Information about http://retrosheet.org/Research/GehringerC/QualityStarts.pdf
Quality Starts 1957-2006 Christopher Gehringer Notes on the…
Tags: baltimore orioles, bill james baseball, csv format, david w smith, diamond mind, eric milton, espn, gehringer, good measure, interested parties, johan santana, mlb, quality start, retrosheet, rob neyer, starting pitcher, starting pitchers, sunset rd, ups, warm ups,Pages: 13
Language: english
Created: Thu Feb 14 13:47:32 2008
Display cached document
Page 1
Page 2
Page 3
Page 4
Page 5
Page 6
Page 7
Page 8
Page 9
Page 10
Page 11
Page 12
Page 13
Quality Starts 1957-2006
Christopher Gehringer
Notes on the information gained in this study:
The information used here was obtained free of charge from and is copyrighted by Retrosheet.
Interested parties may contact Retrosheet at 20 Sunset Rd., Newark, DE 19711.
*Retrosheet data is incomplete for games from 1957 to 1973. However, the season with the most
games missing (1957) is only missing about 52 games, or just over 100 starts. 1999 is also
missing from the data in its entirety. All statements made regarding "since 1957" are missing
these games, but the general amount of the data within should constitute enough information to
come to the same conclusions. However, any "leader lists" could be effected if this data were
available. For information on the games missing from the report, please reference:
http://retrosheet.org/wanted/index.html
**Games where the starting pitcher failed to get through the first batter (or in a few cases, even
face the first batter) were not made eligible to count as Quality Starts. For example, Johan
Santana pitched what would be a Quality Start on 8/6/2002 against the Baltimore Orioles, but was
not listed as the starter, though listed starter Eric Milton was injured in warm-ups and did not
throw a pitch. This happened 22 times since 1957* as far as I can tell, which wouldn't do much
to skew the data.
Other studies cited:
On Quality Starts and Dominant Starts by "redsoxtalk"
http://fantasyscope.wordpress.com/2007/08/30/on-quality-starts-and-dominant-starts/
The Quality Start is a Useful Statistic by David W. Smith
http://www.diamond-mind.com/articles/qstart.htm
Quality Start Still a Good Measure of Quality by Rob Neyer
http://insider.espn.go.com/mlb/insider/columns/story?columnist=neyer_rob&id=2407313
Bill James Baseball Abstract, 1987
Other notes:
Interested parties can email chrisg at sunflower dot com for more information or to
request a .txt file of linescores for starting pitchers since 1957 in csv format (12.4MB)
The Quality Start was invented in 1985 by Philadelphia Inquirer sportswriter John Lowe.
Quite simply, a pitcher is credited with a Quality Start (QS), if he 1) starts the game as
the pitcher, 2) pitches at least six innings (IP) and 3) gives up three earned runs (ER) or
less. Over the past two decades, the statistic has had its detractors. Sporting News writer
Moss Klein once submitted that in principle, a pitcher could pitch exactly six innings and
give up exactly three earned runs in each and every start, giving him a less-than-
outstanding earned run average (ERA) of 4.50.
Because of this opposition, there have been a few studies to prove or disprove the Quality
Start's validity as a statistic, most notably by David W. Smith and Rob Neyer.
Retrosheet founder David W. Smith in the Spring of 1992 did a study for all starts from
1984 to 1991, noting that the ERA of all Quality Starts during the period was 1.91, which
is more-than-outstanding, given the number of innings figured into the equation. Also, he
states, the average number of innings pitched in these starts was close to 7 ½ as opposed
to the minimum, 6, and also the winning percentage of teams with Quality Starts was
better than two out of three. To refute Klein, Smith noted that less than 1000 of the close
to 17,500 Quality Starts between 1984 and 1991 fit the minimum definition of 6 IP and 3
ER (5.7%).
