<$BlogRSDUrl$>

Monday, September 29, 2008

Part II: In Which Corporate Cargo Cults
Bruce & Batter Starting Pitchers  

As I told you in Part I, I’d try to illuminate the second reason San Francisco Chronicle columnist Bruce Jenkins’ lovely feature (supporting his idea that pitchers should go the distance as pitchers in the 50s and 60s did) is a romantic fantasy that's eviscerated by the facts.

Just to remind you, I addressed briefly the more difficult to quantify reason, injuries, in Part I. Part II addresses the better and more reasonably measurable issue: in general, starters perform less well the 3rd and subsequent time through a line-up than they did the 1st two times.

While we look at the stats of the great workhorse pitchers of the 50s and 60s with awe, not just for their complete game and innings-pitched numbers, we also have to be impressed with their overall quality compared to the league.

But I’m going to illustrate here, using the very workhorses that Bruce Jenkins cited, that they were not the same pitcher in the 3rd and subsequent time through the opposition lineup as their overall stats indicate. In fact, you’ll see that more often than not, not only is the best reliever preventing more offense than the workhorse’s 3rd and subsequent time through, but the 2nd-best reliever is generally better at stopping offense, too.

#1 – Overall, starters aren’t as good the 3rd & subsequent time through the line-up as they are the 1st and 2nd time.

I’ll use two stats to illustrate this for you.

As a measure of performance, I’m using OPS (a stat that summarizes both the frequency batters get on base through hits or walks, combined with the frequency of power hits such as doubles and home runs they get) as a measure of what offense a pitcher yields. To compare RELATIVE performance in different situations (through the start and middle of a game versus the later-in-the-game times managers normally need to think about using a reliever to finish off the start’s efforts), I’m using TOPS+.

TOPS+ is the ratio of what the pitcher surrendered in specific situations relative to his overall performance. As an example, if a pitcher overall gave up an OPS of .759, but gave up an OPS of 626 the first time he faced a batter in a game, he’d have a TOPS+ of 94 when facing batters the 1st time (100 is the pitcher’s average performance, lower is better pitching performance, so a TOPS+ below 100 is better than his average).

So, to repeat from Part I…

Times Facing Opp. in Game (Major League Composite, 2008)
  LOWER IS BETTER
     Measure            1st time   2nd time   3rd+ time(s)
   TOPS+                  94       104        113
   OPS                  .726      .765       .800

In 2008 to date, in general, starters as a whole are less effective in a batter’s 2nd plate appearance than in the batter’s 1st, and less effective still in the batter’s 3rd and subsequent appearances against them in the game.
I won’t go into the reasons some advantage shifts to the batter as the game wears on – it’s a well-trodden area worth an essay of this length.

To reiterate, Starters IN GENERAL aren’t as good the 3rd & subsequent time through the line-up as they are the 1st and 2nd time.

#2 – Jenkins’ Workhorse heroes have roughly the same pattern of declining performance in 3rd and subsequent times through the line-up as less illustrious starters do.

So let’s look at Bruce Jenkins’ poster dudes for whom we have data (Baseball-Reference has the numbers going back to 1956). I’m omitting Bob Feller because his career ended too soon for Retrosheet's tracking data. Warren Spahn & Robin Roberts get an asterisk for having a significant portion (the entire first half) of their careers unfolding before the era that Retrosheet has tallied for by-appearance stats. Jim Kaat gets an asterisk because he worked almost 9% of his innings as a reliever, juicing his 1st-time-through-the-line-up numbers in a way that would punish Jenkins’ assertion unfairly.

WORKHORSES' Career TOPS+ (100= average for that pitchers; lower is better for the pitcher)

  Pitcher            1st time   2nd time   3rd+ time(s)
  Marichal            78        100        119
  Seaver              94        100        107
  Ryan               100         93        107
  Spahn*              87         96        116
  Roberts*            92         94        113
  Kaat*               94         97        111
  2008 All Starters   94        104        113

The Great Workhorse pitchers Jenkins cited share, with one exception, the pattern that the entire pool of 2008 starting pitchers do: they are most effective the first time through the lineup. They are measurably, but not alarmingly, less effective the second time through, and significantly less effective the third and subsequent times through. The biggest visible difference between the Workhorses and the 2008 generic hoi polloi is the generic pitchers of 2008 lost more effectiveness in the second time through.
You already saw the exception is Nolan Ryan. While he’s still less effective the 3rd+ time through, he’s actually better the 2nd time through. The difference is wholly in allowing fewer walks the 2nd time through. Another factor that makes Ryan an interesting outlier.

BTW, I scanned through these numbers for all the starters who led their league in complete games at least twice from 1950 through 1969, and the ones who had at least half their career stats in 1956 or later matched the pattern shown as normal in the chart above. There were no more outliers.

OKAY… so even The Great Workhorses are lesser pitchers the 3rd+ time through the line-up, but (Bruce and his BITGODS are asking now) ¿Surely a great pitcher’s relatively weaker effort still trumps a reliever’s normal effort, yes?
On to…

#3 – How do relievers’ efforts (in general) compare to starters efforts the 3rd+ time through the line-up (in general)

Here’s the 2008 data for all major league starters’ 3rd+ time through compared to all relievers’ efforts.
LOWER IS BETTER

                           Avg   OBA  SLG    OPS
All 2008 Starters 3+ .283 .347 .454 .800 All 2008 relievers .254 .333 .397 .730

In the general case, relievers as used in 2008 were more effective than starters-in-their-3rd+ - time through the lineup.
As far as the datums are concerned, the BITGODs are wrong. They would say, though, “Well, that’s because these 2008 starters are babied; if they were like they guys Back In The Good Old Days, their performance would exceed that of their relievers. Which brings us to …

#4 - What’s the difference between a Great Workhorse pitcher’s 3rd+ time through compared to the relievers who came in to give them succor?
To answer this question, we need to use a different method, because there’s no data I can grab that would summarize all the data for relievers who were used in significant innings to relieve Great Workhorse starters over the Workhorses’ entire careers.
To see if there’s a trend, I’m going to take the “most average” season I can find for each of Jenkins’ Workhorses, and compare their seasonal marks for 3rd+ time through the line-up against the relievers who could have come in to take over. If the starter’s less-effective 3rd+ time through are still more effective than the relievers who worked with them that season, it’s an argument the manager frittered away an advantage. But if the reliever’s mark is better than the starter’s less-effective 3rd+ time through, it’s an argument

JUAN MARICHAL
Here’s Marichal’s impressive career line:
  Year Ag Tm  Lg  W   L   G   GS  CG  ERA *lgERA *ERA+ WHIP
  +--------------+---+---+---+---+---+-----+-----+----+-----+
  1960 22 SFG NL   6   2  11  11   6  2.66  3.50  132 1.07
  1961 23 SFG NL  13  10  29  27   9  3.89  3.83   98 1.24
  1962 24 SFG NL  18  11  37  36  18  3.36  3.78  113 1.23 
  1963 25 SFG NL  25   8  41  40  18  2.41  3.19  132 0.99 
  1964 26 SFG NL  21   8  33  33  22  2.48  3.57  144 1.08 
  1965 27 SFG NL  22  13  39  37  24  2.13  3.61  169 0.91 
  1966 28 SFG NL  25   6  37  36  25  2.23  3.71  167 0.85 
  1967 29 SFG NL  14  10  26  26  18  2.76  3.34  121 1.17 
  1968 30 SFG NL  26   9  38  38  30  2.43  2.99  123 1.04 
  1969 31 SFG NL  21  11  37  36  27  2.10  3.53  168 0.99 
  1970 32 SFG NL  12  10  34  33  14  4.12  4.01   97 1.30
  1971 33 SFG NL  18  11  37  37  18  2.94  3.44  117 1.07 
  1972 34 SFG NL   6  16  25  24   6  3.71  3.53   95 1.34
  1973 35 SFG NL  11  15  34  32   9  3.82  3.85  101 1.29
  1974 36 BOS AL   5   1  11   9   0  4.87  3.88   80 1.30
  1975 37 LAD NL   0   1   2   2   0 13.50  3.42   25 2.66
  +--------------+---+---+---+---+---+-----+-----+----+-----+
  16 Yr WL% .631 243 142 471 457 244  2.89  3.55  123 1.10
  +--------------+---+---+---+---+---+-----+-----+----+-----+
  162 Game Avg    17  10  34  33  17  2.89  3.55  123 1.10

His Career ERA+ average was 123 (ERA 23% better than the league average). If you look, you’ll notice that in 1968 Marichal had and ERA+ of 123, the exact same as his career composite. 1967 is pretty close; we can look at that, too.

