1919 World
Series
Chicago
White Sox vs. Cincinnati Reds
BACKGROUND
In 1919, the Chicago White Sox were the favorite to beat the Cincinnati Reds in the World Series. Unfortunately, 8 White Sox players accepted bribes from gamblers to throw the World Series. Ultimately, the Reds won the World Series, 5 games to 3.
Retrosheet Entry on 1919 World Series
Wikipedia Entry for the Black Sox Scandal
GOAL
The goal of this analysis is to answer the following questions:
1. Had the 8 White Sox players not been conspiring with gamblers, who would have likely won the 1919 World Series?
2. Did the 8 White Sox players perform significantly below their abilities to indicate they did in fact throw the 1919 World Series?
PROCEDURE
Using a computer baseball simulation/replay model, Diamond Mind Baseball Version 9, I simulated the 1919 World Series results 100 times and recorded the results in the Encyclopedia module. The following was done:
1. Using the 1919 Homebrew season, I stripped out the other 16 teams and reduced the rosters of the White Sox and Reds to 25 players each, based upon who played during the 1919 World Series and other players with significant playing time who were with their respective ball clubs at the conclusion of the season.
2. I created a 9 game schedule for the 1919 World Series on the same dates the actual World Series was played (scheduled for Game 9).
3. I input and saved the actual starting lineups for the first 8 games and used a “best-guess” lineup for Game 9.
4. I verified the 1919 American and National League era data and resaved every player’s batting and pitching event tables. Each player performs in the simulation at their 1919 regular season statistical tendencies, taking into account the effect of the ballpark and league factors.
5. I simulated the World Series until one team had 5 wins and loaded the data into Encyclopedia. I repeated this step 100 times.
This procedure created a database of 100 World Series’ results, creating a population of potential outcomes of the 1919 World Series. The actual 1919 World Series results was then compared to this population of potential outcomes.
CAVEATS
The accuracy of this analysis is only as good as the model’s simulation engine. The Diamond Mind Baseball Version 9 model has achieved much acclaim as being one of the most accurate statistical simulation/replay models available. Unfortunately, its bias is towards working with late-20th century seasons in how pitching staffs are handled and has limitations in simulating seasons from the deadball era (pre-1927).
Also, the accuracy of this analysis is only as good as the ability to replicate the play of the players and teams through a statistical representation of the players’ abilities, the ballpark factors and the managerial tendencies/strategy of the managers. Much of the statistical representations of the players, ballparks and managers are subject to individual interpretation in which another person may model them differently.
A computer model will simulate real-life in a predictable, defined manner and is limited at replicating the human decision. Likewise, players are defined by strict statistical representations whereas in the real-life, players’ abilities are affected by more than the “laws of statistics.”
A player’s 1919 statistics are compiled during the season against players from every team, players of different abilities. In a World Series against the best team in the other league, a player’s statistical outcomes could be significantly different as he’s facing players who are likely better. Therefore, a .300 hitter during the regular season can’t necessarily be expected to perform statistically at a .300 batting average.
And finally, I am not trained as a statistician. This analysis is done from the basis of a statistics class back in college and a life-long love of baseball. It’s best if you view the results for entertainment purposes and not a serious statistical analysis.
SERIES RESULTS
In the 1919 World Series, the Reds defeated the White Sox 5 games to 3. Here are some stats from my 100 simulations of the 1919 World Series.
White Sox Series Wins: 70 (70.0%)
White Sox Win Series in 5 Games: 7 (7.0%)
White Sox Win Series in 6 Games: 9 (9.0%)
White Sox Win Series in 7 Games: 23 (23%)
White Sox Win Series in 8 Games: 15 (15.0%)
White Sox Win Series in 9 Games: 16 (16.0%)
Reds Win Series in 9 Games: 9 (9.0%)
Reds Win Series in 8 Games: 12 (12.0%)
Reds Win Series in 7 Games: 4 (4.0%)
Reds Win Series in 6 Games: 5 (5.0%)
Reds Win Series in 5 Games: 0 (0.0%)
The actual outcome of the Reds winning in 8 games is certainly within reason as that outcome occurred 12.0% of the time in the simulations. But with the White Sox winning the series occurred 70.0% of the time, I conclude that the White Sox were likely the better team in 1919. With an outcome of winning the series 30.0% of the time, the Reds were probably not the greatest underdog in World Series history and certainly teams who were greater underdogs have won the series before (i.e. 1990 Reds beating the Athletics).