ESPN writer Rob Neyer also did a study in early 2006, comparing the 1985 season, the
year Lowe made his argument, with the 2005 season. His main contention was that the
teams' winning percentage in Quality Starts in each season, was still over two in three
and actually they were just one-tenth of a percentage point (67.3% to 67.4%) apart He
also showed that in 1985, the QSERA was 1.88, and in 2005 it was 2.04 (both happen to
be over a full point below the team with the lowest ERA in either season).
Both Smith and Neyer agree fourteen years apart that the Quality Start is, and has been a
useful and viable statistic since its inception. Neyer contends we could likely come up
with "a different, `better' measure for a starting pitcher's effectiveness... but...
everybody's got his own definition of `better'" but he also states that "there's something
elegant about 3/6/4.50." It has been shown that the Quality Start has been a good statistic
since at least 1984. The purpose of this study is to discuss 1) If it has always been a
viable statistic, and 2) if there are better criteria to define what is a Quality Start.
Using Retrosheet event files and game data, we can come up with a table showing how
often a team won when its pitcher had a Quality Start since 1957:
Year TW TL QSW%
1957 838 381 68.74%
1958 852 369 69.78%
1959 864 393 68.74%
1960 839 429 66.17%
1961 965 463 67.58%
1962 1095 517 67.93%
1963 1208 604 66.67%
1964 1169 573 67.11%
1965 1157 594 66.08%
1966 1137 573 66.49%
1967 1200 625 65.75%
1968 1277 722 63.88%
1969 1440 708 67.04%
1970 1375 691 66.55%
1971 1448 760 65.58%
1972 1459 759 65.78%
1973 1388 676 67.25%
1974 1462 696 67.75%
1975 1391 666 67.62%
1976 1435 737 66.07%
1977 1438 647 68.97%
1978 1559 779 66.68%
1979 1435 709 66.93%
1980 1492 691 68.35%
1981 982 509 65.86%
1982 1423 701 67.00%
1983 1452 683 68.01%
1984 1464 716 67.16%
1985 1453 702 67.42%
1986 1451 684 67.96%
1987 1380 594 69.91%
1988 1575 794 66.48%
1989 1502 731 67.26%
1990 1483 717 67.41%
1991 1467 746 66.29%
1992 1500 758 66.43%
1993 1542 743 67.48%
1994 1078 487 68.88%
1995 1283 585 68.68%
1996 1449 627 69.80%
1997 1511 729 67.46%
1998 1637 752 68.52%
2000 1547 703 68.76%
2001 1617 725 69.04%
2002 1596 778 67.23%
2003 1651 727 69.43%
2004 1566 711 68.77%
2005 1651 796 67.47%
2006 1566 713 68.71%
Over the course of each of the above 49 seasons, teams won, at a minimum, 63.88% of
their pitchers' Quality Starts. What's interesting about what is above is that 1968, the
"year of the pitcher" had the lowest QSW%, which, at first glance, seems a bit strange.
But the fact of the matter is that 1968 was such a good pitching season, that in 48% of the
games with a Quality Start, both pitchers accomplished the feat, making it by far the
season with the highest percentage of games with both starters doing so. Below are the
ten seasons with the highest percentage:
Year TW TL QSW% #QS #G #both #1 %both
pit
1968 1277 722 63.88 1999 1351 648 703 48.0%
1972 1459 759 65.78 2218 1531 687 844 44.9%
1971 1448 760 65.58 2208 1532 676 856 44.1%
1963 1208 604 66.67 1812 1265 547 718 43.2%
1976 1435 737 66.07 2172 1522 650 872 42.7%
1967 1200 625 65.75 1825 1284 541 743 42.1%
1978 1559 779 66.68 2338 1646 692 954 42.0%
1965 1157 594 66.08 1751 1238 513 725 41.4%
1960 839 429 66.17 1268 897 371 526 41.4%
1988 1575 794 66.48 2369 1677 692 985 41.3%
And the ten with the lowest percentage...
