Bret Boone is in the middle of an astonishing season -- as of this writing, ESPN projects him to top his previous season best home run total by 50 percent, his top RBI total by 47 percent, and his hit total by 32 percent. He's on pace to set career highs in batting average (just ahead of his 1994 breakthrough season), and slugging average, topping .500 for the first time, and 63 points ahead of his next best season.
His track record since 1995 is .250/.310/.404 AVG/OBP/SLG over 3000+ at bats. This year, he's blowing that away with production of .322/.360/.554, and certainly ranks as one of the biggest surprises of the year. Coming off so many seasons of mediocrity, one naturally wonders whether this is a fluke season. And the next thought might be, if this is a fluke, how does Boone's season rank among the all-time fluke seasons?
Boone isn't the only all-time fluke candidate playing in 2001, either. Luis Gonzalez is also in the middle of a monster season, although he's been among the top offensive players the past two season, he's 200 points of OPS above his career high, and hitting home runs by the bucketful. His unusually late development and recent track record of high performance makes his improvement this season look less like a fluke than Boone (who's nearest approach to his current level came in 1994), but with the incredible numbers Gonzalez is putting up, we'll look at him as well.
The word "fluke" may have some negative connotation, but here we're just using it to mean a peak season unusually far above the rest of a player's career, or a level of performance that could not be maintained or equaled. I used Sean Lahman's database (available as a free download from www.baseball1.com) to search for fluke seasons throughout history, to see how Boone's 2001 would fit in. I used the following criteria in determining a fluke season:
1) The player must be at least 32 years old through 2000 -- to eliminate players still in the midst of their careers, but allowing some active players clearly past their peak to be considered.
2) The player needs at least five seasons of 400+ plate appearances to qualify, to eliminate the "flash in a pan" player who flamed out quickly and never established a baseline of expected play outside their peak season. Joe Charboneau, for example, won't appear on the list.
3) Only fluke seasons since 1893 (when the pitching distance was set to 60' 6") will be considered.
4) I did not consider any fluke longer than one season. Thus, if a player had two seasons way above the rest of his 20-year career, he won't be considered fluky because he at least did it twice.
Next, I had to decide how to value each player's season. I could have used a single stat like home runs (in which Brady Anderson's 50 HR in 1996 would be the biggest fluke of all time, edging out Davey Johnson's 43 HR in 1973), but that fails to capture what happened in the rest of a player's at-bats. I could have used batting average, or some other rate measure such as OBP. However, since it's possible to put up impressive rate numbers in a limited number of at-bats (Mariano Duncan's .340 batting average over 400 AB in 1996 comes to mind), I decided to consider an aggregate measure of value, rather than just a rate.
There were other considerations. A player who spends one year in Colorado and puts up outstanding numbers (such as Jeffery Hammonds' .335/.395/.529 last year) shouldn't get automatically pegged as having a fluke season, since the park environment is so conducive to hitting. Thus, we will want to account for changes in park and league. We're looking for seasons where, for no apparent reason, everything "clicked" for a player.
A sabermetric approach seemed to be called for. Marginal Lineup Value (MLV) measures what the expected change in run scoring is if you were to insert the player into an otherwise average lineup. MLV is based on the Runs Created metric invented by Bill James, and is adjusted for park and league. MLV rewards both high rates of production and playing time. MLV forms the basis for VORP, a system for total value I have been publishing online for several years. MLV and VORP are described in more detail at stathead.com.
To help give some context for the scale: An MLV of zero means the batter was exactly league average. Carlos Delgado led the AL last year with an MLV of 86.7, and Barry Bonds led the NL with an MLV of 86.3 (a figure he's already topped this year -- he's at 96.5 on pace for an MLV of about 129, which no player other than Babe Ruth has topped). Only 10-15 players in each league typically top a 50 MLV, and anything above 20 can be considered a solid offensive season, well above league average. An MLV of 100 has only been reached 22 times in history (with Barry and Luis on the verge of nos. 23 and 24). Negative MLV is not only possible, but common, since any below average hitter will have a negative MLV.
