouldn’t assume that the opponents of a Bonds, Ruth, or Williams got it exactly right. An OBP of 1.000 in a meaningful proportion of PAs is an exceptionally powerful thing.
One problem that I have with this study is that I don’t think that Bonds would have necessarily had a “normal” number of walks even if pitchers hadn’t walked him intentionally or semi-intentionally so often—he would have still had an above-normal walk rate (even for a power hitter) IMO. Even in his first year in the majors, he had an he walked a lot for someone with the overall hitting stats he had. Don’t get me wrong—if pitchers hadn’t intentionally avoided throwing him strikes to the extent that they did, his walks would have dropped and that would have lead to some decrease in his runs created, but I don’t think it would have been as much as the study suggests.
“Don’t get me wrong—if pitchers hadn’t intentionally avoided throwing him strikes to the extent that they did, his walks would have dropped and that would have lead to some decrease in his runs created, but I don’t think it would have been as much as the study suggests.”
Completely agreed. Allow me to clarify the point of the study: I’m not suggesting that the difference between Bonds’s (or Ruth’s or Williams’s or McGwire’s or Mantle’s) actual walks and his neutralized walks is 100% explained by the manner in which he was pitched. Certainly, some (and perhaps quite a bit) of the difference is explained by Bonds’s particular skill at strike zone judgment.
But obviously not all of it is. Many of the walks Bonds took were egregiously semi-intentional (four straight changeups in the dirt type walks), and obviously a record-shattering number were fully intentional.
So the neutralizing walks exercise isn’t meant to estimate how many walks a Bonds “would have” drawn had pitchers pitched him the way they pitched the average LHB power hitter. It’s to estimate how large the difference is between the walk-driven productivity of an average power hitter and the walk-driven productivity of an extreme outlier such as Bonds.
Some of it (who knows how much) should be credited to Bonds’s walk-drawing skill. But some of it should also be debited from the opponents’ tactical wisdom.
Its been done…
http://baseballinsomnia.com/Baseball_Insomnia/Whats_The_Story/Entries/2009/11/11_How_Many.html
Its been done verry good admin. Thanks !!!
Fascinating. It’s not hard to imagine this study coming to the attention of a few smart GMs around baseball, but how they might react to it—and whether they could implement any significant changes—is another story.
Even if those GMs wanted to alter the approach of pitchers throughout their organization by urging them to challenge hitters more often and convincing them that the increase in long balls would actually be offset by the larger increase in outs made, you still would have human psychology to deal with.
Under pressure of game situations, with macho prowess on the line, I suspect that the more primitive aspects of the psyche would prevail over the rational approach. Home runs have a powerful emotional impact for both hitters and pitchers (not to mention fans), and the stats we see in real life reflect that.
Could it be changed? Maybe, but it’s much more than just a matter of explaining these numbers to your pitching staff. It’s the caveman mentality that has to be overcome. With a bunch of jocks. Good luck with that.
“Certainly, every plate appearance isn’t the same as every other plate appearance. The base/out and score/inning situations significantly impact whether it would be sensible to pitch to Bonds, or anyone else. But on an overall “macro” basis, was my camp correct? Did pitchers walk Bonds too frequently? And if so, what precisely was the cost they paid?”
___________________________________________________
I want to highlight this point that you made. While I think the macro analysis is meaningful, we need to be careful not to assume that the IBB is always wrong. Wouldn’t it be useful to compare the actual number of runs scored in every situation in which Bonds was walked to anticipated runs scored had he been pitched to and also to anticipated runs scored after he was put on base?
For example, in one infamous case, Joe Maddon intentionally walked Josh Hamilton with the bases loaded because in Maddon’s words, he thought Hamilton was too “toasty” right then. So we know that one run scored as a result. Given the pitcher, the batter, the on-deck hitter and the situation and even ignoring for a moment the possibility that a player on a hot streak might be more dangerous than he ordinarily is, how many runs should we anticipate would have scored had Hamilton been pitched to and how many after he was walked? I think only that one run scored and the Rays won, so while that is results based analysis, it still is meaningful if over an entire career similar results occur.
