app.R

library(shiny)
library(shinydashboard)
library(DT)
library(dplyr)
library(ggplot2)
library(tidyr)
library(stringr)
library(scales)
library(maps)
library(mapproj)
library(arrow)
library(readr)
library(xgboost)

# Load data
TM25 <- read_parquet("TM25.parquet")
ccbl_data <- read_parquet("ccbl_data.parquet")

# Function to calculate Stuff+ using your models
calculate_stuff_plus <- function(data) {
  
  # Initialize Stuff+ column first
  data$StuffPlus <- NA
  
  # Check if required columns exist (updated list)
  required_cols <- c("RelSpeed", "SpinRate", "RelHeight", "RelSide", "Extension", 
                     "InducedVertBreak", "HorzBreak", "PitcherThrows")
  missing_cols <- required_cols[!required_cols %in% names(data)]
  
  if(length(missing_cols) > 0) {
    warning("Missing required columns for Stuff+ calculation: ", paste(missing_cols, collapse = ", "))
    return(data)
  }
  
  # Check if models are loaded and xgboost is available
  if(!exists("RHPFBStuff") || !exists("LHPFBStuff") || !exists("RHPBBStuff") || !exists("LHPBBStuff")) {
    warning("Stuff+ models not found")
    return(data)
  }
  
  if(!requireNamespace("xgboost", quietly = TRUE)) {
    warning("xgboost package not available for Stuff+ calculation")
    return(data)
  }
  
  tryCatch({
    # Create missing columns that the models expect
    data$SeamShiftedWakeEffect <- ifelse(is.na(data$SeamShiftedWakeEffect), 0, data$SeamShiftedWakeEffect)
    data$UniquenessScore <- ifelse(is.na(data$UniquenessScore), 0, data$UniquenessScore)
    
    # Define the exact feature columns expected by the models
    model_features <- c("RelSpeed", "SpinRate", "RelHeight", "RelSide", 
                        "Extension", "InducedVertBreak", "HorzBreak", 
                        "SeamShiftedWakeEffect", "UniquenessScore")
    
    # Create prediction matrix for fastballs
    fb_data <- data %>%
      filter(TaggedPitchType %in% c("Fastball", "FourSeamFastBall", "FourSeamFastB", "Sinker", "TwoSeamFastBall")) %>%
      filter(!is.na(RelSpeed), !is.na(SpinRate), !is.na(RelHeight), !is.na(RelSide), 
             !is.na(Extension), !is.na(InducedVertBreak), !is.na(HorzBreak), !is.na(PitcherThrows))
    
    # Create prediction matrix for breaking balls
    bb_data <- data %>%
      filter(TaggedPitchType %in% c("Slider", "Curveball", "Changeup", "ChangeUp", "Sweeper", "Cutter", "Splitter")) %>%
      filter(!is.na(RelSpeed), !is.na(SpinRate), !is.na(RelHeight), !is.na(RelSide), 
             !is.na(Extension), !is.na(InducedVertBreak), !is.na(HorzBreak), !is.na(PitcherThrows))
    
    # Calculate Fastball Stuff+ for RHP
    if(nrow(fb_data %>% filter(PitcherThrows == "Right")) > 0) {
      rhp_fb <- fb_data %>% filter(PitcherThrows == "Right")
      
      # Create matrix with all model features
      rhp_fb_matrix <- as.matrix(rhp_fb[, model_features])
      
      # Apply the same preprocessing as used in model training (center and scale)
      # Get preprocessing info from the model if available
      if(!is.null(RHPFBStuff$preProcess)) {
        rhp_fb_matrix_processed <- predict(RHPFBStuff$preProcess, rhp_fb_matrix)
      } else {
        # Manual centering and scaling if preProcess object not available
        rhp_fb_matrix_processed <- scale(rhp_fb_matrix, center = TRUE, scale = TRUE)
      }
      
      # Convert to xgb.DMatrix
      rhp_fb_dmatrix <- xgboost::xgb.DMatrix(data = rhp_fb_matrix_processed)
      
      # Predict using the model directly (not finalModel)
      if(!is.null(RHPFBStuff$finalModel)) {
        rhp_fb_pred <- predict(RHPFBStuff$finalModel, rhp_fb_dmatrix)
      } else {
        rhp_fb_pred <- predict(RHPFBStuff, rhp_fb_dmatrix)
      }
      
      # Convert to Stuff+ scale (models predict delta_RE, so negative values are better)
      rhp_fb_stuffplus <- 100 + (rhp_fb_pred * -20)
      
      # Find indices in original data
      rhp_fb_indices <- which(data$TaggedPitchType %in% c("Fastball", "FourSeamFastBall", "FourSeamFastB", "Sinker", "TwoSeamFastBall") & 
                                data$PitcherThrows == "Right" &
                                !is.na(data$RelSpeed) & !is.na(data$SpinRate) & !is.na(data$RelHeight) & 
                                !is.na(data$RelSide) & !is.na(data$Extension) & !is.na(data$InducedVertBreak) & 
                                !is.na(data$HorzBreak) & !is.na(data$PitcherThrows))
      
      if(length(rhp_fb_stuffplus) == length(rhp_fb_indices)) {
        data$StuffPlus[rhp_fb_indices] <- rhp_fb_stuffplus
      }
    }
    
