“I don’t really like to run, and that’s why I didn’t go out for track in high school. I ain’t no fool, I see those dudes running around a track for a living. I wouldn’t want to run against them. I wouldn’t want to embarrass myself.”
—Willie Wilson
Since this column started back in March of 2006, I’ve spent a good deal of time on the topic of baserunning. Those efforts have resulted in the creation of five metrics under a common methodology that, when combined, can give us a fairly complete picture of the contributions that runners make on the bases. For readers new to BP–or those simply wishing to get an overview of the framework as a whole–you’ll be happy to know that an essay titled “The Tortoise, the Hare, and Juan Pierre: Translating Baserunning into Runs” describing the system, along with the totals for all major and minor league players from 2005 through 2007, will be included in Baseball Prospectus 2008 (so make sure and get yours pre-ordered today).
Besides preparing the essay for BP2K8, the topic has been on my mind recently because of the discussion surrounding Tim Raines‘ first year of eligibility for the Hall of Fame. Raines doesn’t have gaudy statistical Hall credentials such as 3,000 hits, 500 home runs, or 1,500 RBI, nor was he ever that undefinable but apparently now oh-so-important “most feared hitter in the game.” But there’s no doubt that just below Ricky Henderson, Raines was at the very top of the list in terms of leadoff hitters in the history of baseball. As a result, some of the focus on Raines rightly centered on his baserunning exploits. Dan Rosenheck of the New York Times took up the topic last week and, using the numbers from my baserunning metrics, made the point that in the running 80s the top runners–like Raines–had the talent to contribute over ten runs per year, while those today are three or four runs short of that. In fact, the league leaders from 1982 through 1989 (which includes talent plus variance due to luck) averaged +13.2 total runs on the bases (EqBRR), while from 2000 through 2007 the average has been +10.0 runs. That difference has to do with the greater number of stolen base opportunities that then generate high EqSBR values when accompanied by high success rates, as in the case of Raines. So how much was Raines worth on the bases?
In a post on my blog, I total it up for his career (minus 1999) and come up with a total of 102.5 career EqBRR for Raines. That actually ranks him third behind Willie Wilson at 109.6 (in the blog post I had Willie at 108.7 but was missing the data for 1977, when he contributed just under another run) and Rickey Henderson at 107.1. Of the three, Raines had by far the most value tied up in EqSBR (65 percent, with a value of +66.5 runs due to his 84.7 percent stolen base percentage, which, somewhat paradoxically, probably indicates that perhaps he didn’t run enough) while Henderson was more well-rounded and did better on advancing on hits (+31.2 in EqHAR) and other advancement (+14.6 in EqOAR). Wilson was somewhere in the middle, and did well both in EqSBR at +56.9 and EqHAR at +26.5. As we’ll see shortly, those three remain head and shoulders above the rest of the pack, with the next highest totals below 80 runs.
Raines received just 24.3 percent of the vote, a total the writers should feel embarrassed about, especially when you consider that Jim Rice received 72.2 percent on his way to an all but certain election next year. But despite the disappointment, the queries from Rosenheck prompted me to do something I had put on the back burner for many months: running the baserunning framework using all of the Retrosheet data available going back to 1956. Today, we’ll examine those results, and in the future, yes, plans are in the works to publish these metrics on the site in the stats section in the future.