He’s pretty impressive, even at 9% worse relative to his overall 1968 average, he’s yielding an OPS of only .613 in what is admittedly a pitcher’s year. How about the bullpen behind him? The two relievers with the most use were Frank Linzy & Joe Gibbon. Here are the equivalent lines for their relief work to compare w/the starter's:

1968          G   PA   AB   R   H  2B 3B HR  BB   BA   OBP   SLG   OPS  
+-+----------+---+----+----+---+---+--+--+--+---++-----+-----+-----+-----
Marichal 3rd+ 37  561  530  41 130 13  3 10  19  .245  .275  .338  .613 
Linzy         57  340  304  38  63  9  3  1  25  .207  .265  .266  .532 
Gibbon        29  154  128  12  30  2  0  3  17  .234  .331  .320  .651 

So Linzy is clearly better than Marichal. And Gibbon, the second-best reliever is about as good, though not quite. Herman Franks, one of the fathers of modern bullpen usage, wisely noticed Linzy was better than Marichal the 3rd time through the line-up and that Gibbon, his second best, was about a good, if you could absorb Gibbon's higher walk rate.

Franks, btw, also had a swing man he used as a reliever in 15 games, Bobby Bolin. Here’s the fireballer’s 1968 line:
I Split       G   PA   AB   R   H  2B 3B HR  BB    BA   OBP   SLG   OPS 
+-+----------+---+----+----+---+---+--+--+--+---++-----+-----+-----+----
Bolin         15  155  143   9  30  7  1  2  10   .210  .261  .315  .576

Bolin was their long relief guy (15 games, 155 plate appearances, meaning he was facing about 10 batters per relief appearance). His OPS at .576 was better than Marichal’s .613 in Juan’s 3rd+ time through the line-up, too.

In 1967, Marichal had his next-closest to average year (OPS of 121, a little less great than his career 123). Here’s Marichal’s line in 1967 the 3rd+ time through the line-up, as well as stats for the closer, Linzy, and a couple of other available relievers that year:

1967          G   PA   AB   R   H  2B 3B HR  BB    BA   OBP   SLG   OPS  
------------+---+----+----+---+---+--+--+--+---+-----+-----+-----+-----+
Marichal 3rd+ 26  343  317  40  88  6  1  8  19  .278  .320  .379  .699 
Linzy         57  378  332  21  67  2  1  4  34  .202  .273  .250  .523 
McDaniel      38  249  226  25  55  7  2  4  17  .243  .300  .345  .645 
Bolin         22  192  162  16  38  4  0  2  25  .235  .337  .296  .633 
Gibbon        18  141  121  13  24  6  1  1  15  .198  .292  .289  .581 

Another year of performance for Marichal close to his career average , and another year that Linzy (and at least one other reliever) was good enough to be his equal at the end of a game. While the team also had some relievers who underperformed Marichal’s 3rd+ time through performance, Marichal’s teams didn’t have to suffer when they relieved for him. Relieving Marichal helped his overall performance AND his team’s performance.

TOM SEAVER
Here’s Tom Terrific’s career

  Year Ag Tm  Lg  W   L   G   GS  CG   ERA *lgERA *ERA+ WHIP
  +--------------+---+---+---+---+---++-----+-----+----+----
  1967 22 NYM NL  16  13  35  34  18   2.76  3.38  122 1.20
  1968 23 NYM NL  16  12  36  35  14   2.20  3.02  137 0.98
  1969 24 NYM NL  25   7  36  35  18   2.21  3.63  165 1.03
  1970 25 NYM NL  18  12  37  36  19   2.82  4.01  142 1.07
  1971 26 NYM NL  20  10  36  35  21   1.76  3.40  193 0.94
  1972 27 NYM NL  21  12  35  35  13   2.92  3.35  115 1.11
  1973 28 NYM NL  19  10  36  36  18   2.08  3.63  175 0.97
  1974 29 NYM NL  11  11  32  32  12   3.20  3.59  112 1.16
  1975 30 NYM NL  22   9  36  36  15   2.38  3.45  145 1.08
  1976 31 NYM NL  14  11  35  34  13   2.59  3.30  127 1.06
  1977 32 TOT NL  21   6  33  33  19   2.58  3.87  150 1.01
  1978 33 CIN NL  16  14  36  36   8   2.88  3.58  124 1.18
  1979 34 CIN NL  16   6  32  32   9   3.14  3.78  120 1.15
  1980 35 CIN NL  10   8  26  26   5   3.64  3.61   99 1.18
  1981 36 CIN NL  14   2  23  23   6   2.54  3.56  140 1.11
  1982 37 CIN NL   5  13  21  21   0   5.50  3.68   67 1.61
  1983 38 NYM NL   9  14  34  34   5   3.55  3.64  103 1.24
  1984 39 CHW AL  15  11  34  33  10   3.95  4.16  105 1.17
  1985 40 CHW AL  16  11  35  33   6   3.17  4.32  136 1.22
  1986 41 TOT AL   7  13  28  28   2   4.03  4.25  106 1.33
  +--------------+---+---+---+---+---++-----+-----+----+----
  20 Yr WL% .603 311 205 656 647 231   2.86  3.64  127 1.12
  +--------------+---+---+---+---+---++-----+-----+----+----
  162 Game Avg    16  10  34  33  12   2.86  3.64  127 1.12

Seaver’s career ERA+ is 127 and the seasons he finished closest to that were 1976 (127) and 1978 (124). Here’s his breathtaking line for 3rd time through in 1976, with parallel lines for some of the relievers Mets manager Joe Frazier had in the bullpen: Skip Lockwood, Bob Apodaca & Daffy Sanders.

            G   PA   AB   R   H  2B 3B HR  BB    BA   OBP   SLG   OPS 
----------+---+----+----+---+---+--+--+--+---++-----+-----+-----+-----
Seaver 3rd+ 34  438  401  37  88 14  2  8  32   .219  .280  .324  .604
Lockwood    56  375  333  31  62  6  2  6  34   .186  .265  .270  .535
Apodaca     40  292  251  20  51 14  0  1  25   .203  .280  .271  .551
Sanders     34  202  180  16  42  8  0  4  15   .233  .291  .344  .636

The most- and second most effective relievers were more effective than Seaver, and the third most effective was definitely not as good, but no slouch either.

How about 1978?

Here is Seaver the 3rd+ time and his relievers that year. Doug Bair was the fireman Reds' manager Sparky Anderson used to close games. Manny Sarmiento was used as the next-best reliever, Tom “The Scottish Skeptic” Hume was a swing man, and Pedro Borbon, Sr., was the mop-up pitcher, meant to absorb punishment when the team got blown out early.

              G   PA   AB   R   H  2B 3B HR  BB    BA   OBP   SLG   OPS  
+-----------+---+----+----+---+---+--+--+--+---++-----+-----+-----+-----+
Seaver 3rd+   35  405  362  34  89 18  4 12  34   .246  .308  .417  .725 
Bair          70  416  369  23  87  8  2  6  38   .236  .305  .317  .622 
Borbon        62  418  372  56 102 23  1  6  27   .274  .324  .390  .713 
Sarmiento     59  443  386  56  92 15  3 14  43   .238  .311  .402  .713 
Hume          19  135  124  11  31  3  2  2   8   .250  .293  .355  .648 

All of them, even the mop-up gent, equaled Seaver’s 3rd+ time effectiveness. In average Seaver years, relieving Seaver wasn’t hurting his team’s chances of winning games.

WARREN SPAHN
Half of Spahn’s career is outside the range of Retrosheet’s detailed stats, so we’re going to miss the season that’s the closest match to his career average (1954), but we still get to look at 1959, a close, but slightly better ERA+ season.

