Projected Strike Probability

#############################################################################################
# Compare the projected strike zone probabilities for each of the four versions of movement #
#############################################################################################

#############
# LIBRARIES #
#############

library(RODBC);

####################
# ADDITIONAL FILES #
####################

source("Movement Scripts/Mvt_Fcns.r")

########
# MAIN #
########

# Player's name
first <- 'Julio';
last <- 'Urias';

# Year
year <- '2016';

# Pitch type
pitch_type <- 'CU';

# Batters' handedness
stand <- 'R';

# Set up the database name
database <- paste("MLB",year,sep="");

# Set the pitches that resulted in a swing
swing_cases <- "(des = 'In play, no out' OR
                 des = 'In play, out(s)' OR
                 des = 'In play, run(s)' OR
                 des = 'Foul' OR
                 des = 'Foul (Runner Going)' OR
                 des = 'Foul Bunt' OR
                 des = 'Foul Tip' OR
                 des = 'Swinging Strike' OR
                 des = 'Swinging Pitchout' OR
                 des = 'Missed Bunt' OR
                 des = 'Swinging Strike (Blocked)')";

# Set the pitches that were taken
take_cases <- "(des = 'Called Strike' OR
                des = 'Ball' OR
                des = 'Ball in Dirt' OR
                des = 'Hit By Pitch')";

# Open the channel
channel <- odbcConnect(database);

# Set the query for pitches swung at
Swing_Query = paste("SELECT ax, ay, az, vx0, vy0, vz0, x0, y0, z0
                    FROM pitches WHERE pitch_type = '",pitch_type,"' AND
                    ",swing_cases," AND ab_id IN
                    (SELECT ab_id FROM atbats WHERE
                    pitcher = (SELECT eliasid FROM players WHERE first = '",
                    first,"' AND last = '",last,"') AND stand = '",stand,"')",sep="");

# Download the swing data
Swing <- sqlQuery(channel,Swing_Query);

# Close the channel
close(channel);

# Set the number of points in y at which to sample
Ny <- 100;

# Track the probability of the extrapolation being a strike
Sw_Pitch_Prb_PFX <- Pitch_Track_PFX(Swing,Ny,stand);
Sw_Pitch_Prb_PFX_Drag <- Pitch_Track_PFX_Drag(Swing,Ny,stand);
Sw_Pitch_Prb_W <- Pitch_Track_W(Swing,Ny,stand);
Sw_Pitch_Prb_W_Drag <- Pitch_Track_W_Drag(Swing,Ny,stand);

# Average the values for each array
Np <- length(Swing$x0);

# Plot the average probability by distance for swings
y_inc <- seq(17/12,55,length=Ny);
plot(0,0,,xlim=c(17/12,55),ylim=c(0,1),xaxs='i',yaxs='i',xlab="Distance from Home Plate (feet)", ylab="Called Strike Probability", main=paste(first," ",last," (",year,"): Swings on ",pitch_type," v. ",stand,"HB",sep=""));
lines(y_inc,colMeans(Sw_Pitch_Prb_PFX),col="red",pch=20,xlim=c(17/12,55),ylim=c(0,1),xaxs='i',yaxs='i',xaxt='n',yaxt='n',xlab='',ylab='');
lines(y_inc,colMeans(Sw_Pitch_Prb_PFX_Drag),col="blue",pch=20,xlim=c(17/12,55),ylim=c(0,1),xaxs='i',yaxs='i',xaxt='n',yaxt='n',xlab='',ylab='');
lines(y_inc,colMeans(Sw_Pitch_Prb_W),col="green",pch=20,xlim=c(17/12,55),ylim=c(0,1),xaxs='i',yaxs='i',xaxt='n',yaxt='n',xlab='',ylab='');
lines(y_inc,colMeans(Sw_Pitch_Prb_W_Drag),col="black",pch=20,xlim=c(17/12,55),ylim=c(0,1),xaxs='i',yaxs='i',xaxt='n',yaxt='n',xlab='',ylab='');
legend(2,0.205,c('PFX','PFX D','W','W D'),lty=c(1,1,1,1),lwd=c(1,1,1,1),col=c('red','blue','green','black'));