Crowning a Royal
Let’s start where we left off above: the career leaders in total runs contributed on the bases. The following table shows the top 25 runners from 1956 through 2007 in terms of EqBRR:
Name Years Opps EqGAR Opps EqSBR Opps EqAAR Opps EqHAR Opps EqOAR EqBRR 550 Willie Wilson 76-94 526 6.4 822 56.6 674 14.0 580 26.5 4681 6.0 109.6 8.3 R. Henderson 79-03 930 11.7 1773 42.9 1052 6.7 1005 31.2 7260 14.6 107.1 4.9 Tim Raines 79-02 658 7.6 969 66.5 802 7.7 733 16.8 6121 3.8 102.5 6.1 Davey Lopes 72-87 489 6.4 710 38.4 543 1.8 543 23.4 3929 6.2 76.1 6.7 Paul Molitor 78-98 745 12.0 641 32.0 978 8.0 1019 24.9 7051 -1.1 75.8 4.0 Vince Coleman 85-97 456 7.7 1011 39.1 409 4.4 415 10.7 3303 8.1 70.0 6.9 Luis Aparicio 56-73 759 11.3 693 13.0 693 5.2 794 22.1 6095 12.1 63.8 3.9 Willie Davis 60-79 438 6.4 556 7.1 566 11.3 609 32.7 4545 4.6 62.0 5.1 Ozzie Smith 78-96 724 8.9 757 19.7 704 6.0 830 19.4 6031 5.8 59.9 3.6 Barry Larkin 86-04 427 0.6 422 17.1 525 7.0 594 23.0 4690 8.7 56.5 4.7 Kenny Lofton 91-07 603 12.1 782 11.5 678 -0.8 755 28.6 4930 3.5 54.9 3.9 Mookie Wilson 80-91 372 6.4 449 10.7 359 8.7 376 21.5 2853 7.2 54.4 6.8 Robin Yount 74-93 618 5.3 380 -6.8 862 16.7 921 39.4 6527 -0.5 54.1 3.2 Joe Morgan 63-84 629 1.5 903 34.2 817 1.0 868 -0.6 6375 16.5 52.6 3.0 Johnny Damon 95-07 477 1.9 391 15.1 532 5.8 610 22.6 4316 7.0 52.3 4.5 Lenny Dykstra 85-96 424 6.8 374 19.7 412 5.9 443 9.4 3283 4.9 46.6 5.2 Ron LeFlore 74-82 414 10.2 627 18.5 370 5.2 348 11.8 2793 0.1 45.7 5.5 Otis Nixon 83-98 390 3.9 795 12.6 431 5.7 505 18.4 3555 4.7 45.3 4.4 M. Grissom 89-05 404 2.6 538 16.0 450 8.7 509 13.4 3896 2.8 43.5 4.1 Eric Davis 84-01 227 -1.3 415 24.5 326 2.2 348 11.6 2618 6.2 43.3 6.0 Juan Pierre 00-07 383 9.8 552 -1.0 339 6.0 461 21.4 3203 6.5 42.8 4.8 Craig Biggio 88-07 774 6.3 517 9.0 823 8.6 963 7.3 6822 9.1 40.2 2.2 Willie Mays 56-73 398 2.9 398 2.0 561 8.5 648 19.4 4865 7.0 39.9 3.2 Cesar Cedeno 70-86 367 5.5 759 9.5 454 2.0 499 10.9 3946 10.8 38.8 3.5 Gary Redus 82-94 279 3.1 425 16.3 278 6.1 244 9.8 2120 3.1 38.4 6.3
Davey Lopes (+76.1) and Paul Molitor (+75.8) fall well behind the ruling triumvirate, and are joined by Vince Coleman (+70.0) as the only other three players to reach 70 or more runs. As you look at those players–and the rest of the top 25–you’ll notice that there are a variety of routes these players took to crack this threshold. Like Raines, Lopes was a terrific percentage basestealer, but didn’t fare as well on advancing on groundouts or fly balls. Robin Yount reached 13th (+54.1) on the strength of EqAAR (+16.7) and EqHAR (+39.4), and actually amassed negative values in EqSBR and EqOAR, where pure speed plays a larger role. Joe Morgan excelled in EqSBR and EqOAR but was decidedly average in the other three categories, while Ron LeFlore seemed to be very well-rounded. You’ll also notice that three active players–Kenny Lofton, Johnny Damon, and Juan Pierre–make the list, but almost half the leaders consist of players whose peak seasons occurred in the 1980s.
To the right of the EqBRR column you’ll notice a column titled 550, which is a prorated total per 550 opportunities. This is used as a quick and dirty method to find out who took the most advantage of their opportunities; we’ll take a more sophisticated route below. Willie Wilson (+8.3) comes out well above the rest, with Vince Coleman (+6.9) coming in second, surprisingly followed by–surprising to me anyway–Mookie Wilson (+6.8), Davey Lopes (+6.7), and Gary Redus (+6.3). Players not shown in the table above who did well in the rate statistic and who played 10 or more seasons include Tom Goodwin (+5.6) and Gary Pettis (+5.5). So does this mean that Willie Wilson was the best on the bases in the past 50 years? No metrics are perfect, but it at least provides very strong support for those who think so.