Year TW TL QSW% #QS #G #both #1 %both
pit
2004 1566 711 68.77 2277 1729 548 1181 31.7%
1987 1380 594 69.91 1974 1487 487 1000 32.8%
1996 1449 627 69.80 2076 1561 515 1046 33.0%
2006 1566 713 68.71 2279 1713 566 1147 33.0%
2003 1651 727 69.43 2378 1785 593 1192 33.2%
2002 1596 778 67.23 2374 1780 594 1186 33.4%
2001 1617 725 69.04 2342 1743 599 1144 34.4%
1995 1283 585 68.68 1868 1388 480 908 34.6%
1998 1637 752 68.52 2389 1773 616 1157 34.7%
1977 1438 647 68.97 2085 1547 538 1009 34.8%
Though 1968 had the lowest QSW%, it is mainly because the pitching was so strong that
season that the winning and losing team quite routinely had each pitcher throw a Quality
Start. It is also interesting to note that save for 1987 (a year in which Wade Boggs hit 24
home runs--one of exactly two seasons he hit more than ten, and a well-known offensive
year), the other eight seasons with the lowest percentage of both starters throwing a
Quality Start happened after 1995.
Another fact about 1968 is it has the most Quality Starts per game (62.7%), so it seems
that one of the reasons the QSW% may go down is because of the percentage of games
where both pitchers throw a Quality Start goes up. In 1968, there were 1594 games
looked at, and 648 of them had both pitchers with a Quality Start (40.65%), meaning that
over 20.3% of Quality Starts inherently had to earn the pitchers' team a loss. It is the
only season since 1957 that is over 40% in that department.
The above observations back up Smith and Neyer, that the Quality Start is a good stat.
They also show that it has been a good statistic since at least 1957, as in no season has the
QSW% gone below 63.88% or above 69.91%. Since 1957, the QSW% for major league
baseball is approximately 67.5%. (66749-32173 for the games studied).
To look at Neyer's question, however, is it a great statistic? Yes, 3/6/4.50 is elegant, but
what is better? Most everyone would agree that if their pitcher throws a Quality Start, he
gives his team a good chance of winning the game. Most everyone would agree that two
out of three is a good chance. In today's 162 game schedule, it would mean going 108-
54 (actually, 67.5% would be somewhere between 109 and 110 wins). In virtually any
season, a team with this winning percentage would be in the postseason.
But is there a such thing as a High-Quality Start? Is there a way to determine that a team
should win the game if the pitcher accomplishes a certain feat? Secondly, what would
that winning percentage have to be? 3 in 4? 4 in 5? Also, is there a such thing as a Low-
Quality Start? What would its percentage be? 3 in 5?
Since a Quality Start ends up being a victory for the team about two out of every three
times, or about 67.5%, I would suggest that a high Quality Start would be where a pitcher
gave his team a chance to win about 10% more of the time than that. I would propose a
low-Quality Start would be where a pitcher gave his team a chance to win about 10% less
than that as well.
The true "elegance" of the Quality Start is that, since 1957, it has occurred in 51.6% of
starts, or right around half. This shows not just the simplicity of the six inning, three
earned run formula, but also the criteria, or better, happen just over half the time. Every
game has to have a winner and a loser, so having this as the criteria makes perfect sense.
A High-Quality Start should occur more than a Quality Start, and a Low-Quality Start
should occur less, however, this makes for some troubling information on the low end of
the spectrum. For instance, if a pitcher goes at least five innings and gives up four earned
runs or less, his team is likely to win 61.4% of the time. Also, since 1957, using these
criteria, no season has seen the winning percentage drop below 59.1% (again 1968), or
climb above 62.9% (2003), mainly because of all the Quality Starts figured into the data.
Also, a start with these criteria account for 70% of all starts in the major leagues. It is
likely that both pitchers accomplish the same feat in the game, moving the winning
percentage closer to 50%, but it is hard to find a line to draw in the sand where a team
wins under 60 percent of the time and still tag the word "quality" to the criteria. I have
include the data below, but I also see no validity in a "Low-Quality Start."
