Baseball Articles | Index


By David W. Smith
November, 1991

It's become fashionable in baseball circles in recent years to talk about the significance of the "big bang". Popularized by sports writer Tom Boswell -- with a big assist from Earl Weaver, who is credited with building his Orioles around the three-run homer -- the big bang theory is now used to explain both the origin of the universe as well as the decisive moment in most baseball games.

By definition, a big bang occurs in a baseball game whenever the winning team scores more runs in a single inning than the losing team does in the entire game. (We'll leave cosmogony to the physicists and astronomers). Thus, for example, all shutouts are big bangs. The analyses in this paper are based on the play by play data of Project Scoresheet and The Baseball Workshop.

League summaries for the 1991 season and for 1984-1991 inclusive (16822 total games) are presented in Table 1.


Table 1. Big Bangs and Shutouts

Including			Excluding

Shutouts			Shutouts

1991 AL   530/1134 = 46.7% of games are big bangs 380/984 = 38.6%

1991 NL   453/ 970 = 46.7% of games are big bangs 331/848 = 39.0%

84-91 AL 4186/9062 = 46.2% big bangs 3139/8015 = 39.2%

84-91 NL 3617/7760 = 46.6% big bangs 2591/6734 = 38.5%


The present study was prompted by a desire to test the assertion by some sportscasters (Harry Kalas of the Phillies for one) that the big bang is a common event and therefore important in understanding the playing of the game. On a superficial level, such an assertion is correct -- nearly 50% of all games are indeed big bangs. However, when we try to decide what the significance of that 50% figure is, it becomes evident that the chance of winning with a big bang is much more directly related to the ability of a team's defense to hold the opponents to a small number of runs than it is to the team's offense. Note that "team defense" should be seen as the combined ability of pitchers and fielders, but will be simplified here to refer to just pitchers.

There is an unfortunate tendency in baseball analysis to see a correlation such as this and immediately jump to a conclusion about cause and effect. However, direct evidence to support a cherished hypothesis is usually harder to come by. In the present instance we can express the dilemma in the form of a question: "Does the scoring of many runs in a game lead to a greater tendency to have big innings, or does the greater likelihood of a big inning automatically mean that the team will score more runs in a game?"

It is interesting that the percentage of big bangs in the two leagues (Table 1) is so similar, given the common perception of the AL as a hitter's league and the NL as a pitcher's league. These similarities lead us to consider what the big bang is supposed to represent. Since it is usually seen as a sign of the "big inning", the numbers should be examined to see if there is any merit in what we might call the "Earl Weaver method".

To that end it is interesting to note that the very large majority of big bangs occur when the losing team scores 2 runs or fewer, as shown in Table 2.

Table 2. Percentage of Big Bangs in which Losers Score 0, 1, or 2 Runs

          ----------- AL ------------    ------------ NL ------------

            Big                  % of       Big                  % of

Year      Bangs    0    1    2  Bangs     Bangs    0    1    2  Bangs

1991        530  150  152  119   79.4       479  122  165  101   81.0

1984-1991  4186 1047 1345  950   79.8      3617 1026 1179  812   83.4

Again we see similar percentages over the entire 8-year period, with the NL having a slightly higher frequency of occurrence. This overwhelming occurrence of big bangs in games where the losers score fewer than three runs leads to the conclusion that the big bang is not really a measure of a big offense, but an incidental consequence of a well-pitched game. Add the general perception that the NL is a pitcher's league (whatever that means) and the conclusion is even stronger.

As noted at the start, all shutouts are big bangs. Table 3 gives the chance of a big bang for a team which allows 1 run, 2 runs, or more.

Table 3. Chance of a Big Bang as a Function of Runs Allowed


             Allowing 1 run   Allowing 2 runs   Allowing >2 runs

                  Big               Big                Big

Year        Gms Bangs     %   Gms Bangs     %   Gms  Bangs     %

1991        202   152  75.2   229   119  52.0   553    109  19.7

1984-1991  1735  1345  77.5  1794   950  53.0  4486    844  18.8


             Allowing 1 run   Allowing 2 runs   Allowing >2 runs

                  Big               Big               Big

Year        Gms Bangs     %   Gms Bangs     %   Gms Bangs      %

1991        217   165  76.0   214   101  47.2   417    65   15.6

1984-1991  1583  1179  74.5  1709   812  47.5  3442   600   17.4

Some interesting differences between the two leagues emerge here as well, as the percentages in the last two categories are consistently lower in the NL. This difference presumably reflects the overall lower scoring in the NL than in the AL (8.9 runs by both teams per game in the AL in 1991, 8.2 in NL).