    # Calculate Fastball Stuff+ for LHP
    if(nrow(fb_data %>% filter(PitcherThrows == "Left")) > 0) {
      lhp_fb <- fb_data %>% filter(PitcherThrows == "Left")
      
      lhp_fb_matrix <- as.matrix(lhp_fb[, model_features])
      
      if(!is.null(LHPFBStuff$preProcess)) {
        lhp_fb_matrix_processed <- predict(LHPFBStuff$preProcess, lhp_fb_matrix)
      } else {
        lhp_fb_matrix_processed <- scale(lhp_fb_matrix, center = TRUE, scale = TRUE)
      }
      
      lhp_fb_dmatrix <- xgboost::xgb.DMatrix(data = lhp_fb_matrix_processed)
      
      if(!is.null(LHPFBStuff$finalModel)) {
        lhp_fb_pred <- predict(LHPFBStuff$finalModel, lhp_fb_dmatrix)
      } else {
        lhp_fb_pred <- predict(LHPFBStuff, lhp_fb_dmatrix)
      }
      
      lhp_fb_stuffplus <- 100 + (lhp_fb_pred * -20)
      
      lhp_fb_indices <- which(data$TaggedPitchType %in% c("Fastball", "FourSeamFastBall", "FourSeamFastB", "Sinker", "TwoSeamFastBall") & 
                                data$PitcherThrows == "Left" &
                                !is.na(data$RelSpeed) & !is.na(data$SpinRate) & !is.na(data$RelHeight) & 
                                !is.na(data$RelSide) & !is.na(data$Extension) & !is.na(data$InducedVertBreak) & 
                                !is.na(data$HorzBreak) & !is.na(data$PitcherThrows))
      
      if(length(lhp_fb_stuffplus) == length(lhp_fb_indices)) {
        data$StuffPlus[lhp_fb_indices] <- lhp_fb_stuffplus
      }
    }
    
    # Calculate Breaking Ball Stuff+ for RHP
    if(nrow(bb_data %>% filter(PitcherThrows == "Right")) > 0) {
      rhp_bb <- bb_data %>% filter(PitcherThrows == "Right")
      
      rhp_bb_matrix <- as.matrix(rhp_bb[, model_features])
      
      if(!is.null(RHPBBStuff$preProcess)) {
        rhp_bb_matrix_processed <- predict(RHPBBStuff$preProcess, rhp_bb_matrix)
      } else {
        rhp_bb_matrix_processed <- scale(rhp_bb_matrix, center = TRUE, scale = TRUE)
      }
      
      rhp_bb_dmatrix <- xgboost::xgb.DMatrix(data = rhp_bb_matrix_processed)
      
      if(!is.null(RHPBBStuff$finalModel)) {
        rhp_bb_pred <- predict(RHPBBStuff$finalModel, rhp_bb_dmatrix)
      } else {
        rhp_bb_pred <- predict(RHPBBStuff, rhp_bb_dmatrix)
      }
      
      rhp_bb_stuffplus <- 100 + (rhp_bb_pred * -20)
      
      rhp_bb_indices <- which(data$TaggedPitchType %in% c("Slider", "Curveball", "Changeup", "ChangeUp", "Sweeper", "Cutter", "Splitter") & 
                                data$PitcherThrows == "Right" &
                                !is.na(data$RelSpeed) & !is.na(data$SpinRate) & !is.na(data$RelHeight) & 
                                !is.na(data$RelSide) & !is.na(data$Extension) & !is.na(data$InducedVertBreak) & 
                                !is.na(data$HorzBreak) & !is.na(data$PitcherThrows))
      
      if(length(rhp_bb_stuffplus) == length(rhp_bb_indices)) {
        data$StuffPlus[rhp_bb_indices] <- rhp_bb_stuffplus
      }
    }
    
    # Calculate Breaking Ball Stuff+ for LHP
    if(nrow(bb_data %>% filter(PitcherThrows == "Left")) > 0) {
      lhp_bb <- bb_data %>% filter(PitcherThrows == "Left")
      
      lhp_bb_matrix <- as.matrix(lhp_bb[, model_features])
      
      if(!is.null(LHPBBStuff$preProcess)) {
        lhp_bb_matrix_processed <- predict(LHPBBStuff$preProcess, lhp_bb_matrix)
      } else {
        lhp_bb_matrix_processed <- scale(lhp_bb_matrix, center = TRUE, scale = TRUE)
      }
      
      lhp_bb_dmatrix <- xgboost::xgb.DMatrix(data = lhp_bb_matrix_processed)
      