  Year Ag Tm  Lg  W   L   G   GS  CG  ERA *lgERA *ERA+ WHIP
  +--------------+---+---+---+---+---+-----+-----+----+-----+
  1942 21 BSN NL   0   0   4   2   1  5.74  3.31   58 2.29
  1946 25 BSN NL   8   5  24  16   8  2.94  3.45  117 1.13
  1947 26 BSN NL  21  10  40  35  22  2.33  3.90  168 1.13
  1948 27 BSN NL  15  12  36  35  16  3.71  3.84  103 1.22
  1949 28 BSN NL  21  14  38  38  25  3.07  3.80  124 1.22
  1950 29 BSN NL  21  17  41  39  25  3.16  3.85  122 1.22
  1951 30 BSN NL  22  14  39  36  26  2.98  3.68  123 1.24
  1952 31 BSN NL  14  19  40  35  19  2.98  3.62  121 1.15
  1953 32 MLN NL  23   7  35  32  24  2.10  3.94  188 1.05
  1954 33 MLN NL  21  12  39  34  23  3.14  3.75  119 1.22
  1955 34 MLN NL  17  14  39  32  16  3.26  3.76  115 1.27
  1956 35 MLN NL  20  11  39  35  20  2.78  3.47  125 1.07
  1957 36 MLN NL  21  11  39  35  18  2.69  3.49  130 1.17
  1958 37 MLN NL  22  11  38  36  23  3.07  3.52  114 1.14
  1959 38 MLN NL  21  15  40  36  21  2.96  3.55  120 1.20
  1960 39 MLN NL  21  10  40  33  18  3.50  3.42   98 1.22
  1961 40 MLN NL  21  13  38  34  21  3.02  3.71  123 1.14
  1962 41 MLN NL  18  14  34  34  22  3.04  3.78  124 1.12
  1963 42 MLN NL  23   7  33  33  22  2.60  3.22  124 1.11
  1964 43 MLN NL   6  13  38  25   4  5.29  3.54   67 1.47
  1965 44 TOT NL   7  16  36  30   8  4.01  3.54   88 1.34
  +--------------+---+---+---+---+---+-----+-----+----+----
  21 Yr WL% .597 363 245 750 665 382  3.09  3.65  118 1.19
  +--------------+---+---+---+---+---+-----+-----+----+----
  162 Game Avg    17  11  36  31  18  3.09  3.65  118 1.19

You know the drill by now.

  I Split          G   PA   AB   R   H  2B 3B HR  BB    BA   OBP   SLG   OPS
  +-+------------+---+----+----+---+---+--+--+--+---++-----+-----+-----+-----
  Spahn 3rd+      34  490  458  45 126 22  2  7  22   .275  .309  .378  .687
  McMahon         60  356  310  26  81 10  2  5  37   .261  .338  .355  .693
  Rush            22  202  187  12  41  6  2  2  11   .219  .265  .305  .570
  Jay             15  119  102  14  23  3  0  3  13   .225  .322  .343  .665

BTW: Spahn’s 1959 performance was better the 2nd time through the lineup than the 1st time through, though he still decayed the 3rd+ time through.

Manager Fred Haney used his #6, #5, #4 and even his #3 starters in relief more often than most, in part because his #1 and #2 (Burdette and Spahn) started so often.

The marquee closer on this team was Don “Man on a White Horse” McMahon, and his OPS is a tad less good than Spahn’s, certainly not a big shear-off. Bob “I’m Just Lookin’ For Some” Rush and Joey Jay, both of whom got a decent number of starts, were somewhat better in relief than Spahn the 3rd time through.

In 1954, the Braves had three relievers with ERAs lower than Spahn’s but without the stats to support it, and given the unreliability of using ERA as a way to measure reliever effectiveness, we can’t draw any conclusions for that year.

Given the data we have for the second half of Spahn’s career, there’s no support for the idea that the Braves needed Spahn to pitch complete games if they wanted to win the games he appeared in.

NOLAN RYAN
Here’s the career of the most distinctive outlier in Jenkins’ pile.

  Year Ag Tm  Lg  W   L   G   GS  CG    ERA *lgERA *ERA+ WHIP
  +--------------+---+---+---+---+---++-----+-----+----+----
  1966 19 NYM NL   0   1   2   1   0  15.00  3.61   24 2.66
  1968 21 NYM NL   6   9  21  18   3   3.09  3.02   98 1.25
  1969 22 NYM NL   6   3  25  10   2   3.53  3.63  103 1.26
  1970 23 NYM NL   7  11  27  19   5   3.42  4.01  117 1.39
  1971 24 NYM NL  10  14  30  26   3   3.97  3.40   86 1.58
  1972 25 CAL AL  19  16  39  39  20   2.28  2.92  128 1.13
  1973 26 CAL AL  21  16  41  39  26   2.87  3.52  122 1.22
  1974 27 CAL AL  22  16  42  41  26   2.89  3.41  118 1.27
  1975 28 CAL AL  14  12  28  28  10   3.45  3.52  102 1.43
  1976 29 CAL AL  17  18  39  39  21   3.36  3.31   99 1.32
  1977 30 CAL AL  19  16  37  37  22   2.77  3.91  141 1.34
  1978 31 CAL AL  10  13  31  31  14   3.72  3.63   98 1.41
  1979 32 CAL AL  16  14  34  34  17   3.60  4.06  113 1.27
  1980 33 HOU NL  11  10  35  35   4   3.35  3.28   98 1.29
  1981 34 HOU NL  11   5  21  21   5   1.69  3.28  194 1.12
  1982 35 HOU NL  16  12  35  35  10   3.16  3.32  105 1.21
  1983 36 HOU NL  14   9  29  29   5   2.98  3.38  114 1.19
  1984 37 HOU NL  12  11  30  30   5   3.04  3.31  109 1.15
  1985 38 HOU NL  10  12  35  35   4   3.80  3.45   91 1.29
  1986 39 HOU NL  12   8  30  30   1   3.34  3.58  107 1.12
  1987 40 HOU NL   8  16  34  34   0   2.76  3.92  142 1.13
  1988 41 HOU NL  12  11  33  33   4   3.52  3.32   94 1.24
  1989 42 TEX AL  16  10  32  32   6   3.20  3.97  124 1.08
  1990 43 TEX AL  13   9  30  30   5   3.44  3.96  115 1.03
  1991 44 TEX AL  12   6  27  27   2   2.91  4.06  139 1.00
  1992 45 TEX AL   5   9  27  27   2   3.72  3.83  103 1.31
  1993 46 TEX AL   5   5  13  13   0   4.88  4.16   85 1.41
  +--------------+---+---+---+---+---++-----+-----+----+----
  27 Yr WL% .526 324 292 807 773 222   3.19  3.56  111 1.24
  +--------------+---+---+---+---+---++-----+-----+----+----
  162 Game Avg    13  12  34  33   9   3.19  3.56  111 1.24

The seasons closest to his career averages are 1979 and 1984.

   1979            G   PA   AB   R   H  2B 3B HR  BB     BA   OBP   SLG  OPS 
  +-+------------+---+----+----+---+---+--+--+--+---++-----+-----+-----+-----+
  Ryan 3rd+ PA    29  359  303  32  61 11  1  6  47   .201  .311  .304  .615  
  Clear           52  481  398  48  87  8  1  6  68   .219  .333  .289  .622  
  La Roche        52  373  328  52 104 15  1 12  32   .317  .375  .479  .854  
  Barlow          35  378  338  54 106 13  4  8  30   .314  .373  .447  .820  

The best reliever of the bunch was the closer, Mark Clear. He’s not quite as good as Ryan. I’ve shown the most-used arms out of the ‘pen; there are others, but they are worse.

In 1984, Ryan reversed the normal pattern, with his performance improving each time through the line-up.

  1984           G   PA   AB   R   H  2B 3B HR  BB    BA   OBP   SLG   OPS  
  +-+----------+---+----+----+---+---+--+--+--+---++-----+-----+-----+-----+
  Ryan 3rd+ PA   24  240  220  21  44  4  3  4  16   .200  .255  .300  .555
  Dawley         60  402  351  24  82 13  4  5  35   .234  .298  .336  .635
  DiPino         57  329  285  32  74  9  0  3  36   .260  .343  .323  .665
  Smith          53  304  280  22  60  6  3  5  20   .214  .268  .311  .579
  LaCoss         21  170  146  16  35  5  1  1  20   .240  .331  .308  .640

Again in 1984, Ryan was just plain better in his 3rd time through than any reliever you might send in for him. Not that this group is bad…they’re a pretty good group (and it wasn’t just playing half their games in the Astrodome that built their stats). They’re just not as good as Ryan the 3rd time through.
This is the case Bruce Jenkins has been waiting for – where the Great Workhorse is actually better at the end of the game than anyone in the bullpen even when they’re fresh.