But in order to bring the 1919 series into a context of most other World Series, it’s necessary to look at the outcome had the series only been a best-of-7 affair. If the series were a best-of-7 affair, the White Sox would have won 64 times (64.0%). This makes sense because the longer the series, the more difficult it is for an underdog team to win it.
9 Game Series: White Sox win 70 times (70.0%)
7 Game Series: White Sox win 64 times (64.0%)
5 Game Series: White Sox win 63 times (63.0%)
3 Game Series: White Sox win 55 times (55.0%)
GAME RESULTS
Let’s look at each game individually.
Game 1: At
Redland Field in Cincinnati
Actual Result: Cincinnati (Ruether) 9, Chicago (Cicotte) 1 (Reds take a 1-0 Series lead)
Simulation Result: White Sox win 64 out of 100 times (64.0%)
Game 2: At
Redland Field in Cincinnati
Actual Result: Cincinnati (Sallee) 4, Chicago (Williams) 2 (Reds take a 2-0 Series lead)
Simulation Result: White Sox win 52 out of 100 times (52.0%)
Game 3: At
Comiskey Park in Chicago
Actual Result: Chicago (Kerr) 3, Cincinnati (Fisher) 0 (Reds maintain a 2-1 Series lead)
Simulation Result: White Sox win 50 out of 100 times (50.0%)
Game 4: At
Comiskey Park in Chicago
Actual Result: Cincinnati (Ring) 2, Chicago (Cicotte) 0 (Reds take a 3-1 Series lead)
Simulation Result: White Sox win 70 out of 100 times (70.0%)
Game 5: At
Comiskey Park in Chicago
Actual Result: Cincinnati (Eller) 5, Chicago (Williams) 0 (Reds take a 4-1 Series lead)
Simulation Result: White Sox win 58 out of 100 times (58.0%)
Game 6: At
Redland Field in Cincinnati
Actual Result: Chicago (Kerr) 5, Cincinnati (Ruether) 4 (Reds maintain a 4-2 Series lead)
Simulation Result: White Sox win 49 out of 93 times (52.7%)
Game 7: At
Redland Field in Cincinnati
Actual Result: Chicago (Cicotte) 4, Cincinnati (Sallee) 1 (Reds maintain a 4-3 Series lead)
Simulation Result: White Sox win 46 out of 79 times (58.2%)
Game 8: At
Comiskey Park in Chicago
Actual Result: Cincinnati (Eller) 10, Chicago (Williams) 5 (Reds win Series 5-3)
Simulation Result: White Sox win 30 out of 52 times (57.7%)
Game 9: At
Comiskey Park in Chicago
Actual Result: Not Played
Simulation Result: White Sox win 16 out of 25 times (64.0%)
Overall Simulation Result: White Sox win 435 out of 749 times (58.1%)
The White Sox appear to have the edge in all 9 games of the series with Game 3 the game most in the Reds’ favor with a 50-50 outcome. Eddie Cicotte, who won 29 games with a 1.82 ERA during the 1919 season, is a huge factor in the White Sox outcomes as the 3 games he starts give the White Sox 64.0%, 70.0% and 58.2% favorable outcomes. The White Sox win 64.5% of all games he starts but only 54.3% of the other games. Cicotte is a key factor in the White Sox being the favored team in the Series. Unfortunately, he was one of the 8 players involved in the conspiracy.
PLAYER STATISTICS
It is known that the 8 players accepted money from gamblers to throw the 1919 World Series, but the question remains whether each player purposely played worse to throw the Series. One could argue that players like Joe Jackson and Buck Weaver did not compromise their play and were only taking the gamblers’ money in case the White Sox lost the Series.