# Open the channel
channel <- odbcConnect(database);

# Set the query for pitches taken
Take_Query = paste("SELECT ax, ay, az, vx0, vy0, vz0, x0, y0, z0
                   FROM pitches WHERE pitch_type = '",pitch_type,"' AND
                   ",take_cases," AND ab_id IN
                   (SELECT ab_id FROM atbats WHERE
                   pitcher = (SELECT eliasid FROM players WHERE first = '",
                   first,"' AND last = '",last,"') AND stand = '",stand,"')",sep="");

# Download the pitches taken data
Take <- sqlQuery(channel,Take_Query);

# Close the channel
close(channel);

# Track the probability of the extrapolation being a strike
Tk_Pitch_Prb_PFX <- Pitch_Track_PFX(Take,Ny,stand);
Tk_Pitch_Prb_PFX_Drag <- Pitch_Track_PFX_Drag(Take,Ny,stand);
Tk_Pitch_Prb_W <- Pitch_Track_W(Take,Ny,stand);
Tk_Pitch_Prb_W_Drag <- Pitch_Track_W_Drag(Take,Ny,stand);

windows();

# Plot the average probability by distance for pitches taken
plot(0,0,,xlim=c(17/12,55),ylim=c(0,1),xaxs='i',yaxs='i',xlab="Distance from Home Plate (feet)", ylab="Called Strike Probability", main=paste(first," ",last," (",year,"): Takes on ",pitch_type," v. ",stand,"HB",sep=""));
lines(y_inc,colMeans(Tk_Pitch_Prb_PFX),col="red",pch=20,xlim=c(17/12,55),ylim=c(0,1),xaxs='i',yaxs='i',xaxt='n',yaxt='n',xlab='',ylab='');
lines(y_inc,colMeans(Tk_Pitch_Prb_PFX_Drag),col="blue",pch=20,xlim=c(17/12,55),ylim=c(0,1),xaxs='i',yaxs='i',xaxt='n',yaxt='n',xlab='',ylab='');
lines(y_inc,colMeans(Tk_Pitch_Prb_W),col="green",pch=20,xlim=c(17/12,55),ylim=c(0,1),xaxs='i',yaxs='i',xaxt='n',yaxt='n',xlab='',ylab='');
lines(y_inc,colMeans(Tk_Pitch_Prb_W_Drag),col="black",pch=20,xlim=c(17/12,55),ylim=c(0,1),xaxs='i',yaxs='i',xaxt='n',yaxt='n',xlab='',ylab='');
legend(2,0.205,c('PFX','PFX D','W','W D'),lty=c(1,1,1,1),lwd=c(1,1,1,1),col=c('red','blue','green','black'));

# Merge the two data sets
All_Pitch_Prb_PFX <- rbind(Sw_Pitch_Prb_PFX,Tk_Pitch_Prb_PFX);
All_Pitch_Prb_PFX_Drag <- rbind(Sw_Pitch_Prb_PFX_Drag,Tk_Pitch_Prb_PFX_Drag);
All_Pitch_Prb_W <- rbind(Sw_Pitch_Prb_W,Tk_Pitch_Prb_W);
All_Pitch_Prb_W_Drag <- rbind(Sw_Pitch_Prb_W_Drag,Tk_Pitch_Prb_W_Drag);
All <- list(ax=c(Swing$ax,Take$ax), ay=c(Swing$ay,Take$ay), az=c(Swing$az,Take$az), vx0=c(Swing$vx0,Take$vx0), vy0=c(Swing$vy0,Take$vy0), vz0=c(Swing$vz0,Take$vz0), x0=c(Swing$x0,Take$x0), y0=c(Swing$y0,Take$y0),
            z0=c(Swing$z0,Take$z0));