ROBIN ROBERTS
Like Spahn, we don’t have Retrosheet data for the first half of Roberts’ most excellent career:

  Year Ag Tm  Lg  W   L   G   GS  CG    ERA *lgERA *ERA+ WHIP
  +--------------+---+---+---+---+---++-----+-----+----+----
  1948 21 PHI NL   7   9  20  20   9   3.19  3.95  124 1.42
  1949 22 PHI NL  15  15  43  31  11   3.69  3.96  107 1.34
  1950 23 PHI NL  20  11  40  39  21   3.02  4.06  135 1.18
  1951 24 PHI NL  21  15  44  39  22   3.03  3.84  127 1.10
  1952 25 PHI NL  28   7  39  37  30   2.59  3.66  141 1.02
  1953 26 PHI NL  23  16  44  41  33   2.75  4.20  152 1.11
  1954 27 PHI NL  23  15  45  38  29   2.97  4.03  136 1.02
  1955 28 PHI NL  23  14  41  38  26   3.28  3.96  121 1.13
  1956 29 PHI NL  19  18  43  37  22   4.45  3.73   84 1.23
  1957 30 PHI NL  10  22  39  32  14   4.07  3.80   93 1.15
  1958 31 PHI NL  17  14  35  34  21   3.24  3.95  122 1.19
  1959 32 PHI NL  15  17  35  35  19   4.27  4.10   96 1.17
  1960 33 PHI NL  12  16  35  33  13   4.02  3.88   96 1.22
  1961 34 PHI NL   1  10  26  18   2   5.85  4.07   70 1.51
  1962 35 BAL AL  10   9  27  25   6   2.78  3.69  133 1.13
  1963 36 BAL AL  14  13  35  35   9   3.33  3.45  104 1.07
  1964 37 BAL AL  13   7  31  31   8   2.91  3.59  123 1.25
  1965 38 TOT     10   9  30  25   8   2.78  3.43  123 1.05
  1966 39 TOT NL   5   8  24  21   2   4.82  3.51   73 1.44
  +--------------+---+---+---+---+---++-----+-----+----+----
  19 Yr WL% .539 286 245 676 609 305   3.41  3.86  113 1.17
  +--------------+---+---+---+---+---++-----+-----+----+----
  162 Game Avg    15  12  35  32  16   3.41  3.86  113 1.17

You’ll notice that for the first time, we don’t have a tidy single season to use as an exemplar for his career. In 1957-59, he was close in several ways, not in ERA+, but fairly close in baserunners per inning (WHIP in the table). Of those three years, Roberts’ strongest relative 3rd+ time through was 1957 (better performance than the 2nd time through), and 1958 has the weakest bullpen, so I’ll show them both. In both, Turk Farrell had the linchpin role and Jim Hearn and others backed him up.

  1957            G   PA   AB   R   H  2B 3B HR  BB     BA   OBP   SLG   OPS 
  +-+------------+---+----+----+---+---+--+--+--+---++-----+-----+-----+-----
  Roberts 3rd+    30  371  346  53  89 15  3 17  17   .257  .291  .465  .757 
  Farrell         52  353  305  29  74  7  4  2  36   .243  .324  .311  .635 
  RJ Miller       31  235  215  17  56  9  4  4  14   .260  .302  .395  .697 
  Hearn           32  217  202  22  50 10  1  5  11   .248  .290  .381  .671 
  Morehead        33  221  194  26  53 15  3  1  19   .273  .338  .397  .735 


  1958            G   PA   AB   R   H  2B 3B HR  BB     BA   OBP   SLG   OPS 
  +-+------------+---+----+----+---+---+--+--+--+---++-----+-----+-----+-----
  Roberts 3rd+    32  458  429  47 120 20  4 17  20   .280  .310  .464  .774 
  Farrell         54  401  345  41  84 13  5  7  40   .243  .320  .371  .691 
  Hearn           38  302  269  43  80 10  6  6  26   .297  .357  .446  .803 
  Meyer           32  248  217  17  47  4  2  2  25   .217  .299  .281  .580 

In 1957, manager Mayo Smith had plenty of effective bullpen options to go to for Roberts. In 1958, Smith and successor Eddie Sawyer had a couple of more effective choices (and some scary implode-o-rama Human Hand Grenades).

As with Spahn, given the data for the second half of Roberts’ career, there’s no support for the idea that the Phillies needed Roberts to pitch complete games if they wanted to win the games he appeared in.

JIM KAAT
Here’s Kaat’s marathon career

  Year Ag Tm  Lg  W   L   G   GS  CG   ERA *lgERA *ERA+ WHIP
  +--------------+---+---+---+---+---++-----+-----+----+----
  1959 20 WSH AL   0   2   3   2   0  12.60  3.94   31 2.20
  1960 21 WSH AL   1   5  13   9   0   5.58  3.95   71 1.58
  1961 22 MIN AL   9  17  36  29   8   3.90  4.27  109 1.34
  1962 23 MIN AL  18  14  39  35  16   3.14  4.09  130 1.18
  1963 24 MIN AL  10  10  31  27   7   4.19  3.67   88 1.30
  1964 25 MIN AL  17  11  36  34  13   3.22  3.59  111 1.19
  1965 26 MIN AL  18  11  45  42   7   2.83  3.56  126 1.24
  1966 27 MIN AL  25  13  41  41  19   2.75  3.61  131 1.07
  1967 28 MIN AL  16  13  42  38  13   3.04  3.46  114 1.18
  1968 29 MIN AL  14  12  30  29   9   2.94  3.13  106 1.11
  1969 30 MIN AL  14  13  40  32  10   3.49  3.70  106 1.34
  1970 31 MIN AL  14  10  45  34   4   3.56  3.79  107 1.31
  1971 32 MIN AL  13  14  39  38  15   3.32  3.54  107 1.23
  1972 33 MIN AL  10   2  15  15   5   2.06  3.22  156 1.00
  1973 34 TOT AL  15  13  36  35  10   4.37  3.98   91 1.30
  1974 35 CHW AL  21  13  42  39  15   2.92  3.77  129 1.17
  1975 36 CHW AL  20  14  43  41  12   3.11  3.90  125 1.31
  1976 37 PHI NL  12  14  38  35   7   3.48  3.58  103 1.19
  1977 38 PHI NL   6  11  35  27   2   5.39  4.03   75 1.56
  1978 39 PHI NL   8   5  26  24   2   4.10  3.58   87 1.29
  1979 40 TOT      3   3  43   2   0   3.92  4.03  103 1.38
  1980 41 TOT      8   8  53  14   6   3.94  3.72   94 1.37
  1981 42 STL NL   6   6  41   1   0   3.40  3.56  105 1.45
  1982 43 STL NL   5   3  62   2   0   4.08  3.64   89 1.36
  1983 44 STL NL   0   0  24   0   0   3.89  3.64   93 1.67
  +--------------+---+---+---+---+---++-----+-----+----+----
  25 Yr WL% .544 283 237 898 625 180   3.45  3.71  107 1.25
  +--------------+---+---+---+---+---++-----+-----+----+----
  162 Game Avg    12  10  40  27   8   3.45  3.71  107 1.25

There are five years that are close enough for (inhuman) hand grenades or horseshoes. I’m going to use 1970 and 1971 to exemplify seasons that are close to “average” for Kitty’s kareer.

In 1970, manager Bill Rigney had a great bullpen. And it shows in the way he used the durable Kaat – allowing him to go the distance only four times. In 1971, Rigney’s bullpen ace, Ron Perranowski turned into a blancmange, the much of the ‘pen was shaky, and Kitty went the complete game 15 times.