      if(!is.null(LHPBBStuff$finalModel)) {
        lhp_bb_pred <- predict(LHPBBStuff$finalModel, lhp_bb_dmatrix)
      } else {
        lhp_bb_pred <- predict(LHPBBStuff, lhp_bb_dmatrix)
      }
      
      lhp_bb_stuffplus <- 100 + (lhp_bb_pred * -20)
      
      lhp_bb_indices <- which(data$TaggedPitchType %in% c("Slider", "Curveball", "Changeup", "ChangeUp", "Sweeper", "Cutter", "Splitter") & 
                                data$PitcherThrows == "Left" &
                                !is.na(data$RelSpeed) & !is.na(data$SpinRate) & !is.na(data$RelHeight) & 
                                !is.na(data$RelSide) & !is.na(data$Extension) & !is.na(data$InducedVertBreak) & 
                                !is.na(data$HorzBreak) & !is.na(data$PitcherThrows))
      
      if(length(lhp_bb_stuffplus) == length(lhp_bb_indices)) {
        data$StuffPlus[lhp_bb_indices] <- lhp_bb_stuffplus
      }
    }
    
  }, error = function(e) {
    warning("Error in Stuff+ calculation: ", e$message)
    # Print more detailed error info for debugging
    cat("Detailed error in Stuff+ calculation:\n")
    cat("Error message:", e$message, "\n")
    if(exists("rhp_fb_matrix")) {
      cat("RHP FB matrix dimensions:", dim(rhp_fb_matrix), "\n")
      cat("RHP FB matrix colnames:", colnames(rhp_fb_matrix), "\n")
    }
  })
  
  return(data)
}
# Function to calculate run values (from your TM25 processing code)
calculate_run_values <- function(data) {
  
  # Add count state columns
  data$begin_Strike <- data$Strikes
  data$begin_Ball <- data$Balls
  data$begin_Outs <- data$Outs
  data$end_Strike <- data$Strikes
  data$end_Ball <- data$Balls
  data$end_Outs <- data$Outs
  
  # Update end states based on pitch outcomes
  for (i in 1:nrow(data)) {
    if (!is.na(data$OutsOnPlay[i]) && data$OutsOnPlay[i] == 1) {
      data$end_Outs[i] <- data$begin_Outs[i] + 1
    }
    
    if (data$PitchCall[i] == "StrikeCalled") {
      data$end_Strike[i] <- data$end_Strike[i] + 1
    } else if (data$PitchCall[i] %in% c("BallCalled", "BallinDirt", "BallIntentional")) {
      data$end_Ball[i] <- data$end_Ball[i] + 1
    } else if (data$PitchCall[i] == "BattersInterference") {
      data$end_Ball[i] <- 0
      data$end_Strike[i] <- 0
    } else if (data$PitchCall[i] %in% c("FoulBall", "FoulBallFieldable", "FoulBallNotFieldable")) {
      if (data$begin_Strike[i] < 2) {
        data$end_Strike[i] <- data$end_Strike[i] + 1
      }
    } else if (data$PitchCall[i] %in% c("HitByPitch", "InPlay")) {
      data$end_Strike[i] <- 0
      data$end_Ball[i] <- 0
    } else if (data$PitchCall[i] == "StrikeSwinging") {
      data$end_Strike[i] <- data$end_Strike[i] + 1
    }
    
    if (data$end_Strike[i] == 3) {
      data$end_Strike[i] <- 0
      data$end_Ball[i] <- 0
      data$end_Outs[i] <- data$end_Outs[i] + 1
    }
    
    if (data$end_Ball[i] == 4) {
      data$end_Ball[i] <- 0
      data$end_Strike[i] <- 0
    }
  }
  
  # Create run expectancy table
  simple_RE <- data %>% 
    group_by(GameID, Inning, Top.Bottom) %>% 
    mutate(
      cumulative_runs = cumsum(RunsScored),
      total_runs = max(cumulative_runs),
      runs_remaining = total_runs - cumulative_runs
    ) %>% 
    ungroup() %>% 
    filter(begin_Ball < 4, begin_Strike < 3, begin_Outs < 3) %>% 
    group_by(begin_Ball, begin_Strike, begin_Outs) %>% 
    summarise(
      run_expectancy = round(mean(runs_remaining, na.rm = TRUE), 3),
      .groups = 'drop'
    )
  
  # Join beginning state RE
  data <- left_join(data, simple_RE, 
                    by = c("begin_Strike", "begin_Ball", "begin_Outs"))
  
  # Rename for end state join
  simple_RE_end <- simple_RE %>% 
    rename(
      end_Strike = begin_Strike,
      end_Ball = begin_Ball,
      end_Outs = begin_Outs,
      end_run_expectancy = run_expectancy
    )
  
  # Join end state RE
  data <- left_join(data, simple_RE_end, 
                    by = c("end_Strike", "end_Ball", "end_Outs"))
  