Steve Treder,
I disagree that the Kent vs Santiago hitting behind Bonds is simply saying that better lineups score more runs. If Benito Santiago is less likely to drive in baserunners than Jeff Kent, or even to get on base and keep the inning alive, then the balance of Runs Created is shifted toward Bonds needing to be the man to drive in the runs. Opposing pitchers can exploit that by walking Bonds.
As Giants fan who watched in abject frustration over the last few years of Barry’s career as promising rally after promising rally was snuffed out because he and the runners on base in front of him were left stranded by the hitters behind him, I can attest to the effectiveness of that strategy.
RC is a useful stat, but to use it properly, you have to realize that it assumes a normalized lineup. To say with certainty that it was bad strategy to walk Barry Bonds, a calculation that includes lineup context will be necessary.
“I disagree that the Kent vs Santiago hitting behind Bonds is simply saying that better lineups score more runs. If Benito Santiago is less likely to drive in baserunners than Jeff Kent, or even to get on base and keep the inning alive, then the balance of Runs Created is shifted toward Bonds needing to be the man to drive in the runs.”
Sure, but to what degree? While as a fellow Giants fan I deeply wish Benito Santiago had been a better hitter than he was, the fact is that he wasn’t exactly Bill Bergen out there.
“To say with certainty that it was bad strategy to walk Barry Bonds, a calculation that includes lineup context will be necessary.”
Okay, but perhaps many readers aren’t aware that Bill James performed a study years ago in which he placed 1921 Babe Ruth (oh, yeah!) within a lineup consisting of nothing but abject, utter stiffs, and ran a series of computerized simulations calculating how many runs this lineup would have scored with Ruth walking as frequently as he actually did, versus Ruth being walked 100% of the time—thus completely removing every single, double, triple, and home run Ruth could have contributed.
The result was that the 100%-walked Ruth lineup would score significantly more runs.
On-base percentage really, really matters in run production, a whole lot.
Steve Treder,
Maybe I’m not as impressed by computer simulations as you are. The outcome is greatly dependent on the assumptions in the model you give the computer.
I wonder if Bill James tried this scenario?: The 2 or 3 hitters above Babe in the lineup all have high OBP’s, but then there is a big dropoff in both OBP and power in the 2-3 hitters behind the Babe. That would actually be a lot closer to the situation Barry Bonds found himself in.
Steve, I think that many of us are saying that your conclusion…
Through the heart of Bonds’s awesome offensive peak, the fear he generated in opposing pitchers caused them to present the Giants with the gift of about 20 extra runs per year—in other words, about two wins per year—simply by walking him so vastly more frequently than a left-handed-hitting slugger typically walks.
…can’t be supported by your analysis. The runs, maybe, but not the wins. Looking at Tango’s walk/don’t walk chart, over 30% of plate appearances in close games (3 runs or less) indicate that the opposing team should “go with its gut, walk Bonds or DEFINITELY walk Bonds.” Nine percent indicate walk or DEFINITELY walk.
If I’m interpreting your tables correctly, Bonds was walked 944 times more than your standard. That’s seven percent of all plate appearances—less than the nine percent and WAY less than the 30%. If you estimate that about 67% of games are settled by three runs or less, then seven percent appears to be about right.
However, you could argue that opponents didn’t walk Bonds enough!
“If I’m interpreting your tables correctly, Bonds was walked 944 times more than your standard. That’s seven percent of all plate appearances—less than the nine percent and WAY less than the 30%. If you estimate that about 67% of games are settled by three runs or less, then seven percent appears to be about right.”
Well, that’s 944 times, or 7%, more than the average LHB power hitter over his entire career. From 2001 onward, his percentage was vastly higher than that.
And remember that we’re not talking about his percentage being higher than zero walks, but being higher than an already-higher-than-average walk rate: the walk rate of LHB power hitters, the most dangerous category of all hitters.
Sure, but there is play in that 7% figure, all the way up to 25% or so, depending on the matchup. Doesn’t change my issue with your conclusion.