  1970            G   PA   AB   R   H  2B 3B HR  BB    BA   OBP   SLG   OPS 
  +-+------------+---+----+----+---+---+--+--+--+---++-----+-----+-----+-----
  Kaat 3rd+       32  279  251  36  83  9  2 13  20   .331  .379  .538  .917 
  Perranowski     67  471  417  38 108  6  3  7  42   .259  .325  .338  .664 
  S. Williams     68  456  408  34  85 11  2  8  32   .208  .271  .304  .575 
  T. Hall         41  317  275  27  47  8  5  6  36   .171  .268  .302  .570 
  Zepp            23  136  130  10  28  4  1  3   5   .215  .243  .331  .573 

1971 G PA AB R H 2B 3B HR BB I BA OBP SLG OPS +-+-----------+---+----+----+---+---+--+--+--+---+------+-----+-----+-----+ Kaat 3rd+ 35 409 386 40 104 18 2 8 14 .269 .297 .389 .686 S. Williams 55 381 318 43 72 13 2 5 45 .226 .336 .327 .663 T. Hall 37 257 217 19 40 7 1 4 35 .184 .295 .281 .576 Perranowski 47 290 241 49 76 7 4 4 31 .315 .396 .427 .824

Even in the wobbly 1971 bullpen, the second-most effective reliever was an improvement over Kaat’s 3rd+ trip through opponents’ line-ups.

NOTE: If anyone wants to crawl through Kaat’s other close-to-Kaat-average seasons and see if there are counter-examples where he’s better than the second-best reliever, feel free to do it and let me know what you find.

Kaat, interesting informant to intelligent sportswriters everywhere, is yet another one of Jenkins’ Workhorses who undermine The BITGOD Prescription for which he campaigns.

SUMMARY & CONCLUSION
There are no measurements or statistics that support the idea that getting pitchers to go the distance more will help their teams or the quality of their own performances. Even ignoring the possibility of injuries or wear-ünd-tear(no mean oversight), teams with ordinary bullpens don’t risk losing more games than they would by letting their better pitchers-who-are-not-the-reincarnation-of-Nolan-Ryan pitch a gaggle of CGs.

Most pitchers who are relievers are not as good as most pitchers who are starters, and certainly the average reliever is far inferior to the top-ranked starts. BUT that doesn’t mean a decent reliever won’t outperform in his first handful of batters compared to a Great Workhorse in the latter’s 3rd+ time through the line-up.

I’m not suggesting that no young starter will ever again be an outlier in the direction of Nolan Ryan. It might turn out that Tim Lincecum (who was better his third time through the lineup than the second time through both this year and last) is the next possible one. But you can’t turn someone into Nolan Ryan by wishing or by running most young starters through a protocol that will turn most of them into the Steaming Rubble of a Brandon Duckworth career.

The BITGOD Prescription is a patent medicine, a intoxicating tonic for the nostalgic. Love it if you will, but stop pretending it’s a prescription that will make teams more successful. There’s just not evidence to support it.


Thursday, September 25, 2008

Corporate Cargo Cults:
Bruce Jenkins, The Duke of Moral Hazard &
His Young Pitcher Slaughterhouse  

These are the dark ages of pitching. It is a time of cowardice and fear, oblivious to the lessons of history. If there's a bond among starting pitchers of the pitch-count era, it's that they were born too late. One of life's great truisms is to finish what you start.

It's what you tell your kids, your surgeon, your contractor. This once applied to baseball, with precision, but now there's a new law: Just quit. Let somebody else finish the job. You did your part, now go be a cheerleader.

Pitch counts have destroyed not only the elements of pride and accomplishment among starting pitchers, but the art of winning. If one thing characterized the great pitchers of the past, from Bob Feller to Warren Spahn to Tom Seaver, it's that they learned how to win.

You don't get that from a "quality start" and a nice, early shower. It's when you understand the difference between a breezy sixth inning and a stressful ninth, when you brought that victory home, and can't wait to do it again. – Bruce Jenkins

Every line of work has its own cargo cults, a set of energetically-pursued methods based on a mass delusion. When they get institutionalised as S.O.P., they can be really scary. When San Francisco Chronicle sports columnist Bruce Jenkins campaigns for Baseball’s most resilient Cargo Cult, it’s downright Resident Evil: Apocalypse scary (truly heart-pounding but at the same time so patently ridiculous you can’t believe anyone would bother wasting their craft on it).

We’ll get back to Jenkins and baseball in a minute, but first a little more about Cargo Cults in general.

AIRSTRIPS CREATE AIRPLANES

The best known Anthropology example of the Cargo Cult are some post- WWII Pacific islander religions.

During the war, outsiders had built airfields to land or transfer supplies to support the war, and some of the material goods got skimmed off for the residents. The war ended, the shipments stopped coming through.

That ended the goodies. The residents wanted to goodies back. Having noticed that air transports followed air fields, they presumed, therefore, that air fields generated cargo planes, so they built rudimentary landing strips they assumed would generate the traffic that provided the material goods.

Like native cults, Corporate Cargo Cults are rarely fabricated out of nothing – they’re usually based on an historical success or failure of remarkable proportion.

    * “Less regulation is always better than the amount we currently have”

    * “No one ever lost their job buying Microsoft system software”

    * “Real estate is the one investment you can never lose money in”

“Leadership” tends to institutionalize whatever response seemed to work in the crisis moment even if the causality was questionable.

Later, when the crisis has passed, people forget the context of the response and merely repeat their past behavior.

As a species, we generally benefit from automatic responses (not rethinking from scratch every response to every stimulus). So, it’s easy to take a general tendency and assume that it’s going to work in all cases, regardless of context. That’s the easiest path to walk hard.

Worse than the absurdity of the Cargo Cult, one of its critical attributes is that it actually undermines the chances for organizational success.

THE DUKES OF MORAL HAZARD

There aren’t many Cargo Cults in Baseball – there are superstitions and irrelevant rituals, to be sure (a pitcher not stepping on the line when walking from mound to dugout after a half-inning, or a player not changing undershirt during a hot streak). But such quirks rarely affect performance for better or worse.

Baseball, unlike Banking or Real Estate or most finance or service endeavors, is relentlessly focused on performance right now and how actions affect performance in the future, and Baseball management relentlessly focuses on facts to fine-tune behaviors. In Baseball, unlike corporate or academic or non-profit arenas, decisions are almost always measured, and decision-makers almost always held accountable. So in Baseball, Cargo Cult adherence tends to take one out of the pool with pretty good certainty and usually with some whoop-axe alacrity, too.

Baseball’s worst Cargo Cult has been on the wane for about a decade, the religion built around the value/virtue of starting pitchers throwing complete games (sometimes also the value of having a pitcher be a 20-game winner, or throwing 250 innings for a season).

The practice was changed because of belief, substantiated heavily as the general case, that heavy use of a pitcher leads to more injuries and, in many cases, prematurely ends more careers than more regulated use does. More scientific front offices and pitching coaches followed the vanguard of data-equipped, fact-based leaders like the Oakland Athletics’ and New York Mets’ Rick Peterson and started rebuilding the protocols of when to pull a starter, especially a young one (because the effects of wear vary by many factors, but most markedly by the experience (functional age) of the pitcher).

And as the adherence to the practice has been on the wane, the Cargo Cult’s diminishing number of adherents seem, as the hangers-on to any shrinking ideology or religion do, more strident and entrenched, more sure they are right. Instead of being about team wins, the argument sounds more like an exhortation to some moral imperative, as though pitching 7 strong innings and yielding the mound to a fresh arm was un-manly (cue The Four Feathers).

Bruce Jenkins, one of the more insightful and readable columnists around, has been on a Starters Should Pitch Workloads Like They Did Back In The Good Old Days (BITGOD) crusade for a while, and his recent masterwork was a pair ‘o big features (August 26 - 27) about the lost art of the complete game.

Fun, interesting, well-written, and completely eviscerated…not by examples he omitted, but by the very examples he damningly included. His citations generally make the case AGAINST using pitchers more often and for longer outings.

I recommend you read the whole piece, but here are what I consider the essential points (after the exhortationalistic epigramme at the top of this entry).

    * Pitchers from previous eras consider regulating starters’ outing length by pitch count is absurd.

    * By definition, most relievers are generally not as good as starters (if they were as good, they’d be starters).

    * Pulling a starter and replacing him in the late innings with a lesser pitcher undermines team accomplishments.