  # Handle end of inning (3 outs = 0 run expectancy)
  data$end_run_expectancy <- ifelse(data$end_Outs >= 3, 0, data$end_run_expectancy)
  
  # Calculate delta run expectancy
  data$delta_RE <- data$end_run_expectancy - data$run_expectancy + data$RunsScored
  
  return(data)
}

TMcolors <- c(
  "Fastball" = "dodgerblue2",
  "FourSeamFastBall" = "dodgerblue2", 
  "FourSeamFastB" = "dodgerblue2",
  "OneSeamFastBall" = "yellow",
  "TwoSeamFastBall" = "yellow",
  "Sinker" = "yellow",
  "Cutter" = "orange",
  "Changeup" = "#01FF70",
  "ChangeUp" = "#01FF70",
  "Slider" = "tomato",
  "Sweeper" = "gold3",
  "Curveball" = "#B10DC9",
  "Splitter" = "#01FF70",
  "Knuckleball" = "lightblue",
  "Other" = "gray60"
)

ui <- dashboardPage(
  dashboardHeader(title = "NCAA Pitcher Dashboard"),
  
  dashboardSidebar(
    sidebarMenu(
      menuItem("Percentiles", tabName = "pitcher", icon = icon("baseball-ball")),
      menuItem("Pitch Movement and Arsenal", tabName = "movement", icon = icon("chart-line")),
      menuItem("Seam Orientation", tabName = "seam", icon = icon("circle-dot")),
      menuItem("Location and Usage", tabName = "location", icon = icon("crosshairs"))
    )
  ),
  
  dashboardBody(
    tabItems(
      # Original Pitcher Analysis Tab
      tabItem(tabName = "pitcher",
              fluidRow(
                box(
                  title = "Controls", status = "primary", solidHeader = TRUE, width = 12, color = "teal",
                  fluidRow(
                    column(4,
                           selectInput("league", "Select League:",
                                       choices = c("Cape Cod League" = "ccbl", "NCAA 2025" = "ncaa2025", "NCAA 2024" = "ncaa2024"),
                                       selected = "ccbl"
                           )
                    ),
                    column(4,
                           selectInput("pitcher", "Select Pitcher:",
                                       choices = NULL,
                                       selected = NULL
                           )
                    ),
                    column(4,
                           br(),
                           actionButton("update", "Update Analysis", class = "btn-primary")
                    )
                  )
                )
              ),
              
              fluidRow(
                box(
                  title = "Pitcher Percentiles", status = "success", solidHeader = TRUE, width = 12,
                  plotOutput("percentile_plot", height = "550px")
                )
              ),
              
              fluidRow(
                box(
                  title = "Pitcher Statistics", status = "info", solidHeader = TRUE, width = 12,
                  DT::dataTableOutput("pitcher_stats")
                )
              )
      ),
      
      # Enhanced Pitch Movement Tab
      tabItem(tabName = "movement",
              fluidRow(
                box(
                  title = "Movement Controls", status = "primary", solidHeader = TRUE, width = 12, color = "teal",
                  fluidRow(
                    column(3,
                           selectInput("movement_league", "Select League:",
                                       choices = c("Cape Cod League" = "ccbl", "NCAA 2025" = "ncaa2025", "NCAA 2024" = "ncaa2024"),
                                       selected = "ccbl"
                           )
                    ),
                    column(3,
                           selectInput("movement_pitcher", "Select Pitcher:",
                                       choices = NULL,
                                       selected = NULL
                           )
                    ),
                    column(3,
                           selectInput("plot_type", "Plot Type:",
                                       choices = c("Movement Profile" = "movement", "Pitch Breaks" = "breaks"),
                                       selected = "movement"
                           )
                    ),
                    column(3,
                           br(),
                           actionButton("update_movement", "Update Movement", class = "btn-primary")
                    )
                  )
                )
              ),
              
              fluidRow(
                box(
                  title = "Pitch Visualization", status = "success", solidHeader = TRUE, width = 12, color = "teal",
                  plotOutput("movement_plot", height = "700px")
                )
              ),
              
              fluidRow(
                box(
                  title = "Release Points", status = "info", solidHeader = TRUE, width = 6, color = "teal",
                  plotOutput("release_plot", height = "550px")
                )
              ),
              
              fluidRow(
                box(
                  title = "Pitch Arsenal Summary", status = "info", solidHeader = TRUE, width = 12, color = "teal",
                  DT::dataTableOutput("movement_stats")
                )
              )
      ),
      