Some may be wondering where this puts Lou Brock, since he did not crack the top 25. Let’s take a look at his seasonal totals:
Year Opps EqGAR Opps EqSBR Opps EqAAR Opps EqHAR Opps EqOAR Opps EqBRR 1961 0 0.0 0 0.0 0 0.0 0 0.0 7 0.1 7 0.1 1962 40 1.4 25 -0.4 27 0.1 24 2.7 226 0.1 342 4.0 1963 42 -0.9 40 -1.3 28 -0.1 43 -0.2 299 -0.2 452 -2.8 1964 43 -0.3 64 0.1 29 0.0 53 0.7 309 -0.5 498 -0.1 1965 55 2.7 99 -2.8 34 0.7 35 -0.9 304 -1.5 527 -1.8 1966 39 0.4 96 4.2 47 1.0 46 2.5 325 -0.4 553 7.7 1967 47 -0.3 78 -0.7 32 0.0 63 0.3 371 1.5 591 0.7 1968 62 -0.3 76 4.3 36 -0.3 58 -1.4 420 -0.3 652 1.9 1969 66 1.0 69 3.8 38 0.3 50 -0.1 395 -1.5 618 3.6 1970 51 0.5 66 1.2 42 0.8 64 3.5 399 0.3 622 6.3 1971 66 -0.4 84 3.0 67 -2.2 80 -0.5 504 0.3 801 0.3 1972 52 -0.5 88 1.6 51 0.9 51 -0.8 383 0.2 625 1.4 1973 62 2.5 98 1.9 45 1.3 40 2.1 413 1.7 658 9.5 1974 54 0.6 153 5.4 56 -4.1 61 -0.3 438 0.6 762 2.2 1975 47 0.7 75 2.5 48 -0.3 40 0.4 327 -0.1 537 3.3 1976 28 -0.2 78 -1.1 24 -0.4 43 -0.5 276 -0.2 449 -2.4 1977 30 -0.9 59 -3.4 26 0.4 51 1.2 245 0.0 411 -2.6 1978 25 0.4 23 0.9 22 0.2 22 1.0 175 0.1 267 2.6 1979 21 -0.5 35 -1.8 31 0.0 41 0.7 246 0.1 374 -1.5 830 5.8 1306 17.4 683 -1.5 865 10.3 6062 0.4 9746 32.4
His career total of +32.4 runs puts him just outside of the top 25, sandwiching him between Lonnie Smith (+33.0) and Delino DeShields (+32.2) in 37th place. He’s hurt by three of his four final seasons and some very average to below-average running from 1963 through 1965. Despite leading the league in 1966 and 1973, he would periodically rate very poorly in EqAAR or EqHAR, thereby holding his totals down. His runs per 550 opportunities sits at +1.8 ranking him 94th, just ahead of Hank Aaron (+1.8 and +25.2 EqBRR).
On the flip side of the coin, the following table shows the 25 runners who find themselves on the bottom of pile:
Name Years Opps EqGAR Opps EqSBR Opps EqAAR Opps EqHAR Opps EqOAR EqBRR 550 Frank Thomas 90-07 355 -8.2 50 -6.1 631 -0.4 722 -18.0 4956 -4.6 -37.1 -3.0 Jorge Posada 96-07 243 -4.5 37 -9.2 327 -3.3 371 -16.6 2661 -3.6 -37.1 -5.6 Gary Carter 74-92 379 -7.0 90 -16.8 510 -4.8 535 -7.6 4095 -1.0 -37.4 -3.7 Dave Parker 73-91 380 3.5 277 -30.6 606 -6.8 671 -8.1 4709 4.4 -37.6 -3.1 Jim Thome 91-07 319 -5.7 39 -7.9 474 -8.8 639 -16.9 4057 1.7 -37.6 -3.7 Tim Wallach 80-96 428 -0.1 122 -29.5 500 1.5 569 -8.5 4138 -1.4 -38.1 -3.6 C. Delgado 93-07 281 -3.0 20 -6.2 453 -5.8 519 -24.0 3693 0.2 -38.7 -4.3 Boog Powell 61-77 319 -4.8 44 -10.0 462 3.0 517 -22.7 3718 -5.4 -39.9 -4.3 Fred McGriff 86-04 383 -6.7 104 -9.7 555 -1.2 623 -19.1 4368 -3.1 -39.9 -3.6 Alvin Davis 84-92 208 -2.7 24 -8.4 355 -3.2 354 -21.2 2620 -4.8 -40.3 -6.2 H.Baines 80-01 361 -4.7 68 -12.2 673 -7.3 694 -12.8 5023 -3.8 -40.9 -3.3 Rusty