    * Back in the 1950’s, 1960’s and 1970’s, pitchers could go complete games because it was expected of them and they were conditioned to go the distance. The reason they don’t anymore is because they’re not trained to nor expected to. Jenkins: “If your job security depends on finishing a game - with 160 pitches, if that's what it takes - then you don't think twice about it, nor does your manager, general manager or owner. The act becomes as mundane as covering first base or laying down a bunt.”.

    * Contemporary marquee starters such as Tim Lincecum, Carlos Zambrano, Dan Haren and Scott Kazmir should be notching complete games at the historical, not contemporary rate, but the new practice is depriving them of that opportunity, and perhaps they enjoy the game less, and perhaps should be respected less for their accomplishments as a result.

    * If teams not only expected more complete games, but moved from a five-man to a four-man rotation as Jenkins suggests (loading the expectation of 20% more outings per season for each starting pitcher), middle relievers would pitch less and those relievers are “pretty much a joke on many teams”.

    * Teams should launch a Counter-Reformation against the pitch count reforms and start training pitchers to go the distance. It will benefit the pitchers physically and emotionally and the teams with better potential performance, and injury rates won’t be much different from today’s pitch count-driven results.

There were some great examples of BITGOD workhorses: Robin Roberts, Warren Spahn, Juan Marichal, Tom Seaver, Bob Feller. And Nolan Ryan and Jim Kaat, both heavily-worked pitchers who worked well into their 40s. A sidebar featured pitchers less known to fans who achieved great feats of game endurance, such as Joe Oeschger and Leon Cadore who both went the distance in a 22-inning game back in 1920, and Tom Cheney, who tossed a 166-pitch (thanks, 'Neck) complete game in 1962.

THE REALITY OF THE DUKE OF MORAL HAZARD’S EXAMPLES
The reality of the pitchers he cites generally (not universally) makes a powerful counter-case to his religion.

Here’s why.

Some of those pitchers broke down because of their heroic workloads. All the others for whom we have Retrosheet data generally pitched late innings less effectively than the relievers who replaced them. That is, while a great starter’s composite performance was/is better than a merely decent reliever’s, a great starter’s late in the game (the time a reliever would generally replace the starter) performance is inferior to that of a decent reliever from that team. We have, thanks to Retrosheet.Org and Baseball Reference, detailed game-by-game and seasonal statistics for teams’ starting pitcher outings from 1956 through the present. Where I have such statistical data, we’ll look at it.

In this, Part I, entry, I’m going to cite some explanation of the pre-1956 careers of some of the workhorses Jenkins cites. In Part II of this entry, I’m going to go through all the 1956- and on pitchers Retrosheet and Baseball Reference document whose efforts Jenkins uses to support his beliefs and comment on each one.

METHOD NOTES: To compare starters with their relievers, I’m going to use ERA+ (pitcher’s ERA relative to the league average) as an indicator to determine a season that was “average” for that starter. So, for example, if Jim Kaat’s career composite ERA+ was 107 (seven percent better than the leagues he pitched in), I will find the full season he had with the ERA+ closest to 107.

Then, I’m going to find the 2nd-best reliever on the starter’s team that year. I’ll choose based on a mix of science and art – I’ll generally look over a pair of relievers: the marquee reliever and then the best from the remainder of relievers who pitched serious innings. Of that pair, I’ll choose the second best performer.

I’m choosing the second best because unlike a Strat-O-Matic tussle, a real manager has imperfect knowledge of the statistical probabilities of each individual’s success or failure. The manager finds out about effectiveness in a range of relief situations in an ongoing small-sample experiments through the season. So we need to saddle the manager with a good, not the best, choice, to relieve the great starter, a more “known” contributor.

I’m going to use the stat OPS (On Base Plus Slugging Average) as an indicator of effectiveness, of what a pitcher gave up to opponents. I’m going to compare the 2nd-best reliever’s OPS to the OPS of the subject starter’s performance facing batters in batters’ 3rd and subsequent plate appearances.

To reiterate, it doesn’t matter to the team’s overall effectiveness whether a starter’s performance the first two times he faces batters is better than a reliever’s performance; a reliever almost never replace a starter who’s cruising early in a game. But by the 3rd or 4th time in a game a starter is seeing the batter, the pitcher is more likely to have some fatigue, and the batter more locked into the pitcher’s stuff.

I’m also going to refer to TOPS+, the ratio of what the pitcher surrendered in specific situations relative to his overall performance. As an example, if a pitcher overall gave up an OPS of .759, but gave up an OPS of 626 the first time he faced a batter in a game, he’d have a TOPS+ of 94 when facing batters the 1st time (better than his average).

MISSING THE STRIKE ZONE
The complexity of analyzing Jenkins’ proposed initiative is that his argument aims at one cause and then presumes a pair of different effects. The BITGODs aim for more complete games, and moral virtue issues aside, assume that pitcher injuries won’t increase. These, I’ll call the Injury and the Moral Rectitude issues.

He also believes that the teams would surrender fewer runs/win more games if good starters both worked longer and started more often (The BITGOD Prescription). Again, Jenkins suggests that if you replace reliever innings with incrementally more great starter innings, you get better composite team pitching performance. This, I’ll call the Team Performance issue.

I’m not going to try to take on the full scope of the injury question. I think it’s a major factor, but injury data is as hard to nail down as a gyrating sea cucumber addicted to crack and hosed down with extra virgin olive oil. And Retrosheet and (even the miraculous) Baseball Reference keep no stats on Moral Rectitude, so, for now at least, I’m only going to deal with it in this Part I, and only a little.

Let’s focus for now on Team Performance. ¿Would a team’s overall pitching benefit from The BITGOD Prescription? The question should be: “Is a fresh reliever worse than a great starter by batters' 3rd and later plate appearance against that starter?” The vast majority of starters are less effective the 3rd and subsequent plate appearances against a batter than they were in the 1st plate appearance they face the batter.

  
  Times Facing Opp. in Game (Major League Composite, 2008) 
  I  Split         OPS   tOPS+ 
  +-+------------++-----++-----+ 
  1st PA in G    .726   94 
  2nd PA in G    .765  104 
  3rd+ PA in G   .800  113 

source: Baseball-Reference.Com

In 2008 to date, in general, starters as a whole are less effective in a batter’s 2nd plate appearance than in the batter’s 1st, and less effective still in the batter’s 3rd and subsequent appearances against them in the game. Okay, so that’s true IN GENERAL.

But the BITGODs aren’t suggesting human punching bags like the Orioles’ hapless Radicalhams Liz pitch complete games and go every fourth start; how about the great (mostly Hall of Fame) workhorses Jenkins cites? And how about the contemporary marquee starters he hopes will be the vanguard of the Counter-Reformation leading us all back to the Chef’s Salad Days of Infinite Virtue and World Series Hardware? We’ll take a quick look at Lincecum and his cohort, too, in Part II.

SOME FACTS: PITCHERS CITED AS HEROIC
For the purpose of addressing the injury issues a little, let's look at the career trajectories of starters called out in a sidebar by Jenkins as heroic performers.

Jenkins writes:

In 1904, a 30-year-old Yankees pitcher named Jack Chesbro led the American League with 48 complete games.

Let’s take a quick look at Happy Jack Chesbro’s career.

  
  Year Ag Tm  Lg  W   L  G   GS CG   IP   ERA *lgERA *ERA+
  +--------------+---+---+---+---+---++------+-----+-----+----+ 
  1899 25 PIT NL  6   9  19  17 15  149.0 4.11 3.81    93 
  1900 26 PIT NL 15  13  32  26 20  215.7 3.67 3.62    99 
  1901 27 PIT NL 21  10  36  28 26  287.7 2.38 3.25   137 
  1902 28 PIT NL 28   6  35  33 31  286.3 2.17 2.75   127 
  1903 29 NYY AL 21  15  40  36 33  324.7 2.77 3.11   112 
  1904 30 NYY AL 41  12  55  51 48  454.7 1.82 2.70   148 
  1905 31 NYY AL 19  15  41  38 24  303.3 2.20 2.91   133 
  1906 32 NYY AL 23  17  49  42 24  325.0 2.96 2.96   100 
  1907 33 NYY AL 10  10  30  25 17  206.0 2.53 2.79   110 
  1908 34 NYY AL 14  20  45  31 20  288.7 2.93 2.46    84 
  1909 35 TOT AL  0   5  10   5  2   55.7 6.14 2.52    41 
  +--------------+---+---+---+---+---++------+-----+-----+----+ 
  11 Yr WL%.600  198-132 392 332 260 2896  2.68 2.96   110 
  +--------------+---+---+---+---+---++------+-----+-----+----+ 
  162 Game Avg    18-12   36  31  24 272.0 2.68 2.96   110 
  source: Baseball-Reference.Com

Happy Jack was built up as a young pitcher to take increasing workloads, not unlike an early version of the BITGOD Prescription. Then, at age 30, he pitched the 454 innings, going 41-12 for the Highlanders/Yankees. He had an ERA+ of 148 (an ERA 48% better than the league as a whole).