Let’s take a look at each player’s statistics from the 1919 World Series in relation to their statistics from the 100 simulations. The statistics from the 100 simulations are expressed as an average per Series. The Percentile uses the Excel formula to determine where the actual Series statistic falls within the range of the 100 simulations.
|
Happy
Felsch |
AVG |
OBP |
SPC |
G |
AB |
H |
2B |
3B |
HR |
R |
RBI |
BB |
K |
|
100 Simulations |
.249 |
.304 |
.334 |
7.49 |
29.25 |
7.28 |
0.81 |
0.32 |
0.35 |
3.10 |
4.49 |
2.15 |
1.75 |
|
Actual Series |
.192 |
.222 |
.231 |
8 |
26 |
5 |
1 |
0 |
0 |
2 |
3 |
1 |
4 |
|
Percentile |
29.4% |
21.2% |
26.1% |
|
|
|
|
|
|
|
|
|
|
|
Chick
Gandil |
AVG |
OBP |
SPC |
G |
AB |
H |
2B |
3B |
HR |
R |
RBI |
BB |
K |
|
100 Simulations |
.254 |
.285 |
.282 |
7.49 |
28.96 |
7.36 |
0.50 |
0.03 |
0.08 |
2.31 |
3.02 |
1.14 |
1.34 |
|
Actual Series |
.233 |
.258 |
.300 |
8 |
30 |
7 |
0 |
1 |
0 |
1 |
5 |
1 |
3 |
|
Percentile |
45.0% |
38.3% |
55.5% |
|
|
|
|
|
|
|
|
|
|
|
Joe
Jackson |
AVG |
OBP |
SPC |
G |
AB |
H |
2B |
3B |
HR |
R |
RBI |
BB |
K |
|
100 Simulations |
.306 |
.407 |
.402 |
7.49 |
27.80 |
8.52 |
0.71 |
0.40 |
0.38 |
3.98 |
4.55 |
4.80 |
0.60 |
|
Actual Series |
.375 |
.394 |
.563 |
8 |
32 |
12 |
3 |
0 |
1 |
5 |
6 |
1 |
2 |
|
Percentile |
74.7% |
41.3% |
89.9% |
|
|
|
|
|
|
|
|
|
|
|
Fred
McMullin |
AVG |
OBP |
SPC |
G |
AB |
H |
2B |
3B |
HR |
R |
RBI |
BB |
K |
|
100 Simulations |
.263 |
.330 |
.290 |
3.28 |
3.00 |
0.79 |
0.03 |
0.01 |
0.01 |
0.25 |
0.40 |
0.24 |
0.30 |
|
Actual Series |
.500 |
.500 |
.500 |
2 |
2 |
1 |
0 |
0 |
0 |
0 |
0 |
0 |
0 |
|
Percentile |
72.7% |
67.6% |
71.7% |
|
|
|
|
|
|
|
|
|
|
|
Swede
Risberg |
AVG |
OBP |
SPC |
G |
AB |
H |
2B |
3B |
HR |
R |
RBI |
BB |
K |
|
100 Simulations |
.210 |
.263 |
.238 |
7.49 |
24.91 |
5.23 |
0.38 |
0.03 |
0.09 |
2.16 |
2.13 |
1.79 |
2.40 |
|
Actual Series |
.080 |
.233 |
.160 |
8 |
25 |
2 |
0 |
1 |
0 |
3 |
0 |
5 |
3 |
|
Percentile |
12.1% |
36.6% |
26.4% |
|
|
|
|
|
|
|
|
|
|
|
Buck
Weaver |
AVG |
OBP |
SPC |
G |
AB |
H |
2B |
3B |
HR |
R |
RBI |
BB |
K |
|
100 Simulations |
.259 |
.282 |
.303 |
7.49 |
31.80 |
8.25 |
0.76 |
0.08 |
0.16 |
3.62 |
3.04 |
0.88 |
0.69 |
|
Actual Series |
.324 |
.324 |
.500 |
8 |
34 |
11 |
4 |
1 |
0 |
4 |
0 |
0 |
2 |
|
Percentile |
83.8% |
75.6% |
95.9% |
|
|
|
|
|
|
|
|
|
|
|
Eddie
Cicotte |
ERA |
WHIP |
W |
L |
S |
G |
GS |
CG |
SHO |
INN |
H |
ER |
BB |
K |
|
100 Simulations |
1.95 |
1.07 |
1.51 |
0.60 |
0.00 |
2.82 |
2.79 |
1.19 |
0.49 |
21.7 |
18.27 |
4.71 |
4.90 |
6.03 |
|
Actual Series |
2.91 |
1.11 |
1 |
2 |
0 |
3 |
3 |
2 |
0 |
21.7 |
19 |
7 |
5 |
7 |
|
Percentile |
22.4% |
41.8% |
|
|
|
|
|
|
|
|
|
|
|
|
|
Lefty
Williams |
ERA |
WHIP |
W |
L |
S |
G |
GS |
CG |
SHO |
INN |
H |
ER |
BB |
K |