Your (*) disclaimer is up-front, but it means this computation just can’t address the question “Did pitchers walk Bonds too frequently?” If we say that those walks convert into his non-walked rate stats, we’re assuming the conclusion. We already know he didn’t hit for the 1.000/1.000 (OBP/SLG) that a walk gives him.
You said there “We’re assuming the pitchers in these PAs are providing him with the same manner of pitches they did in his actual PAs when he didn’t walk.” In terms of what-if strategy, they can’t just do that.
For one thing, those BB-outcome pitchers are probably worse pitches, and more righties, as people mentioned above. But there’s also a more fundamental problem.
We’re selectively changing history on just the BB outcomes. The pitcher can’t see the BB outcome and then do it over differently. He needs to throw each pitch with the goal of avoiding a BB. Take for example a full count, where he aimed for the corner, missed, BB. What if that walk hadn’t happened? Well, what made it hypothetically not happen? “Aim for that corner *and hit it*” is not a how-to-pitch-to-Bonds strategy we can prescribe. What can we really say besides take a little off, throw it nearer the middle? And that’s going to give different rate stats in the non-BB outcomes.
To translate away some BBs, we need to turn them into stats that reflect “what if pitchers were changing their approach so those BBs wouldn’t have happened?” Which are different that the historical non-BB-outcome stats.
Head exploded
Awesome article
“To translate away some BBs, we need to turn them into stats that reflect “what if pitchers were changing their approach so those BBs wouldn’t have happened?” Which are different that the historical non-BB-outcome stats.”
OK, fine. In what way would they be different, exactly?
My assumption, as articulated in the (*) portion of the article, is that for purposes of this exercise, constant production as already presented in non-BB PAs is reasonable.
HOW different should they be from that? In what way?
I would be quite ready to incorporate different rates than the constants, but I don’t see a reliable basis for imputing anything else.
Certainly, the constants are an imperfect estimate. But in the absence of anything better, this is the best we can work with.
And, clearly, the bottom line is that futzing with these rates of production in the newly-non-BB PAs is farting around at the margins. Unless one is willing to assume (without underlying basis) dramatic changes in production rates in these PAs, one way or the other, then the outcome of the exercise won’t be meaningfully changed.
I think part of what Studes is implying is that you can’t just look at this on a macro (season or longer) level. You should be looking at individual PAs. Specially derived RE24 tables for the relevant players would be one way to look at it. Another would be to use Markov chains or some similar form to model the situation. Again specialized for the players involved.
“HOW different should they be from that? In what way?”
How different a strategy are you giving the pitchers, to avoid these walks? What strategy? Then we can talk about how much it affects the outcomes.
Oh, I’m not saying it’s really an answerable question. But if we tell a pitcher to avoid walking guys, and we don’t bump up his smarts or his stuff, they’ll hit him some unknown amount better, right? Either he’s taking something off, or he’s pitching to a smaller zone, or what am I missing?
So I don’t see it’s a matter of assuming changes “one way or the other”. I think we can be pretty sure which way, we just don’t know how much.
To be clear, we *don’t* know how much, and I’m not suggesting you should make up some fudge factor and re-run the numbers.
I think the numbers you ran are great, just to interpret them as a lower limit. If pitchers had pitched to avoid walking McGwire, he would have hit *at least* 82 HRs. Which is even cooler, right?
The lower limit just happens not to be as useful on the question of whether it’s better strategy for the pitcher to stay in the zone. Well, it’s a hard question to answer.
(BTW, I’d think it would be a reasonable first approximation to figure zero change to rates in translating away IBBs. Since there the strategy switch is just pitch normally.)
We really need to look at what actually happens AFTER the walk to get a gauge on how much the opposing team was helping or hurting themselves by walking him. I’ve actually done some preliminary checking of this, and (don’t have the exact numbers handy at the moment) while walks with less than two out tended to help the Giants, with two outs the batters coming up after Bonds were horrendous-something like 2 for 47, with virtually no RBIs, and obviously no possibility of further runs scoring.