Don’t imagine for a second that even with the Manliness build up, 454 innings and 48 complete games didn’t affect his career. The following year, he had 25% fewer starts (38) and while still an excellent pitcher, his superiority waned to an ERA+ of 133 (an ERA 33% better than the league). The year after that, increased starts, lesser performance yet, and the year after that, at age 33, his last useful year, significantly fewer starts and an ERA+ of 100 (that is, league average). And he became Crappy Jack for the next two years and was out of the majors.

Chesbro is in the Hall of Fame, by the way, and that supports Jenkins a tad, though to be fair, Chesbro’s installation is all about his durability over a medium length career and what he did in that single 1904 season.

It looks like his magnificent record year took a toll and he was never the same pitcher again. We can’t prove it was cause=workload and effect=diminished performance. But since Jenkins cited his heroism and something Lincecum should be admiring, it’s worth looking at what Chesbro’s career trajectory was and asking the question “Is that a good career trajectory for Lincecum?”

Jenkins writes:

In a 16-inning, complete-game win against Baltimore in 1962, Washington's Tom Cheney threw 228 pitches.

Here’s Cheney’s career (nice Jeff Merron narrative on him):

  
  Year Ag Tm Lg   W L  G  GS   IP   ERA *lgERA *ERA+ 
  +--------------+---+----+---+---++------++-----+-----+----+ 
  1957 22 STL NL  0 1   4  3   9.0  5.00 4.00   80 
  1959 24 STL NL  0 1  11  2  11.7  6.94 4.22   61 
  1960 25 PIT NL  2 2  11  8  52.0  3.98 3.76   95 
  1961 26 TOT     1 3  11  7  29.7 10.01 3.90   39 
  1962 27 WSA AL  7 9  37 23 173.3  3.17 4.05  128 
  1963 28 WSA AL  8 9  23 21 136.3  2.71 3.74  138 
  1964 29 WSA AL  1 3  15  6  48.7  3.70 3.70  100 
  1966 31 WSA AL  0 1   3  1   5.3  5.06 3.47   69 
  +--------------+---+---+---+---++------++-----+-----+----+ 
  8 Yr WL% .396 19 29 115 71 466.0  3.77 3.88  103 
  +--------------+---+---+---+---++------++-----+-----+----+ 
  162 Game Avg   6 10  42 25 170.3  3.77 3.88  103 
  source: Baseball-Reference.Com

By the way, after that magnificent September 12 start, he came back six days later and couldn’t get out of the 4th inning, and then rested for twelve days, and pitched a very good start.

In Cheney’s next season, 1963, he started with four complete games, all great performances, on as little as three days rest. And at age 28, that was basically the end of his starting career. The rest of the season he had 17 starts, going 4-9 and performing around the league average. He broke down in August. While his season ERA+ was his best, if you subtract those first four starts, he was average.

And he was never a regular major league starter again.

Cheney (while he will always have a place in my heart for graciously giving me the second ballplayer autograph I ever got, a swell childhood memory) is not a support for the heroic complete game as a career-builder.

Jenkins writes:

-- New York Giants pitcher Joe McGinnity, known as "Iron Man," didn't start pitching in the major leagues until he was 28. Five times, he pitched both ends of a doubleheader. He worked an astounding 434 innings in the 1903 season, and over his 10-year career racked up 247 wins and 314 complete games. Get this, though: Wandering through the minors until he was 52, collected 204 more wins.

And

Nolan Ryan, known as much for his walks as his strikeouts, routinely surpassed 150 pitches as his career progressed (27 years, 222 complete games and 5,386 innings pitched). In 1974, according to beat writers in attendance, Ryan threw 259 pitches in a 12-inning win over Kansas City.

These two Hall of Fame pitchers support The BITGOD Prescription, though it’s also worth noting each was a once-in-a-generation freak of nature. Both pitched forever and had very good ERA+ marks. But note that Iron Joe didn’t have a particularly good major league season after the age of 35 for either quantity or quality. Ryan was very good even though age 44, though his game was predicated more on durability and pure strikeout power than on excellent ERA marks.

By the way, I don’t know if those beat writers Jenkins mentioned were remembering the wrong year, but Ryan didn’t go more than 10 innings in any game against the Royals in 1974. They may have been thinking about a 13 inning game against Boston where Ryan walked ten and struck out 19. An historical note -- my man Cecil Cooper, leading off for the Bosox, struck out six times in his eight at-bats that game against Ryan; what do we call that...a Golden Cust?

As freaks of nature, we can recognize that pair of iron men are figures to be respected and (ideally) emulated. But as freaks of nature, we also need to recognize their success, especially Ryan’s, might be significantly genetic, and not something coaching, nutrition or pilates can reify.

Jenkins writes:

It's remarkable enough that on May 1, 1920, Brooklyn and Boston played a 1-1 tie that lasted 26 innings. Incredibly, pitchers Leon Cadore and Joe Oeschger each went the distance. Historians estimate that Cadore threw 345 pitches, Oeschger 319. MBB NOTE: The game ended as a tie.

Here’s Leon “The Caddy” Cadore’s career:

  
  Year Ag Tm Lg    W  L   G GS CG    IP ERA *lgERA*ERA+ 
  +--------------+---+---+---+---+---+------+-----+-----+----+ 
  1915 24 BRO NL   0  2   7  2  1  21.0 5.57  2.77  50 
  1916 25 BRO NL   0  0   1  0  0   6.0 4.50  2.67  59 
  1917 26 BRO NL  13 13  37 30 21 264.0 2.45  2.78 113 
  1918 27 BRO NL   1  0   2  2  1  17.0 0.53  2.79 527 
  1919 28 BRO NL  14 12  35 27 16 250.7 2.37  2.97 125 
  1920 29 BRO NL  15 14  35 30 16 254.3 2.62  3.23 123 
  1921 30 BRO NL  13 14  35 30 12 211.7 4.17  3.89  93 
  1922 31 BRO NL   8 15  29 21 13 190.3 4.35  4.10  94 
  1923 32 TOT      4  2   9  5  3  38.3 4.46  3.88  87 
  1924 33 NYG NL   0  0   2  0  0   4.0 0.00  3.67  inf 
  +--------------+---+---+---+---+---+------+-----+-----+----+ 
  10 Yr WL% .486  68 72 192 147 83 1257 3.14  3.33 106
  +--------------+---+---+---+---+---+------+-----+-----+----+ 
  162 Game Avg    13 14  38  29 16 252  3.14  3.33 106

We don’t have game-by-game stats for 1920, but if you look at his ERA+ progression, The Caddy looked like a pretty fine young pitcher.

That is, until 1920, the year of his 300+ pitch outing. He never again notched an ERA that equaled or bettered the league average.