Staub 63-85 519 -5.9 92 -8.3 686 -3.4 773 -26.1 5668 2.7 -41.1 -2.9 Rick Cerone 75-92 245 -2.5 33 -11.7 258 -2.7 336 -21.6 2123 -3.0 -41.4 -7.6 G. Luzinski 70-84 277 -4.0 69 -8.9 461 -4.9 533 -26.2 3537 1.5 -42.5 -4.8 W. McCovey 59-80 358 -8.8 59 -7.7 506 -0.6 633 -25.6 4365 -0.1 -42.8 -4.0 Bob Boone 72-90 429 0.3 109 -25.0 406 3.2 486 -18.1 3620 -3.8 -43.4 -4.7 Eddie Murray 77-97 539 -8.5 171 -10.7 845 -8.1 821 -19.4 6245 2.8 -43.9 -2.8 Tony Pena 80-97 374 0.8 150 -20.0 419 -3.6 455 -17.9 3415 -3.2 -43.9 -5.0 E. Martinez 87-04 394 -7.3 71 -9.5 553 -1.0 634 -23.7 4488 -2.6 -44.0 -3.9 Ln. Parrish 77-95 311 -5.0 69 -15.2 453 -9.8 471 -10.2 3333 -4.8 -44.9 -5.3 Mike Piazza 92-07 224 -4.3 31 -8.4 444 -6.8 476 -26.3 3394 -1.6 -47.4 -5.7 H. Killebrew 56-75 345 -9.1 42 -7.6 534 -0.3 612 -24.7 4291 -5.8 -47.6 -4.5 Wade Boggs 82-98 708 -7.7 65 -17.6 901 -2.2 1057 -12.2 7027 -8.5 -48.1 -2.7 Todd Zeile 89-04 331 -6.2 111 -20.7 478 -2.4 596 -18.6 3712 -0.4 -48.4 -5.1 Ted Simmons 68-88 449 -8.0 62 -18.0 603 -1.6 677 -25.1 4757 2.0 -50.6 -4.3
In an earlier blog post, I had noted that Todd Zeile (-48.4) and Wade Boggs (-48.1) were at the bottom; in calculating the additional Retrosheet years, Ted Simmons just barely gets under them at -50.6 runs. It should probably come as no surprise that nine of the twenty-five were primarily catchers, with the rest first basemen, corner outfielders, and a few third basemen and DHs. It’s also apparent that none of them–save perhaps Dave Parker–was ever considered above average in terms of running speed.
Parker’s case is interesting because he did finish his career above average in EqGAR and EqOAR, yet was almost 40 runs below average overall. His seasonal totals reveal that his poor stolen bases percentages dragged him down, -30.6 over the course of his career (154 stolen bases and 113 caught stealing); a representative season might be his 17-for-36 performance in 1977. On those very rare occasions (exactly twice in his career) when he took more caution–such as 1979, when he was 20 of 24 and contributed +1.1 runs in EqSBR–he ended up at +7.5 overall. Looking at Parker’s career:
Year Opps EqGAR Opps EqSBR Opps EqAAR Opps EqHAR Opps EqOAR Opps EqBRR 1973 9 0.0 2 0.0 10 -0.7 8 0.6 68 -0.2 97 -0.2 1974 14 -0.1 6 -1.4 12 0.3 17 0.6 115 1.2 164 0.6 1975 35 1.3 14 -2.1 42 -2.9 45 -0.8 307 -0.4 443 -4.9 1976 22 -0.2 28 -0.9 39 1.4 39 2.0 260 -1.3 388 1.0 1977 32 -0.4 37 -6.4 44 -0.9 58 0.7 360 -0.8 531 -7.8 1978 15 0.1 28 -0.8 45 -1.6 40 0.4 335 1.5 463 -0.4 1979 30 2.2 24 1.1 46 1.4 46 1.7 416 1.2 562 7.5 1980 22 0.0 17 -1.4 33 -1.9 39 0.6 253 0.1 364 -2.6 1981 12 0.3 8 0.2 12 0.1 15 1.8 97 0.6 144 3.0 1982 9 -0.9 12 -1.2 21 0.6 24 -0.6 130 0.3 196 -1.8 1983 25 0.9 23 -2.9 31 0.3 39 -0.7 256 -0.6 374 -3.0 1984 15 0.5 21 -2.6 37 0.2 41 -1.5 291 0.8 405 -2.5 1985 24 0.3 17 -4.1 44 -1.0 46 -1.3 296 -0.9 427 -7.0 1986 23 0.3 8 -2.9 27 -0.6 38 -1.4 295 1.1 391 -3.6 1987 24 0.3 11 -0.9 39 -0.2 37 -4.1 265 -0.1 376 -5.0 1988 14 -0.6 1 -0.6 23 -0.8 26 -0.9 189 0.5 253 -2.4 1989 18 -0.5 0 0.0 32 0.2 30 -2.0 252 0.6 332 -1.8 1990 24 0.6 13 -2.9 42 0.1 46 -2.4 302 0.0 427 -4.6 1991 13 -0.4 7 -0.9 27 -0.6 37 -0.9 222 0.7 306 -2.1 380 3.5 277 -30.6 606 -6.8 671 -8.1 4709 4.4 6643 -37.6
Parker’s extremes, however, are not the largest historically. There were twelve other players with larger differences in separate seasons, including Kenny Lofton’s +13.3 in 1993 coupled with a -8.5 in 1997, and Bobby Bonds’ -7.2 in 1979 and a +13.6 in 1972. But the largest difference in two seasons belongs to Maury Wills, who scored an amazing +20.6 in his record-breaking 1962 thanks to a +13.8 in EqSBR, and a -7.5 in 1966 when he was caught stealing 24 times in 62 attempts. Both totals were the extremes in the league, and Wills’ 1962 performance was the top single-season value of all time.
From a rate perspective, you probably won’t be surprised to learn that pitchers dominate the bottom of the list. John Smiley (-17.4) and Jose DeLeon (-17.2) take the bottom two spots; further, fully 65 of the bottom 100 rates for players playing 10 or more seasons belong to pitchers. We already knew pitchers don’t contribute much on the bases, so I’ve filtered them out of the following table that shows the bottom ten runners ordered by rate:
Name Years Opps EqGAR Opps EqSBR Opps EqAAR Opps EqHAR Opps EqOAR EqBRR 550 John Bateman 63-72 152 -2.0 19 -4.6 140 -3.1 163 -13.1 1299 -0.7 -23.5 -7.3 Pat Borders 88-05 165 -0.2 29 -7.4 187 -7.1 199 -10.1 1486 -2.6 -27.4 -7.3 Rick Cerone 75-92 245 -2.5 33 -11.7 258 -2.7 336 -21.6 2123 -3.0 -41.4 -7.6 Ron Hodges 73-84 100 -0.8 26 -6.2 117 -1.1 103 -7.0 826 -1.3 -16.5 -7.7 Todd Greene 96-06 52 0.6 4 -0.7 52 -3.5 59 -4.5 455 -0.8 -8.8 -7.8 Buck Martinez 69-86 123 -2.9 19 -6.1 142 -0.5 180 -13.3 1218 -1.3 -24.1 -7.9 M.LaValliere 84-95 188 -1.4 23 -6.5 140 -3.3 170 -15.2 1437 -2.4 -28.8 -8.1 Jamie Quirk 75-92 115 -1.4 22 -8.5 116 -1.8 133 -7.3 994 -1.3 -20.4 -8.1 Randy Knorr 91-01 28 -0.3 1 -0.7 37 -1.1 44 -4.5 277 -0.3 -6.9 -9.9 Eddie Perez 95-05 68 -1.5 4 -1.7 50 -0.5 77 -8.8 536 -0.8 -13.4 -10.0
Here we see illustrated in stark relief the idea behind the old saying “runs well for a catcher,” as all ten of the worst baserunners measured by the rate stat wore the tools of ignorance. The first non-catcher on the list was reserve outfielder and pinch-hitter Jerry Lynch at 17th (-6.7), who played for the Reds and Pirates in the fifties and sixties and who played a critical role in the 1961 NL pennant race.