|
100 Simulations |
2.32 |
1.16 |
1.15 |
0.81 |
0.00 |
2.52 |
2.52 |
0.95 |
0.32 |
18.8 |
17.13 |
4.84 |
4.66 |
7.34 |
|
Actual Series |
6.61 |
1.22 |
0.00 |
3.00 |
0 |
3 |
3 |
0 |
0 |
16.3 |
12 |
12 |
8 |
4 |
|
Percentile |
1.0% |
42.9% |
|
|
|
|
|
|
|
|
|
|
|
|
There are 3 categories of players here. First, there are 2 players (Jackson & Weaver) who exceeded the average statistics for the 100 simulations. Second, there are 5 players (Felsch, Gandil, Risberg, Cicotte and Williams) who played below the average statistics for the 100 simulations. And third, there’s McMullin, who was a bench player who’s playing time likely had no effect on the Series results.
Please keep in mind that except for Williams’ ERA, the players all performed in the 1919 Series within the range of outcomes determined through the 100 simulations. Except for Williams’ ERA, each player’s Series statistics, taken outside of any evidence of a conspiracy, are all plausible.
Let’s take a look at the players’ statistics grouped together.
|
All 6
Batters |
AVG |
OBP |
SPC |
G |
AB |
H |
2B |
3B |
HR |
R |
RBI |
BB |
K |
|
100 Simulations |
.257 |
.309 |
.313 |
7.49 |
145.7 |
37.43 |
3.19 |
0.87 |
1.07 |
15.42 |
17.63 |
11.00 |
7.08 |
|
Actual Series |
.255 |
.293 |
.369 |
8 |
149 |
38 |
8 |
3 |
1 |
15 |
14 |
8 |
14 |
|
Percentile |
49.9% |
37.6% |
80.9% |
|
|
|
|
|
|
|
|
|
|
|
Both
Pitchers |
ERA |
WHIP |
W |
L |
S |
G |
GS |
CG |
SHO |
INN |
H |
ER |
BB |
K |
|
100 Simulations |
2.12 |
1.11 |
2.66 |
1.41 |
0.00 |
5.34 |
5.31 |
2.14 |
0.81 |
40.5 |
35.40 |
9.55 |
9.56 |
13.37 |
|
Actual Series |
4.50 |
1.16 |
1 |
5 |
0 |
6 |
6 |
2 |
0 |
38.0 |
31 |
19 |
13 |
11 |
|
Percentile |
2.4% |
41.7% |
|
|
|
|
|
|
|
|
|
|
|
|
When looking at the statistics collectively, it appears like the batters’ Series statistics fall near the average for the 100 simulations. Possibly, the 6 position players were able to play at normal abilities during most of the time and only choose to hit into an out only during critical junctures in the game. Or possibly players like Jackson and Weaver were able to counter the poor statistics of Felsch, Gandil and Risberg.
But looking at the pitchers’ Series statistics, it’s clearly obvious that the Cicotte and Williams were able to extract a much greater pull on the outcomes of the game. Most notably is their ERA falling almost outside the 100 simulation range and their 1-5 W-L record being significantly worse than their 2.7-1.4 W-L record in the 100 simulations. In fact, in only 1 of the 100 simulations did Cicotte and Williams team up for 5 Series losses and only twice teaming up for 4 Series losses. In only 16 simulations did they team up for only 0 or 1 win.
DATABASE
Here is the 1919 Diamond Mind Baseball database I used for this analysis.