On the other hand, pitchers going seven innings and giving up two earned runs or less
end up pitching their team to victory in over 76% of games since 1957, and they only
account for just over 38% of starts in that span. The low for one season (again in 1968--
because of the same issues discussed above) was 72.76%, and the high (1958) was
79.55%. Below are the two tables. The first is with the minimum of five innings and
four runs. The second is with the minimum of seven innings and two runs:
Bill James wanted to call the Quality Start a "Johnson Game." His reasoning was that
people had a problem with the term "Quality." It makes sense to rename the statistic, but
the statistic that would be renamed may not be the most sensible. The problem being is
that if a pitcher hit the minimum requirement of six innings and three earned runs exactly,
his team lost the game more often then they won it over the past 50 years or so. Actually
the record is 2302 wins versus 3295 losses (41%), which might seem to back up Moss
Klein's argument.
However, it is important to keep in mind that this is a minimum requirement, and that the
purpose of the Quality Start was to say that a team with one had a reasonable chance of
winning the game, and it does just that. The problem with any statistic of this nature is
that you have to have minimums. Any better minimums would raise the winning
percentage to a range where the statistic would not be used for its specific purpose. Thus,
even with the low winning percentage for the minimum requirements, the Quality Start is
a good statistic, and as shown above, it is a more reliable statistic now than before.
Even though the winning percentage for the minimum requirements has gone up
significantly since 1957, the winning percentage for anyone fitting the requirements
hasn't really changed all that much. It is still very close to 2/3. Actually, QSW% for
1957 and 2006 differ by only .03%, making them each other's most comparable seasons,
and in 1994 (55-50) and again in 2000 (98-95), teams with the minimum requirement for
a Quality Start actually won over 50% of their games. In the last ten years, the record is
838-936 (47%). It will be interesting to see what the next ten years brings, as more and
more teams go to the bullpen and starting pitchers get even less innings. It looks as if
people reaching the minimum will soon have a winning record, if not be right around
50% for the foreseeable future:
Winning % in QS With Minimum Requirements
60.0%
50.0%
40.0%
Win %
30.0%
20.0%
10.0%
0.0%
57
60
63
66
69
72
75
78
81
84
87
90
93
96
00
03
06
19
19
19
19
19
19
19
19
19
19
19
19
19
19
20
20
20
Year
By definition, The Quality Start is about giving your team a chance to win, and not about
stating your team should win. Thus, I propose the "High-Quality Start" with the
minimums of seven innings and two runs. Seven innings and two runs gives a team odds
of winning of better than three in four, and the odds of winning with the minimum
requirements throughout the past 50 years is about 54.5%, stating the pitcher's team
should win the game, and that is something worthy of being titled a "Johnson Game."
The question here is did Walter Johnson fans feel their team had a reasonable chance
with him on the hill, or did they feel like they were going to win?