Bret Boone is on pace for an MLV of 50.9 (i.e., an average team would score about 51 runs more with Boone in the lineup than they would if they had a league average hitter). His second-best season was an MLV of 20.6 in 1994. The difference between his best and second-best season is 30.3 runs. If Boone maintains that pace, he'll join only 20 other players to have had a fluke season 30 or more runs above their next-best season.
Fluke Season 2nd-best Season
NAME YEAR MLV PA YEAR MLV PA GAP
--------------- ----- ------ ---- ----- ----- ---- -----
CASH,NORM 1961 94.7 668 1965 34.1 548 60.5
GONZALEZ,LUIS 2001 104.7 ??? 1999 51.5 693 53.2
DUFFY,HUGH 1894 88.8 606 1891 39.9 601 48.9
GASTON,CITO 1970 49.5 627 1972 4.0 403 45.5
HOLMES,TOMMY 1945 74.8 710 1944 31.6 694 43.2
GENTILE,JIM 1961 71.1 593 1960 30.9 459 40.2
REYNOLDS,CARL 1930 55.9 590 1929 16.9 541 39.0
PETROCELLI,RICO 1969 56.0 634 1971 20.6 646 35.4
HARPER,TOMMY 1970 46.5 685 1972 11.5 632 35.0
SEYMOUR,CY 1905 66.4 634 1903 31.4 594 35.0
ZIMMERMAN,HEINIE 1912 68.3 601 1913 34.2 494 34.1
TRAVIS,CECIL 1941 61.3 661 1938 28.1 629 33.2
HICKMAN,JIM 1970 46.4 608 1972 13.4 422 33.1
BELL,LES 1926 38.0 635 1929 5.3 533 32.7
CAMINITI,KEN 1996 71.3 628 1995 39.0 596 32.3
STONE,GEORGE 1906 71.5 640 1907 39.3 661 32.2
CAREW,ROD 1977 81.4 688 1975 49.3 600 32.1
SANDERS,REGGIE 1995 48.2 561 1992 16.9 437 31.4
BOONE,BRET 2001 50.9 ??? 1994 20.6 419 30.3
GILKEY,BERNARD 1996 61.8 648 1993 31.5 617 30.3
LAZZERI,TONY 1929 66.9 617 1932 36.6 594 30.3
VAUGHAN,ARKY 1935 79.7 603 1934 49.4 654 30.3
WILLIAMS,JIMMY 1899 62.9 683 1901 33.0 559 29.9
VOSMIK,JOE 1935 51.5 683 1934 21.8 440 29.7
SAX,STEVE 1986 46.9 695 1989 17.7 704 29.2
ANDERSON,BRADY 1996 57.6 677 1997 28.6 693 29.0
ROSEN,AL 1953 81.3 688 1952 53.2 646 28.1
WILSON,HACK 1930 93.9 691 1929 65.9 654 28.0
Few will be surprised at Norm Cash's 1961 topping the list. A lifetime .271 hitter, he had the best overall offensive season in the league 1961, with a .361/.487/.662 AVG/OBP/SLG, including 41 HR, 132 RBI, and 119 runs scored. He never came within 75 points of that batting average again, nor closer than 85 points of OBP and 150 points of SLG. He never topped 100 RBI or 100 runs scored in any other season. His 193 hits were 25 better than his next best season. Cash has been the poster child for fluke seasons for some time.
As an aside, one player on the list but warrants special mention. That player is Hugh Duffy, who's .440/.502/.694 season in 1894 not only won the Triple Crown, but was worth 88.8 MLV, nearly 50 runs better than his next best season. Oddly enough, a .500+ OBP wasn't enough to lead the league, as he finished second in OBP to Billy Hamilton's .523. However, Duffy's next-best season came in 1891, prior to the pitching distance changing. If we use 1891 as his second-best season, Duffy would have the second-biggest fluke on the list, but closer to No. 3 than No. 1. If we use his next best season at 60' 6", which is 1893, Duffy would nearly equal Norm Cash's all-time fluke.