Before moving on to some analysis, let’s take a look at one more set of leader boards by showing the career leaders and trailers in each of the five metrics:
Advancing on Ground Outs Stolen Bases and Pickoffs Name Yrs EqGAR Name Yrs EqSBR Brett Butler 17 13.7 Tim Raines Sr 22 66.5 Steve Sax 14 12.9 Willie Wilson 19 56.6 Kenny Lofton 16 12.1 Rickey Henderson 24 42.9 Paul Molitor 21 12.0 Vince Coleman 13 39.1 Rickey Henderson 24 11.7 Davey Lopes 16 38.4 ------------------------------- ---------------------------- Carl Yastrzemski 23 -8.6 Bob Bailey 17 -31.2 Willie McCovey 22 -8.8 Brett Butler 17 -33.0 Pete Rose 24 -9.0 Pete Rose 24 -33.3 Harmon Killebrew 20 -9.1 Buddy Bell 18 -34.0 Joe Torre 17 -12.5 Alfredo Griffin 18 -34.4 Advancing on Air Outs Advancing on Hits Name Yrs EqAAR Name Yrs EqHAR Robin Yount 20 16.7 Robin Yount 20 39.4 Willie Wilson 19 14.0 Willie Davis 18 32.7 Pete Rose 24 11.9 Rickey Henderson 24 31.2 George Brett 21 11.3 Glenn Beckert 11 29.4 Willie Davis 18 11.3 Kenny Lofton 16 28.6 ------------------------------- ---------------------------- Dave Kingman 16 -8.5 Ted Simmons 21 -25.1 Glenn Hubbard 12 -8.6 Willie McCovey 22 -25.6 Jim Thome 16 -8.8 Rusty Staub 23 -26.1 Lance Parrish 19 -9.8 Greg Luzinski 15 -26.2 Tony Perez 23 -15.2 Mike Piazza 15 -26.3 Advancing on Balks, PB, and WP Name Yrs EqOAR Joe Morgan 22 16.5 Rickey Henderson 24 14.6 Luis Aparicio 18 12.1 Cesar Cedeno 17 10.8 Bert Campaneris 19 10.8 ------------------------------- Clete Boyer 15 -5.8 Harmon Killebrew 20 -5.8 Chris Chambliss 16 -6.3 Sammy Sosa 17 -6.6 Wade Boggs 17 -8.5
Tony Perez far outdistances the competition in giving up runs on fly outs at -15.2, but when we look closer we find that while he did record negative values in 15 of his 23 seasons, it was 1980 that really did him in, when he was thrown out at the plate five times in 33 opportunities (costing him -5.5 runs). Mercifully, the 1981 season was shortened, as Perez was on a pace to match 1980 by being nailed three times in 21 chances. We also see that Big Red Machine teammate Pete Rose did very well in advancing on fly outs, while seemingly unable to do so on ground outs, and being simply awful in attempting to steal bases. Also, as noted before, we see here why Robin Yount did so well in the career rankings, as he took the top spot in both EqAAR and EqHAR.
Two surprising players are Glenn Beckert (with his excellent showing in EqHAR) and George Brett (with his ability to advance on fly outs). Beckert managed to record positive values on advancing on hits in ten of his eleven seasons, and +3.5 runs or more five times, recording the second highest single season total at +5.7 in 1970. Not only was he able to take the extra base, he enhanced his numbers by not getting thrown out: he was caught trying to advance just five times in 491 career chances. Brett’s case was much the same: he was thrown out advancing on fly balls just eight times in 807 career chances (and not at all from 1981-1987, but three times in 1989) while advancing 216 times.
Pure Baserunning
Thus far we’ve looked at career totals in the various metrics and created a simple rate statistic that takes all five metrics into account, scaled to roughly approximate the number of opportunities a runner would get in a single season. But we can do better. The “per 550” rate statistic has two major weaknesses: it treats all opportunities equally, and it’s not really a measure of “pure” baserunning since it includes EqSBR, which is almost entirely discretionary.
The first of these weaknesses can be thought of as two distinct issues. First, the simple rate statistic makes the assumption that an opportunity to advance in one metric is the same as that in another. We know, however, that that isn’t the case. In the extreme example an opportunity–as counted by EqOAR–has much less value than one credited for EqHAR, as evidenced both by the far greater number of opportunities counted for EqOAR and the much smaller range in terms of runs it represents. As a result, it’s not really fair to weight them equally when players accumulate disproportionate numbers of opportunities in one or more metrics. Second, Per 550 assumes that all opportunities within a metric are created equal. Those who have read my previous columns on this topic will know that context plays a large role in assigning expected and actual run values to the underlying events in the various metrics, so in a sense the framework is built around the idea that opportunities vary widely in their quality. For example, a runner who finds himself on second base with one out disproportionately will have his ability to rack up large EqGAR numbers reduced, since taking third in that situation nets his team relatively fewer runs than advancing with nobody out. While runners are credited fairly within the metric itself, since each opportunity is judged in context, taking a simple total divided by opportunities approach won’t always be fair.