      # Seam Orientation Tab
      tabItem(tabName = "seam",
              fluidRow(
                box(
                  title = "Seam Orientation Controls", status = "primary", solidHeader = TRUE, width = 12, color = "teal",
                  fluidRow(
                    column(3,
                           h5("Data Source: Cape Cod League Only"),
                           p("Seam orientation analysis requires 3D spin axis data only available in CCBL dataset.", 
                             style = "color: #666; font-style: italic;")
                    ),
                    column(3,
                           selectInput("seam_pitcher", "Select Pitcher:",
                                       choices = NULL,
                                       selected = NULL
                           )
                    ),
                    column(3,
                           checkboxInput("show_seam_curve", "Show Seam Curve", value = TRUE),
                           numericInput("point_size", "Point Size:", value = 3, min = 0.5, max = 3, step = 0.1)
                    ),
                    column(3,
                           numericInput("point_alpha", "Point Transparency:", value = 1.5, min = 0.1, max = 2, step = 0.05),
                           selectInput("filter_pitch_type", "Filter by Pitch Type:",
                                       choices = c("All Pitches" = "all"),
                                       selected = "all"),
                           br(),
                           actionButton("update_seam", "Update Chart", class = "btn-primary")
                    )
                  )
                )
              ),
              
              fluidRow(
                box(
                  title = "Seam Orientation Chart", status = "success", solidHeader = TRUE, width = 12, color = "teal",
                  plotOutput("seam_plot", height = "700px")
                )
              ),
              
              fluidRow(
                box(
                  title = "Seam Orientation Metrics", status = "info", solidHeader = TRUE, width = 12, color = "teal",
                  DT::dataTableOutput("seam_metrics_table")
                )
              )
      ),
      
      # Location and Usage Tab (Fixed)
      tabItem(tabName = "location",
              fluidRow(
                box(
                  title = "Chart Controls", status = "primary", solidHeader = TRUE, width = 12, color = "teal",
                  fluidRow(
                    column(6, 
                           selectInput("location_league", "Select League:",
                                       choices = c("Cape Cod League" = "ccbl", "NCAA 2025" = "ncaa2025", "NCAA 2024" = "ncaa2024"),
                                       selected = "ccbl")
                    ),
                    column(6, 
                           selectInput("location_pitcher", "Select Pitcher:", choices = NULL)
                    )
                  )
                )
              ),
              fluidRow(
                box(title = "Pitch Location Heatmaps", width = 6, status = "info", solidHeader = TRUE,
                    plotOutput("location_plot", height = "450px")),
                box(title = "Pitch Usage by Count", width = 6, status = "warning", solidHeader = TRUE,
                    plotOutput("count_usage_plot", height = "450px"))
              ),
              fluidRow(
                box(
                  title = "Location Summary Statistics", status = "success", solidHeader = TRUE, width = 12, color = "teal",
                  DT::dataTableOutput("location_stats")
                )
              )
      )
    )
  )
)

# Define server logic
server <- function(input, output, session) {
  
  # Data preprocessing function - UPDATED to include Stuff+ calculation
  preprocess_data <- function(data, league_type) {
    if (league_type == "ccbl") {
      # Cape Cod League preprocessing
      data <- data %>%
        filter(!TaggedPitchType %in% c("Knuckleball", "Undefined", "Other")) %>%
        mutate(
          RelSpeed = as.numeric(RelSpeed),
          SpinRate = as.numeric(SpinRate),
          InducedVertBreak = as.numeric(InducedVertBreak),
          HorzBreak = as.numeric(HorzBreak),
          Extension = as.numeric(Extension),
          ExitSpeed = as.numeric(ExitSpeed),
          Angle = as.numeric(Angle),
          RelSide = as.numeric(RelSide),
          RelHeight = as.numeric(RelHeight),
          PlateLocSide = as.numeric(PlateLocSide),
          PlateLocHeight = as.numeric(PlateLocHeight),
          