Win % Win % Win %
19
57.00
58.00
59.00
60.00
61.00
62.00
63.00
64.00
68.00
70.00
72.00
74.00
76.00
78.00
80.00
82.00
60.00
61.00
62.00
63.00
64.00
65.00
66.00
67.00
68.00
69.00
70.00
71.00
57
1957 1957 19
59
1959 1959 19
61
1961 1961 19
63
1963 1963 19
65
1965 1965 19
67
1967 1967 19
69
1969 1969 19
71
1971 1971 19
73
1973 1973 19
75
1975 1975 19
77
1977 1977 19
79
1979 1979 19
81
1981 1981 19
Quality Starts
Year
Year
Year
83
7 Innings/2 Runs
5 Innings/4 Runs
1983 1983 19
85
1985 1985 19
87
1987 1987 19
89
1989 1989 19
91
1991 1991 19
93
1993 1993 19
95
1995 1995 19
97
1997 1997 20
00
2000 2000 20
02
2002 2002 20
04
2004 2004 20
06
2006 2006
Lists:
Most Quality Starts since 1957:
Name TW TL
Don Sutton 336 147
Nolan Ryan 331 150
Tom Seaver 309 145
Gaylord Perry 312 141
Steve Carlton 311 136
Roger Clemens 331 108
Phil Niekro 298 139
Tommy John 303 128
Bert Blyleven 291 138
Greg Maddux 304 122
Tom Glavine 282 107
Jim Kaat 256 109
Frank Tanana 240 125
Ferguson 254 104
Jenkins
Best QSW% since 1957 (minimum 100 QS):
Name TW TL QSW%
Sandy Koufax 160 39 0.804
Mark Mulder 93 23 0.8017
Scott McGregor 124 31 0.8
Whitey Ford 151 40 0.7906
Jeff Fassero 83 22 0.7905
Andy Pettite 150 40 0.7895
Billy Pierce 86 23 0.789
Storm Davis 93 25 0.7881
Ron Guidry 153 42 0.7846
Dennis Leonard 132 37 0.7811
Steve Stone 103 29 0.7803
Kirk Reuter 113 32 0.7793
Most High-Quality Starts since 1957:
Name TW TL HQS
Don Sutton 258 73 331
Roger Clemens 257 65 322
Gaylord Perry 243 78 321
Nolan Ryan 249 70 319
Steve Carlton 249 69 318
Tom Seaver 237 81 318
Phil Niekro 226 69 295
Greg Maddux 219 69 288
Tommy John 219 67 286
Bert Blyleven 215 60 275
Jim Palmer 211 51 262
Ferguson 201 58 259
Jenkins
Bob Gibson 199 46 245
Jim Kaat 179 53 232
Frank Tanana 168 64 232
Best HQSW% since 1957 (minimum 100 HQS):
Name TW TL HQSW%
Mike Flanagan 120 18 86.96%
Denny McClain 100 15 86.96%
Sandy Koufax 132 20 86.84%
Andy Pettite 93 16 85.32%
David Wells 120 21 85.11%
Warren Spahn 108 20 84.38%
Mike Cuellar 138 26 84.15%
Ross Grimsley 89 17 83.96%
Scott McGregor 89 17 83.96%
Steve Renko 86 17 83.50%
Whitey Ford 116 23 83.45%
Frank Viola 126 25 83.44%
Dennis Leonard 100 20 83.33%
Most Low Quality Starts since 1957:
Name TW TL QS
Don Sutton 382 217 599
Nolan Ryan 369 227 596
Steve Carlton 366 198 564
Phil Niekro 350 204 554
Tom Seaver 353 195 548
Greg Maddux 362 179 541
Gaylord Perry 338 202 540
Team records in Quality Starts, Non Quality Starts and Overall since 1957:
Other observations:
In reading about a proposed "Dominant Start," it seems to have merit. However, since
1957, the difference in strikeouts per inning in a Quality Start as opposed strikeouts per
inning in a non-Quality Start only differ by two-hundredths of a strikeout, it might be
viable, but it would not really be related to the Quality Start. Walks per inning, however,
go up 0.18 per inning if the pitcher does not have a quality outing. This holds true by
season, or for all the data as a whole.
There could be a statistic based on the High-Quality Start, but more research is likely
needed on the subject.
Bert Blyleven should be in the Hall of Fame, and I hate the Twins.
No team historically has won over 40% of their games where they didn't have a quality
start. The best is Anaheim at 37.8% (Though the NQSW% as the California Angels is
one of the worst--Los Angeles Angels is pretty average) The Yankees are second in
QSW% to the Milwaukee Braves and in NQSW% to the Angels.
Further study:
How often did it happen that a player had a Quality Start after six innings but stayed in
the game and lost it? What are the win loss records for those players? Is there a certain
team, player or manager that has that happen more than others? How much does good
hitting play into it? It seems that QSW% and NQSW% are pretty standard across the
board, but the better hitting teams seem to be a little higher.