As far as the second weakness goes, it turns out that when many folks talk about baserunning, they’re not really thinking about discretionary stolen base attempts, but instead that combination of speed, risk taking, and judgment that goes into evaluating situations, and that when aggregated leads us to assert that a particular player is a good baserunner. Since EqSBR is primarily comprised of stolen base attempts (with a few pickoffs thrown in) our new rate statistic, which we’ll christen Equivalent Baserunning Rate (EqBRRate), will omit this aspect.
So to eliminate both of these weaknesses, we’ll define EqBRRate as the ratio of actual or total runs to expected runs contributed across the four remaining metrics. Since both values for individual opportunities consider the context and are weighted appropriately (an EqOAR opportunity has both a lower expected and usually actual run value associated with it than an opportunity of EqHAR) and since we’re eliminating EqSBR, both weaknesses discussed above addressed. To illustrate how this works, let’s consider Chone Figgins‘ 2007 season:
Metric Opps TotRuns ExpRuns EqOAR 314 0.1 1.2 EqGAR 31 3.2 2.6 EqAAR 44 5.7 4.7 EqHAR 53 10.1 5.6 442 19.0 14.2
Taking the context of the opportunities into account, we would have expected Figgins to net +14.2 runs across the four metrics, but he actually contributed +19.0 runs. When we divide the total by the expected we get a ratio of 1.34, indicating that he contributed, or manufactured if you will, 34 percent more runs than expected.
Of course, when creating a rate statistic, one needs to form a cutoff, since otherwise players with small samples will dominate the leaders and trailers. Because the magnitude of the opportunities for EqOAR is much greater than the other metrics, I’ve divided EqOAR opportunities by ten and then summed all the opportunities to create a Scaled Opportunities value that can be used in single-season and career comparisons. So for Figgins, we calculate this value to be 159. For the career leaders we’ll use a fairly arbitrary cutoff of 550 opportunities, and for single-season leaders we’ll use 75.
Armed with this metric, we can now reveal the career leaders and trailers in EqBRRate, our measure of pure baserunning ability stretching from 1956 through 2007:
Name Years Span EqBRR ScaledOpps EqBRRate Chone Figgins 6 2002-2007 31.0 756 1.35 Tom Goodwin 12 1991-2004 34.1 1017 1.33 Juan Pierre 8 2000-2007 42.8 1499 1.31 Grady Sizemore 4 2004-2007 18.4 633 1.30 Cristian Guzman 7 2000-2007 11.8 876 1.29 Mookie Wilson 12 1980-1991 54.4 1386 1.28 Willie Davis 18 1960-1979 62.0 2058 1.27 Ray Durham 12 1995-2007 35.6 1819 1.26 Tony Scott 11 1973-1984 6.9 758 1.26 Jose Reyes 5 2003-2007 25.6 685 1.26 Ken Landreaux 11 1977-1987 18.7 977 1.25 Gerald Young 8 1987-1994 11.9 559 1.24 Glenn Beckert 11 1965-1975 26.4 1517 1.24 Lonnie Smith 17 1978-1994 33.0 1586 1.23 Dave Hollins 11 1990-2002 12.3 737 1.23 Scott Podsednik 7 2001-2007 21.3 684 1.23 Oddibe McDowell 7 1985-1994 23.2 773 1.23 David DeJesus 5 2003-2007 3.4 648 1.22 Tony Graffanino 11 1996-2007 10.4 739 1.22 Gary Pettis 11 1982-1992 38.2 1140 1.22 ---------------------------------------------------------------- Clay Dalrymple 12 1960-1971 -26.7 654 0.71 Javier Lopez 14 1992-2006 -36.1 1127 0.71 Pat Borders 16 1988-2005 -27.4 687 0.71 Alvin Davis 9 1984-1992 -40.3 1176 0.69 Dave Valle 13 1984-1996 -20.5 681 0.69 Steve Balboni 11 1981-1993 -16.1 587 0.68 Ed Bailey 10 1956-1965 -32.9 834 0.67 Mike Piazza 15 1992-2007 -47.4 1471 0.67 Buck Martinez 17 1969-1986 -24.1 560 0.67 Jody Davis 10 1981-1990 -26.9 705 0.66 Cecil Fielder 13 1985-1998 -34.1 1010 0.66 Rich Gedman 13 1980-1992 -25.2 702 0.65 Fred Kendall 12 1969-1980 -20.6 572 0.65 John Bateman 10 1963-1972 -23.5 580 0.65 Darrin Fletcher 13 1989-2002 -24.4 722 0.65 Smoky Burgess 11 1956-1966 -22.9 565 0.59 Gus Triandos 10 1956-1965 -22.6 662 0.58 Willie Aikens 8 1977-1985 -26.1 584 0.58 Mike LaValliere 12 1984-1995 -28.8 634 0.57 Bengie Molina 8 2000-2007 -26.6 678 0.57
Chone Figgins takes the title of the best pure baserunner of the last 50 years, while Bengie Molina… well, let’s just say that coaches don’t have to concern themselves with waving him around very often.