          is_csw = ifelse(PitchCall %in% c("StrikeSwinging", "StrikeCalled"), 1, 0),
          is_swing = ifelse(PitchCall %in% c("StrikeSwinging", "FoulBallNotFieldable", "InPlay",
                                             "FoulBallFieldable", "FoulBall"), 1, 0),
          is_whiff = ifelse(PitchCall == "StrikeSwinging" & is_swing == 1, 1,
                            ifelse(is_swing == 1, 0, NA)),
          in_zone = case_when(
            PlateLocSide >= -0.95 & PlateLocSide <= 0.95 &
              PlateLocHeight >= 1.6 & PlateLocHeight <= 3.5 ~ 1,
            TRUE ~ 0
          ),
          in_zone_whiff = ifelse(PitchCall == "StrikeSwinging" & in_zone == 1, 1,
                                 ifelse(is_swing == 1 & in_zone == 1, 0, NA)),
          chase = ifelse(is_swing == 1 & in_zone == 0, 1,
                         ifelse(is_swing == 0 & in_zone == 0, 0, NA)),
          is_hit = ifelse(PlayResult %in% c("Single", "Double", "Triple", "HomeRun"), 1,
                          ifelse(PlayResult %in% c("FieldersChoice", "Error", "Out", "Sacrifice"), 0, NA)),
          is_hard_hit = ifelse(PitchCall == "InPlay" & ExitSpeed >= 95, 1,
                               ifelse(PitchCall == "InPlay" & ExitSpeed < 95, 0, NA)),
          is_barrel = ifelse(!is.na(ExitSpeed) & !is.na(Angle) & PitchCall == "InPlay" &
                               (ExitSpeed * 1.5 - Angle) >= 117 &
                               (ExitSpeed + Angle) >= 124 &
                               Angle <= 50 & ExitSpeed >= 98, 1,
                             ifelse(PitchCall == "InPlay", 0, NA)),
          is_plate_appearance = ifelse(
            PitchCall %in% c("InPlay", "HitByPitch") | KorBB %in% c("Strikeout", "Walk") | 
              PlayResult %in% c("StrikeoutSwinging", "StrikeoutLooking"), 1, 0
          ),
          is_at_bat = case_when(
            PitchCall == "InPlay" & !PlayResult %in% c("StolenBase", "Sacrifice", "CaughtStealing", "Undefined") ~ 1,
            KorBB == "Strikeout" ~ 1,
            PlayResult %in% c("StrikeoutSwinging", "StrikeoutLooking") ~ 1,
            TRUE ~ 0
          ),
          is_walk = case_when(
            is_plate_appearance == 1 & KorBB == "Walk" ~ 1,
            is_plate_appearance == 1 & KorBB != "Walk" ~ 0,
            TRUE ~ NA_real_
          ),
          Pitcher = str_replace_all(Pitcher, "(\\w+), (\\w+)", "\\2 \\1"),
          is_k = case_when(
            is_at_bat == 1 & KorBB == "Strikeout" ~ 1,
            is_at_bat == 1 & KorBB != "Strikeout" ~ 0,
            PlayResult %in% c("StrikeoutSwinging", "StrikeoutLooking") ~ 1,
            TRUE ~ NA_real_
          )
        )
    } else if (league_type %in% c("ncaa2025", "ncaa2024")) {
      # NCAA preprocessing (both 2024 and 2025)
      data <- data %>%
        mutate(
          RelSide = as.numeric(RelSide),
          RelHeight = as.numeric(RelHeight),
          PlateLocSide = as.numeric(PlateLocSide),
          PlateLocHeight = as.numeric(PlateLocHeight),
          is_csw = case_when(
            PitchCall %in% c("StrikeSwinging", "StrikeCalled") ~ 1,
            TRUE ~ 0
          ),
          is_swing = case_when(
            PitchCall %in% c("StrikeSwinging", "FoulBallNotFieldable", "InPlay",
                             "FoulBallFieldable", "FoulBall") ~ 1,
            TRUE ~ 0
          ),
          is_whiff = case_when(
            PitchCall == "StrikeSwinging" & is_swing == 1 ~ 1,
            PitchCall != "StrikeSwinging" & is_swing == 1 ~ 0,
            TRUE ~ NA_real_
          ),
          in_zone = case_when(
            PlateLocSide >= -0.95 & PlateLocSide <= 0.95 &
              PlateLocHeight >= 1.6 & PlateLocHeight <= 3.5 ~ 1,
            TRUE ~ 0
          ),
          in_zone_whiff = case_when(
            PitchCall == "StrikeSwinging" & in_zone == 1 ~ 1,
            is_swing == 1 & in_zone == 1 ~ 0,
            TRUE ~ NA_real_
          ),
          chase = case_when(
            is_swing == 1 & in_zone == 0 ~ 1,
            is_swing == 0 & in_zone == 0 ~ 0,
            TRUE ~ NA_real_
          ),
          is_hit = case_when(
            PlayResult %in% c("Single", "Double", "Triple", "HomeRun") ~ 1,
            PlayResult %in% c("FieldersChoice", "Error", "Out", "Sacrifice") ~ 0,
            TRUE ~ NA_real_
          ),
          is_hard_hit = case_when(
            ExitSpeed >= 95 & PitchCall == "InPlay" ~ 1,
            ExitSpeed < 95 & PitchCall == "InPlay" ~ 0,
            TRUE ~ NA_real_
          ),
          is_barrel = case_when(
            !is.na(ExitSpeed) & !is.na(Angle) & PitchCall == "InPlay" &
              (ExitSpeed * 1.5 - Angle) >= 117 &
              (ExitSpeed + Angle) >= 124 &
              Angle <= 50 & ExitSpeed >= 98 ~ 1,
            is.na(ExitSpeed) | is.na(Angle) ~ NA_real_,
            PitchCall == "InPlay" ~ 0
          ),
          is_plate_appearance = ifelse(
            PitchCall %in% c("InPlay", "HitByPitch") | KorBB %in% c("Strikeout", "Walk") | 
              PlayResult %in% c("StrikeoutSwinging", "StrikeoutLooking"), 1, 0
          ),
          is_at_bat = case_when(
            PitchCall == "InPlay" & !PlayResult %in% c("StolenBase", "Sacrifice", "CaughtStealing", "Undefined") ~ 1,
            KorBB == "Strikeout" ~ 1,
            PlayResult %in% c("StrikeoutSwinging", "StrikeoutLooking") ~ 1,
            TRUE ~ 0
          ),
          is_walk = case_when(
            is_plate_appearance == 1 & KorBB == "Walk" ~ 1,
            is_plate_appearance == 1 & KorBB != "Walk" ~ 0,
            TRUE ~ NA_real_
          ),
          is_k = case_when(
            is_at_bat == 1 & KorBB == "Strikeout" ~ 1,
            is_at_bat == 1 & KorBB != "Strikeout" ~ 0,
            PlayResult %in% c("StrikeoutSwinging", "StrikeoutLooking") ~ 1,
            TRUE ~ NA_real_
          )
        )
    }
    