Here’s Joe “The Big Ouch” Oeschger’s career:

  
  Year Ag Tm  Lg    W  L  G GS CG  IP     ERA *lgERA *ERA+ 
  +--------------+---+---+---+---+---+------+-----+-----+----+ 
  1914 22 PHI NL    4  8  32 10  5 124.0  3.77  2.92    77 
  1915 23 PHI NL    1  0   6  1  1  23.7  3.42  2.74    80 
  1916 24 PHI NL    1  0  14  0  0  30.3  2.37  2.64   111 
  1917 25 PHI NL   15 14  42 30 18 262.0  2.75  2.81   102 
  1918 26 PHI NL    6 18  30 23 13 184.0  3.03  2.98    98 
  1919 27 TOT NL    4  4  17 12  6 102.7  3.94  2.98    75 
  1920 28 BSN NL   15 13  38 30 20 299.0  3.46  3.04    88 
  1921 29 BSN NL   20 14  46 36 19 299.0  3.52  3.67   104 
  1922 30 BSN NL    6 21  46 23 10 195.7  5.06  4.02    79 
  1923 31 BSN NL    5 15  44 19  6 166.3  5.68  3.99    70 
  1924 32 TOT NL    4  7  29 10  0  94.3  4.01  4.21   105 
  1925 33 BRO NL    1  2  21  3  1  37.0  6.08  4.18    69
  +--------------+---+---+---+---+---+------+-----+-----+----+ 
  12 Yr WL% .414  82 116 365 197 99 1818  3.81  3.36    88 
  +--------------+---+---+---+---+---+------+-----+-----+----+ 
  162 Game Avg     9 14   44  23 11 219.7 3.81  3.36    88

The Big Ouch must have been a pretty decent prospect. The Phils (not a doormat in that time) brought him up at age 22. He was middling in accomplishment up until the legendary 300+ pitch game, had his best season the following year, but was never more than a spot starter after that. I don’t see Oeschger as either supporting or undermining the Jenkins Protocol; he was middlingly useful and inconsistent before his Phyrric Tie and equally so after, the Mike Morgan of the Post-Great War Era.

Jenkins also wrote about Allie Reynolds in the hero sidebar, but while he was a great performer, he wasn’t particularly heroic. In an era where many starters pitched every 4th game, he started less frequently, but was available as an occasional reliever, so he pitched over 220 innings most years. I suspect the copydesk messed up Jenkins' assertion, but I can’t decode what that would be from what’s left.

CONCLUSIONS…AND MORE NEXT TIME
I see no clear pattern within Jenkins’ cited heroes. A couple of freaks of nature (durability outliers), a couple of guys whose careers crapped out pretty soon after their heroics, and one guy who seemed as middling after as before. Jenkins certainly can’t use the data as a convincing support for his Prescription as a standard approach.

Look, what if you were paying Tim Lincecum, and you wanted the maximum value out of him? What if you could have 32 starts limited to 100-105 pitches a year but with an alternative; what if you could have him for 40 starts a year, but there was one in five chance he’d break down as a result and have a Leon Cadore career? Or what if it was a one in ten chance or a one in three chance.

Knowing that Lincecum is an outlier in many respects (his mechanics, his power-to-size ratios, his intellect, his past training), perhaps he’s an outlier in durability, as well. At what odds do you take the chance of having him be the next Kerry Wood or Mark Prior (outliers who didn’t transcend fatigue and whose starting careers imploded like neutron bombs, leaving no survivors to mourn)? Good question, and one I leave to you to answer.

But this just addresses Injury and Moral Rectitude issues. What about the Team Performance issue – how much better off is a team leaving in the great starter?

In the next entry, I’ll answer that by examining each of Jenkins’ cited workhorses, all of whom had lengthy careers that survived or thrived on BITGOD workloads, and compare their performances to the relievers their teams might have deployed in their steads.

Monday, September 08, 2008

Planning the Intentional Walk: Succession Theory Baseball-Style  

I got a few interesting questions on my last entry about the Tampa Bay Rays' manager, Joe Maddon, intentionally walking Josh Hamilton with the bases loaded and a small enough lead that it put the winning run to the plate, a move that goes against a hundred years of standard operating procedure.

One question that first struck me as uninterestingly obvious was actually grist for MBB-thought.

"When an intentional walk is called for and the manager intends to pull the pitcher and replace him, why is it always that the first pitcher throws the four intentional balls, and not the replacement?"

Baseball has some conventional wisdom about this close-to-universal practice, some of which I think is just bullspittin', much of which is solid reasoning, and some of which is a valuable lesson for managers beyond baseball.

Management will frequently replace a failing team member with a new talent at the wrong moment, getting the new person to drag a late or inadequate project over the finish line. This can be even wrong-er when the replacement is a manager.

Let's look at the standard Baseball arguments for letting the pitcher who's getting replaced issue the walk.

#1 - It preserves the effort of the next pitcher, saving pitches for the person who's going to be doing more serious work. #2 - It potentially gives the incoming pitcher a little more time to get warmed up (not universally needed, and sometimes an additional increment of time that's a negative). #3 - It puts the E.R.A. accounting for that runner on the pitcher who got the team into the jam, not the subsequent pitcher, who's trying to rescue the incumbent. #4 - Controlling your best stuff is tough for most pitchers at the easiest of times.
So in trying to work out of a jam, why would we start the new pitcher's effort with a sequence of throwing not-strikes?

The various objectives of the s.o.p. involve the statistical/historical, the building or dimunition of morale, and, most importantly, the implications for execution.

BEYOND BASEBALL
A new client I had known only a little brought me in to help fix a department that was in a long-term morale morass that cascaded into attitudes that undermined performance. When I nosed around a little and started gathering people's stories and tales and myths about the history of the department, I came to suspect that the problem was a pair of really sorry management practices that combined to make a difficult situation almost impossible to get untangled. If they'd followed Baseball's intentional walk model, they'd have had a lot better chance to solve it. The department had had three different managers in the previous 34 months.

I could discern easily from employee evaluations he'd written and meeting minutes that the first of the three was a person who had craft skills in the department's area, but zero management training that took and was emotionally and intellectually disengaged from the requirements of running the department. He pushed the department into an accountability-free zone; people were almost never given feedback about good work and blame was assigned only when things got awful (denial being practised until other departments screamed in pain loudly enough). When the executive team decided to get rid of the failing manager, they hired a big-time consulting practice to audit the department and make staffing recommendations. The result targeted the manager and five of the eleven staff as underperformers who should be replaced. Ironically, the report used as a key element of the evaluation the employee evaluations of the useless manager (how masterful is that? And those guys charge 7x what I do).

Having paid the enormous pelf the big-time consultants charged, the executive team were bound to the results, though they skipped one piece of the execution. One of the targeted staff had been with the company from the beginning and no one wanted to lay her off though they were committed to laying her off. And they didn't want to lay off the others, leave her in place, only to have to lay her off later (creating multiple "layoffs"). So they made a sub-optimal decision to lay off the manager, recruit another and let the new manager execute the staff layoffs.

Manager number two lasted about eight months, and I'm pretty confident she would have worked out if not saddled with the staff layoffs. She hesitated to execute them until she'd had a chance to examine the files and see the individuals in action, and this accountable approach actually worked against her, because the executive team didn't want to relent a micron on the pre-fab decision and by the time she delivered the pink slips, the deed was associated with her. The remaining staff assumed the purge was her design, and did nothing to cover her back. Performance indicators hinted that process ran tighter, but morale just swirled the bowl.

Manager number two lasted a few months, but had little staff support, a reputation as a loser in the overall corporation, and, demoralized herself, left with zero notice (she told them Friday she would not be back Monday - a classic "at will" consequence) when she was recruited by an out of town rival who didn't care about her reputation but did care about her knowledge of my client's inner workings.

Manager number three was shipped in from another country, a star from a related company. And he was a star. But in the six weeks it took to bring him on, he was pre-victimised by a classic big-organization ploy; a peer manager saw the power vacuum as an opportunity to move a troubled worker with management aspirations herself into the demoralized & adrift department. By the manager number three arrived, the toxic ambition monster had asserted her right to boss everyone around and his first ten weeks were spent not addressing morale or performance issues, but cutting through the unofficial wall of silence about how Toxie was on his roster (the manager who done the dump was well-established with a lot of loyalty from peers and the executive team). Then he had to corral and then cut Toxie out of the herd. I could give you the whole Trail of Tears, but you can see how it inevitably all went downhill and why.

If you've been in the working world for a few years, you have seen some or all of these pieces on display yourself. Planning for succession (and success) is much weaker in the business and academic and non-profit arenas than in Baseball.

Baseball knows you don't dig a hole for the successor if you can avoid it. You do everything reasonable to make sure the successor doesn't carry apparent responsibility for the incumbent's jams. In Baseball, they try to give you adequate time to suss out the situation and get warmed up for it.

Why can't non-Baseball management emulate the National Pastime's s.o.p and let the incumbent issue the intentional walks?

This page is powered by Blogger. Isn't yours?


free website counter