Fluke Season 2nd-best Season (1893-)
NAME YEAR MLV PA YEAR MLV PA GAP
--------------- ----- ------ ---- ----- ----- ---- -----
DUFFY,HUGH 1894 88.8 606 1893 29.2 611 59.6
One could argue that there still is something not quite right about the list above. Norm Cash, while never quite reaching the heights of 1961 again, put together a more than respectable career, with a lifetime OBP of .371 and a SLG of .488. He hit 377 HR, placing in the top 10 in HRs nine times and in OPS six times. Cash wasn't a one-season wonder -- he was a productive player throughout his career. It's just that his one season was so mind-bogglingly good that we think of him as a fluke. Luis Gonzalez is a similar case.
Let's try another example. Is Heinie Zimmerman's 1912 really more fluky than, say Les Bell's? Zimmerman was a fine hitter for many years, and placed in the top 10 in major offensive categories dozens of times. Bell's 1926 was virtually his only appearance anywhere on the leader boards in his 9 seasons in the big leagues. This is reflected in the relative magnitude of the second-best seasons: Zimmerman's in 34.2, Bell's is just 5.3 Keeping in mind that the number reflects runs above average, Bell's season arguably feels more fluky because is was the one shining moment in an otherwise ordinary career. Zimmerman's season was the standout in a career with many good years.
With that in mind, let's revise the idea of a fluke. The gap between the best and second-best seasons is still important -- let's require a 20-run difference to be a fluke candidate. Next, we'll divide the gap by the MLV of the player's second-best season. This "fluke ratio" will attempt to balance between large gaps, which we view as necessary for flukiness, and productive second-best seasons, which are more the hallmark of a good player hitting a peak than a fluke way out of the context of the rest of his career.
For example, the lowest "fluke ratio" among the 65 players in the 20 run gap group is Hank Aaron, who posted a 94.1 MLV in 1959, and a 73.4 MLV in 1963. Both seasons are outstanding MVP-caliber years, and few players were more consistent year-in, year-out than Hammerin' Hank. Calling 1959 a fluke season is almost an insult. His fluke ratio is 0.282, the lowest of any player with a 20+ run gap, which matches our intuition that we're not seeing what we think of as a fluke when looking at Aaron's career.
So, who are the players with the largest fluke ratios? I've listed all the players with a ratio greater than 1, meaning that their fluke season was at least twice as far from league average as their next-best season, and was 20 or more runs better than their second-best season as well.
Fluke Season 2nd-best Season FL
NAME YEAR MLV PA YEAR MLV PA GAP RATIO
-------------- ----- ----- ---- ----- ----- ---- ---- ------
GASTON,CITO 1970 49.5 627 1972 4.0 403 45.5 11.505
BELL,LES 1926 38.0 635 1929 5.3 533 32.7 6.181
VIRDON,BILL 1956 31.2 543 1955 6.2 571 25.0 4.041
HARPER,TOMMY 1970 46.5 685 1972 11.5 632 35.0 3.041
RIZZUTO,PHIL 1950 34.9 716 1942 9.9 603 25.0 2.528
COWENS,AL 1977 36.3 655 1982 10.4 607 25.9 2.482
HICKMAN,JIM 1970 46.4 608 1972 13.4 422 33.1 2.479
REYNOLDS,CARL 1930 55.9 590 1929 16.9 541 39.0 2.315
PEARSON,ALBIE 1963 35.8 673 1961 12.1 526 23.8 1.970
SANDERS,REGGIE 1995 48.2 561 1992 16.9 437 31.4 1.859
CASH,NORM 1961 94.7 668 1965 34.1 548 60.5 1.775
PETROCELLI,R 1969 56.0 634 1971 20.6 646 35.4 1.721
HOFMAN,SOLLY 1910 41.2 542 1909 15.4 581 25.8 1.679
SAX,STEVE 1986 46.9 695 1989 17.7 704 29.2 1.648
DEMETER,DON 1962 40.2 601 1965 15.3 418 24.9 1.630
BIGBEE,CARSON 1922 35.6 671 1921 14.2 673 21.4 1.513
BOONE,BRET 2001 50.9 ??? 1994 20.6 419 30.3 1.471