Since it is clear that most big bangs occur when the losers score fewer than 3 runs, it is useful to consider the general chance of winning in all games where a team allows fewer than 3 runs, whether it is a big bang or not. Table 4 has the appropriate numbers.

Table 4. Won-Loss Record when Allowing < 3 Runs

                   AL                 NL

Year         Win Loss   Pct    Win  Loss   Pct

1991         581   98  .856    553   102  .844

1984-1991   4576  701  .867   4318   783  .847

Again we have the conclusion that the big bang isn't really measuring a big offensive performance. Big bangs predominantly occur when the winner doesn't allow many runs.

Where does the appeal of the big bang come from? Certainly a big outburst of scoring in an inning is dramatic, and a large lead may be good for the manager's digestion. Nevertheless, the scoring of, say 6 runs, in an inning during a game in which the opponents are being held to two runs is hardly a meaningful indication of the value of an overpowering offense. Let's focus on the offensive side in isolation for a moment by considering how often the "big inning" actually occurs.

Table 5. Number and % of Innings with Different Numbers of Runs


Year      Innings  0 Runs          1 Run         2 Runs        > 3 Runs

1991        20440   14882  (72.8)   2994  (14.6)   1369  (6.7)   499  (2.4)

1984-1991  162312  117625  (72.5)  24610  (15.2)  11293  (7.0)  3985  (2.5)


Year      Innings  0 Runs           1 Run        2 Runs        > 3 Runs

1991        17441   12857  (73.7)    2592 (14.9)   1140 (6.5)    344  (1.9)

1984-1991  139991  103257  (73.8)   20885 (14.9)   9073 (6.5)   2849  (2.0)

Once again the NL shows slightly lower scoring than the AL, but the patterns are similar between the two leagues and demonstrate that the "big inning" is a very uncommon event. Returning to the definition I offered of a big inning as one in which 4 or more runs are scored, then it seems unreasonable to base strategy on an event that occurs about 2% of the time.

We certainly know that different team have different strategies, depending on their personnel and their home parks. Therefore it is reasonable to expect a significant variation between different teams in the ability to put a big inning on the board. Table 6 presents results on a team by team basis for 1991, with Texas, Milwaukee, and Detroit leading the way, while hapless Montreal trails badly.

Table 6. Pct of Innings in which each Team Scored 4 or More Runs in 1991

Texas Rangers            3.3

Milwaukee Brewers        3.1

Detroit Tigers           3.0

Chicago White Sox        2.9

Kansas City Royals       2.8

Pittsburgh Pirates       2.8

Oakland Athletics        2.6

Minnesota Twins          2.5

California Angels        2.4

New York Mets            2.2

Boston Red Sox           2.2

Atlanta Braves           2.2

San Diego Padres         2.0

St.Louis Cardinals       2.0

Seattle Mariners         1.9

Philadelphia Phillies    1.9

Los Angeles Dodgers      1.9

Baltimore Orioles        1.9

Houston Astros           1.8

Cincinnati Reds          1.8

San Francisco Giants     1.7

New York Yankees         1.7

Cleveland Indians        1.7

Toronto Blue Jays        1.6

Chicago Cubs             1.5

Montreal Expos           1.0

As with our other scoring measures, we find that the top of this list is dominated by AL teams and the bottom mostly has NL teams. The final table is a list of the number of big bangs achieved and allowed by each team from 1984-1991.

Table 7. Big Bangs Achieved and Allowed by Each Team, 1984-1991

New York Mets           359  240

Toronto Blue Jays       357  234

Los Angeles Dodgers     348  308

Kansas City Royals      332  310

San Diego Padres        329  292

Boston Red Sox          324  269

Houston Astros          318  332

Minnesota Twins         316  318

Oakland Athletics       315  268

St.Louis Cardinals      313  286

California Angels       310  299

Cincinnati Reds         305  309

Montreal Expos          300  287

Detroit Tigers          299  280

New York Yankees        298  292

Chicago White Sox       291  311

San Francisco Giants    290  300

Pittsburgh Pirates      289  299

Milwaukee Brewers       286  309

Baltimore Orioles       275  323

Seattle Mariners        271  328

Chicago Cubs            269  303

Philadelphia Phillies   260  319

Texas Rangers           259  324

Cleveland Indians       253  321

Atlanta Braves          237  342

Most Achieved in One Season

Los Angeles Dodgers    1985   57

Most Allowed in One Season

Baltimore Orioles      1988   60

In closing we should ask: is there any value to the notion of the big bang? It led us to some interesting conclusions about scoring, but it doesn't seem to offer a clearcut indication of which team is going to win. As many analysts have noted over the years, a winning team requires a balance between scoring runs and preventing the opponents from scoring. Simply addressing the offensive half or defensive half of the equation in isolation cannot be expected to give the whole picture.