    # CRITICAL FIX: Calculate Stuff+ for the processed data
    tryCatch({
      data <- calculate_stuff_plus(data)
    }, error = function(e) {
      warning("Error calculating Stuff+: ", e$message)
      # Ensure StuffPlus column exists even if calculation fails
      if(!"StuffPlus" %in% names(data)) {
        data$StuffPlus <- NA
      }
    })
    
    # Ensure StuffPlus column exists regardless
    if(!"StuffPlus" %in% names(data)) {
      data$StuffPlus <- NA
    }
    
    return(data)
  }
  
  # Create summary stats for percentiles - UPDATED to include Stuff+
  create_summary_stats <- function(data) {
    # Check if AutoHitType column exists, if not create a simplified GB% calculation
    if(!"AutoHitType" %in% names(data)) {
      gb_calc <- quote(100 * mean(Angle[PitchCall == "InPlay"] < 10, na.rm = TRUE))
    } else {
      gb_calc <- quote(100 * (sum((AutoHitType == "GroundBall" & PitchCall == "InPlay"), na.rm = TRUE) / 
                                sum(PitchCall == "InPlay", na.rm = TRUE)))
    }
    
    summary_data <- data %>%
      group_by(Pitcher) %>%
      summarise(
        `FB Velo` = mean(RelSpeed[TaggedPitchType == "Fastball"], na.rm = TRUE),
        `FB IVB` = mean(InducedVertBreak[TaggedPitchType == "Fastball"], na.rm = TRUE),
        `Avg EV` = mean(ExitSpeed, na.rm = TRUE),
        `Hard-Hit%` = 100 * mean(is_hard_hit, na.rm = TRUE),
        `GB%` = !!gb_calc,
        `Zone%` = round(mean(in_zone, na.rm = TRUE) * 100, 1),
        `Chase%` = 100 * mean(chase, na.rm = TRUE),
        `Whiff%` = 100 * mean(is_whiff, na.rm = TRUE),
        `Z-Whiff%` = 100 * mean(in_zone_whiff, na.rm = TRUE),
        `K%` = 100 * mean(is_k, na.rm = TRUE),
        `BB%` = 100 * mean(is_walk, na.rm = TRUE),
        Extension = mean(Extension, na.rm = TRUE),
        `Stuff+` = mean(StuffPlus, na.rm = TRUE),  # ADD STUFF+ HERE
        .groups = "drop"
      )
    
    # Filter out rows with all NA values, but be more lenient
    summary_data %>%
      filter(rowSums(is.na(.) | is.infinite(as.matrix(.))) < ncol(.) - 1) # Allow if at least one valid metric
  }
  
  # Location heatmap function
  locations <- function(data, player) {
    
    # Filter and check for required columns
    if(!"PlateLocSide" %in% names(data) || !"PlateLocHeight" %in% names(data)) {
      return(ggplot() + 
               annotate("text", x = 0, y = 0, label = "Missing PlateLocSide or\nPlateLocHeight columns", size = 4) +
               theme_void())
    }
    
    filtered_data <- data %>%
      filter(Pitcher == player, TaggedPitchType != "Other", !is.na(TaggedPitchType)) %>%
      filter(!is.na(PlateLocSide), !is.na(PlateLocHeight), !is.na(BatterSide)) %>%
      filter(is.finite(PlateLocSide), is.finite(PlateLocHeight))
    
    if(nrow(filtered_data) == 0) {
      return(ggplot() + 
               annotate("text", x = 0, y = 0, label = "No location data\navailable for this pitcher", size = 4) +
               theme_void())
    }
    
    # Calculate pitch usage and filter
    pitch_usage_data <- filtered_data %>%
      group_by(TaggedPitchType) %>%
      summarise(pitch_count = n(), .groups = "drop") %>%
      mutate(usage_pct = 100 * pitch_count / nrow(filtered_data)) %>%
      filter(usage_pct >= 2)
    
    if(nrow(pitch_usage_data) == 0) {
      return(ggplot() + 
               annotate("text", x = 0, y = 0, label = "No pitches used ≥2%\nof the time", size = 4) +
               theme_void())
    }
    