What’s interesting about the top part of this table is that it doesn’t contain the career leaders in stolen bases–or for that matter EqBRR–but instead is populated by primarily fast runners with presumably good judgment. I’ve included the EqBRR career totals as well, which make it obvious that Tony Scott and David DeJesus aren’t much for stealing bases, but certainly could or can run the bases. This table also reveals that elite runners like Figgins are worth approximately 25-30 percent more runs than an average runner, while plodders might cost their team 30-40 percent. You’ll also notice that nine of the top twenty runners were active in 2007. While this may simply be random, it just may reflect a widening of the gap between the best and worst runners of recent years that creates more space for better runners to excel. It should also be noted that the bottom of the list doesn’t consist exclusively of catchers this time and instead throws in a few players like Willie “Nothing like Mays” Aikens, Cecil Fielder, Steve Balboni, and Alvin Davis.
To finish up today we’ll explore two questions that come up frequently in email from readers: aging and correlation. Once we have a rate statistic we can easily analyze how that rate varies with age. To do this we can simply weight EqBRRate by Scaled Opportunities for all seasons by age (leaving out pitchers) to come up with the following graph:
The graph shows that starting at an above average rate, players improve in their running from age 20, and peak at 23. We can speculate that this increase in performance in the first three seasons–despite already likely having achieved maximum running speed–could be the result of a learning curve that players undergo as they adjust to the league. After declining fairly slowly through age 27, the descent picks up speed through the late 20s and through their mid 30s before slowing around age 34, no doubt due to the selection bias inherent in a graph like this, as players are selected out of the league for many reasons, perhaps including their baserunning.
We can also break the curves down by position with results that shouldn’t be too shocking:
Center fielders (using data from 1996-2006) are clearly the best baserunners overall (as evidenced by our top 20, more than half of whom played center), remaining 10-20 percent above average from ages 22 through 35. Second baseman and shortstops also remain above average throughout, as do all outfielders from 1956 through 1995. Next, we find left fielders (again, since 1996) just slightly ahead of right fielders for most of the span, with third basemen trailing somewhat. First basemen and designated hitters, often interchangeable, track fairly well with each other. And, of course, on the bottom of the list we find our catchers, who start out at about 90 percent of average, and descend from there.
On the question of correlation, what readers generally want to know is how persistent baserunning is as a skill from year-to-year. In order to measure this, we can split the careers of non-pitchers into even and odd seasons, and using our 75 Scaled Opportunities threshold we can pair those career halves for players with at least three seasons in each half. This resulted in data for 696 players. Performing a correlation on those pairs results in the plot with the linear regression line we see below.
The correlation coefficient here is a fairly healthy 0.64, resulting in an R-square (coefficient of determination) of 0.42. This suggests that 42 percent of a player’s EqBRRate can be attributed to skill and 58 percent to other factors, including causes such as luck and coaching. I did check whether aging–for example, that fact that when splitting a career consisting of consecutive seasons into even and odd halves, one of the halves will necessarily include, on average, seasons when the player was younger–would have relevance here by first normalizing the EqBRRate using the aging curve above; overall, it had no effect on the strength of the correlation.
To put this result in perspective, we can note that for the same time period the career half correlation of determination for batting average was 0.56 (using a 200 AB seasonal cutoff), indicating that the skill component involved in baserunning (using these metrics anyway) is somewhat smaller than that contained in batting average.
Thank you for reading
This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.
Subscribe now