American vs. National League wOBA Vales.
This article is originally published at https://www.datascienceriot.com/
Traditionally, statistics like wOBA (weighted on-base average) have been calculated using league averages. While building the baseballDBR package, I thought it would be interesting to group the American and National when making wOBA calculations. In theory, there should be parity across the two leagues, but that is not always the case.
In order to calculate wOBA values for each league, the baseballDBR package uses a ported version of Tom Tango’s SQL incantation to calculate wOBA using the Baseball Databank. While Tango admits, this calculation is not perfect, it normally has a plus/minus of less than one one-thousandth of a percent compared to Fangraphs’ values.
Gathering wOBA Modifiers by League
install.packages("baseballDBR")library(baseballDBR)
# Load data from Baseball Databank
get_bbdb(table = c("Batting", "Pitching", "Fielding"))
# Run wOBA values for seperate leagues.
# Subset for 1900, when the AL was created. Exclude the Federal League.
w_vals <- wOBA_values(BattingTable = Batting, FieldingTable = Fielding, PitchingTable = Pitching, Sep.Leagues = TRUE) %>%
subset(yearID >= 1900 & lgID != "FL", select = c("yearID", "lgID", "lg_woba"))
head(w_vals)## # A tibble: 6 x 3
## yearID lgID lg_woba
## <int> <chr> <dbl>
## 1 1900 NL 0.3383157
## 2 1901 AL 0.3281682
## 3 1901 NL 0.3203771
## 4 1902 AL 0.3279358
## 5 1902 NL 0.3085920
## 6 1903 AL 0.3002770Plot Leauge wOBA Vales by Year
library(ggplot2)
ggplot(w_vals, aes(x=yearID, y=lg_woba, color=lgID)) + geom_point(shape=1) +
scale_colour_hue(l=50) + geom_smooth() The plot shows the modern parity we expected. It also shows the effects of the “dead ball era” prior to 1920. However, what is interesting is the increase in. league wOBA between 1920 and 1930 in the American League. It should be mentioned, the stat “league wOBA” is the average on-base percentage (OBP) for each league.
Using OBP to Find Outliers
There were obviously players in the American League between 1920 and 1930 that were blowing the curve, performing above average. We can drill down deeper to find out exactly who those players were. Note, since the league wOBA represents a league average OBP, we will use OBP instead of wOBA to find our outliers.
# Set NA to zero because sacrifice fly was not a tracked stat prior to 1954, but were still counted as at-bats and in the OBP calculation.
Batting$SF[is.na(Batting$SF)] <- 0
# Add OBP to data frame.
Batting$OBP <- OBP(Batting)
# Find the outliers. Subset for more than 200 at-bats to exclude part-time players.
al_outliers <- subset(Batting, OBP >= .5 & lgID == "AL" & yearID >= 1920 & yearID <= 1930 & AB >= 200)
al_outliers## playerID yearID stint teamID lgID G AB R H X2B X3B HR RBI SB CS BB SO IBB HBP SH SF GIDP OBP
## 18520 ruthba01 1920 1 NYA AL 142 457 158 172 36 9 54 137 14 14 150 80 NA 3 5 0 NA 0.533
## 19040 ruthba01 1921 1 NYA AL 152 540 177 204 44 16 59 171 17 13 145 81 NA 4 4 0 NA 0.512
## 20097 ruthba01 1923 1 NYA AL 152 522 151 205 45 13 41 131 17 21 170 93 NA 4 3 0 NA 0.545
## 20637 ruthba01 1924 1 NYA AL 153 529 143 200 39 7 46 121 9 13 142 81 NA 4 6 0 NA 0.513
## 21719 ruthba01 1926 1 NYA AL 152 495 139 184 30 5 47 150 11 9 144 76 NA 3 10 0 NA 0.516There you have it, George Herman Ruth, blowing the curve once again!
Thanks for visiting r-craft.org
This article is originally published at https://www.datascienceriot.com/
Please visit source website for post related comments.