HOLMES,TOMMY 1945 74.8 710 1944 31.6 694 43.2 1.369
VOSMIK,JOE 1935 51.5 683 1934 21.8 440 29.7 1.363
PORTER,DARRELL 1979 37.0 662 1983 15.8 515 21.2 1.345
GENTILE,JIM 1961 71.1 593 1960 30.9 459 40.2 1.303
DUFFY,HUGH 1894 88.8 606 1891 39.9 601 48.9 1.228
PARRISH,LARRY 1979 43.0 587 1986 19.4 518 23.6 1.221
TRAVIS,CECIL 1941 61.3 661 1938 28.1 629 33.2 1.182
SEYMOUR,CY 1905 66.4 634 1903 31.4 594 35.0 1.115
SEWELL,JOE 1923 51.4 658 1926 24.5 651 26.9 1.098
MOSTIL,JOHNNY 1926 44.6 689 1925 21.4 707 23.2 1.083
CHASE,HAL 1916 41.4 562 1906 20.1 613 21.4 1.065
SCHOENDIENST,R 1953 42.7 624 1952 21.0 662 21.8 1.037
GONZALEZ,LUIS 2001 104.7 ??? 1999 51.5 693 53.2 1.033
ANDERSON,BRADY 1996 57.6 677 1997 28.6 693 29.0 1.013
A familiar name is on top, though Cito Gaston is better known today as a former manager than for his playing career. Cito set career highs in batting average (.318, next best was .269, on-base percentage (.364 vs. .313), and slugging percentage (.543 vs. .405). His lifetime line of .256/.298/.397 doesn't make you think All-Star, but for one shining moment in 1970, he was. Don't feel too bad for him, though. The two World Series he won as a manager more than make up for his ranking atop our fluke list.
A few other names on the list are familiar. Reggie Sanders could never stay healthy long enough to live up to the hype. Brady Anderson's power surge makes the list, as does Rico Petrocelli's 1969 when he joined Ernie Banks as the other pre-Trinity shortstop to slam 40 HRs in a season. I can still remember reading a baseball article after the 1977 baseball season that concluded with the rhetorical question: "How would you like Al Cowens' future?" In hindsight, Dave Winfield's future (who was the same age as Cowens) would have been a better choice.
Luis Gonzalez falls pretty far down this list. Although he's on pace for the second-largest fluke gap ever, he was a good enough hitter in both 1999 and 2000 that he doesn't come close to topping the fluke ratio list. The main reason is that his second-best MLV is higher than any player on either of the lists above, other than Hack Wilson's 1929. It's very hard to put as much distance between your first and second-best seasons as between your second best and league average when your second best is so good itself.
Bret Boone makes the list at around No. 17. No matter how you look at it, whether gap or ratio, Boone is having one of the flukiest seasons on record, but still isn't close to the biggest flukes ever. Norm Cash or Cito Gaston probably deserve that "honor" -- Cash, for the sheer magnitude of the fluke; and Gaston, since he was never appreciably above average any other year. In Boone's defense, one could argue that Boone is getting shafted because we didn't consider that Boone's second-best year, 1994, was a strike-shortened one, and he may have gone on to post a high MLV had their not been a labor stoppage.
Of course, the deeper truth is we don't really know if this is a fluke season for Boone or not. He could establish this as a new level of performance, and continue hitting like this for several more years. In retrospect, 2001 would look like the year he stepped it up, rather than a one-year aberration in an otherwise undistinguished career. The players on the list above don't have the same benefit, since their careers are over or in decline. To see whether Bret Boone or Luis Gonzalez will stay on this list, we'll have to look to 2002 and beyond.
The team of writers from the Baseball Prospectus (tm) writes twice a week for ESPN.com during the baseball season. You can check out more of their work at their web site at baseballprospectus.com. Keith Woolner can be reached at kwoolner@baseballprospectus.com
| |
|