# Determine the indices related to balls and strikes based on probabilities
Index <- ball_strike_prob(All,stand);

# Set the ball and strike data
St_Pitch_Prb_PFX <- All_Pitch_Prb_PFX[Index$S,];
St_Pitch_Prb_PFX_Drag <- All_Pitch_Prb_PFX_Drag[Index$S,];
St_Pitch_Prb_W <- All_Pitch_Prb_W[Index$S,];
St_Pitch_Prb_W_Drag <- All_Pitch_Prb_W_Drag[Index$S,];
Ba_Pitch_Prb_PFX <- All_Pitch_Prb_PFX[Index$B,];
Ba_Pitch_Prb_PFX_Drag <- All_Pitch_Prb_PFX_Drag[Index$B,];
Ba_Pitch_Prb_W <- All_Pitch_Prb_W[Index$B,];
Ba_Pitch_Prb_W_Drag <- All_Pitch_Prb_W_Drag[Index$B,];

windows();

# Plot the average probability by distance for strikes
plot(0,0,,xlim=c(17/12,55),ylim=c(0,1),xaxs='i',yaxs='i',xlab="Distance from Home Plate (feet)", ylab="Called Strike Probability", main=paste(first," ",last," (",year,"): Strikes on ",pitch_type," v. ",stand,"HB",sep=""));
lines(y_inc,colMeans(St_Pitch_Prb_PFX),col="red",pch=20,xlim=c(17/12,55),ylim=c(0,1),xaxs='i',yaxs='i',xaxt='n',yaxt='n',xlab='',ylab='');
lines(y_inc,colMeans(St_Pitch_Prb_PFX_Drag),col="blue",pch=20,xlim=c(17/12,55),ylim=c(0,1),xaxs='i',yaxs='i',xaxt='n',yaxt='n',xlab='',ylab='');
lines(y_inc,colMeans(St_Pitch_Prb_W),col="green",pch=20,xlim=c(17/12,55),ylim=c(0,1),xaxs='i',yaxs='i',xaxt='n',yaxt='n',xlab='',ylab='');
lines(y_inc,colMeans(St_Pitch_Prb_W_Drag),col="black",pch=20,xlim=c(17/12,55),ylim=c(0,1),xaxs='i',yaxs='i',xaxt='n',yaxt='n',xlab='',ylab='');
legend(2,0.205,c('PFX','PFX D','W','W D'),lty=c(1,1,1,1),lwd=c(1,1,1,1),col=c('red','blue','green','black'));

windows();

# Plot the average probability by distance for balls
plot(0,0,,xlim=c(17/12,55),ylim=c(0,1),xaxs='i',yaxs='i',xlab="Distance from Home Plate (feet)", ylab="Called Strike Probability", main=paste(first," ",last," (",year,"): Balls on ",pitch_type," v. ",stand,"HB",sep=""));
lines(y_inc,colMeans(Ba_Pitch_Prb_PFX),col="red",pch=20,xlim=c(17/12,55),ylim=c(0,1),xaxs='i',yaxs='i',xaxt='n',yaxt='n',xlab='',ylab='');
lines(y_inc,colMeans(Ba_Pitch_Prb_PFX_Drag),col="blue",pch=20,xlim=c(17/12,55),ylim=c(0,1),xaxs='i',yaxs='i',xaxt='n',yaxt='n',xlab='',ylab='');
lines(y_inc,colMeans(Ba_Pitch_Prb_W),col="green",pch=20,xlim=c(17/12,55),ylim=c(0,1),xaxs='i',yaxs='i',xaxt='n',yaxt='n',xlab='',ylab='');
lines(y_inc,colMeans(Ba_Pitch_Prb_W_Drag),col="black",pch=20,xlim=c(17/12,55),ylim=c(0,1),xaxs='i',yaxs='i',xaxt='n',yaxt='n',xlab='',ylab='');
legend(2,0.985,c('PFX','PFX D','W','W D'),lty=c(1,1,1,1),lwd=c(1,1,1,1),col=c('red','blue','green','black'));