    # Filter to qualifying pitch types
    player_data <- filtered_data %>%
      filter(TaggedPitchType %in% pitch_usage_data$TaggedPitchType) %>%
      mutate(
        BatterSide = case_when(
          BatterSide %in% c("Right", "R") ~ "Vs Right",
          BatterSide %in% c("Left", "L") ~ "Vs Left",
          TRUE ~ paste("Vs", BatterSide)
        ),
        TaggedPitchType = factor(TaggedPitchType)
      )
    
    if(nrow(player_data) == 0) {
      return(ggplot() + 
               annotate("text", x = 0, y = 0, label = "No qualifying pitch data\navailable", size = 4) +
               theme_void())
    }
    
    # Check if we have enough data points for density calculation
    min_points_check <- player_data %>%
      group_by(TaggedPitchType, BatterSide) %>%
      summarise(n_points = n(), .groups = "drop") %>%
      filter(n_points >= 3)  # Need at least 3 points for density
    
    if(nrow(min_points_check) == 0) {
      return(ggplot() + 
               annotate("text", x = 0, y = 0, label = "Insufficient data points\nfor density calculation", size = 4) +
               theme_void())
    }
    
    # Create the plot
    p <- player_data %>%
      ggplot(aes(PlateLocSide, PlateLocHeight)) +
      # Only add density if we have sufficient data
      {if(nrow(player_data) >= 10) 
        stat_density_2d(
          aes(fill = after_stat(ndensity)),
          geom = "raster", contour = FALSE, n = 20
        )
      } +
      # Add individual points if density fails or not enough data
      {if(nrow(player_data) < 10)
        geom_point(alpha = 0.6, size = 2)
      } +
      facet_grid(BatterSide ~ TaggedPitchType, labeller = label_value) +
      geom_rect(
        xmin = -8.5/12, xmax = 8.5/12,  # Convert to feet (assuming inches input)
        ymin = 18/12, ymax = 42/12,
        fill = NA, color = "black", linewidth = 0.4
      ) +
      coord_fixed() +
      scale_x_continuous("", limits = c(-2, 2)) +
      scale_y_continuous("", limits = c(0, 5)) +
      theme_void() +
      {if(nrow(player_data) >= 10)
        scale_fill_gradientn(
          colors = c("#315aa1", "grey80", "#d72029"),
          values = scales::rescale(c(0, 0.5, 1)),
          limits = c(0, 1),
          guide = "none"
        )
      } +
      theme(
        strip.text.x = element_text(size = 12, face = "bold"),
        strip.text.y.right = element_text(size = 12, angle = 0, hjust = 0, face = "bold"),
        strip.placement = "outside",
        strip.background = element_blank(),
        panel.background = element_rect(fill = "#ffffff", color = NA),
        legend.position = "none",
        plot.title = element_text(hjust = 0.5, size = 14, face = "bold")
      ) +
      labs(title = paste(player, "- Pitch Locations"))
    
    return(p)
  }
  
  # Count usage function (reverted to original pie chart style)
  count_usage <- function(data, player) {
    
    # Define pitch colors for use in the function
    pitch_colors <- c(
      "Fastball" = "dodgerblue2", "FourSeamFastBall" = "dodgerblue2", "FourSeamFastB" = "dodgerblue2",
      "OneSeamFastBall" = "yellow", "TwoSeamFastBall" = "yellow", "Sinker" = "yellow",
      "Cutter" = "orange", "Changeup" = "#01FF70", "ChangeUp" = "#01FF70",
      "Slider" = "tomato", "Sweeper" = "gold3", "Curveball" = "#B10DC9",
      "Splitter" = "#01FF70", "Knuckleball" = "lightblue",
      "Other" = "gray60", "Undefined" = "gray40"
    )
    
    data <- data %>%
      filter(Pitcher == player, TaggedPitchType != "Other") %>%
      group_by(TaggedPitchType) %>%
      mutate(n = 100 * n() / nrow(data %>% filter(Pitcher == player))) %>%
      filter(n >= 2) %>%
      ungroup()
    
    if(nrow(data) == 0) {
      return(ggplot() + 
               annotate("text", x = 0, y = 0, label = "No count data\navailable for this pitcher", size = 4) +
               theme_void())
    }
    
    data %>%
      mutate(
        count = case_when(
          Balls == 0 & Strikes == 0 ~ "0-0",
          Balls == 3 & Strikes == 2 ~ "3-2",
          Balls > Strikes ~ "Behind",
          Strikes > Balls ~ "Ahead",
          Balls == Strikes ~ "Tied"
        ),
        BatterSide = ifelse(BatterSide == "Right", "Vs Right", "Vs Left")
      ) %>%
      group_by(count, BatterSide) %>%
      mutate(total_count = n()) %>%
      group_by(count, TaggedPitchType, BatterSide) %>%
      summarise(n = n(), total = first(total_count), .groups = "drop") %>%
      mutate(percentage = n / total * 100) %>%
      arrange(desc(percentage)) %>%
      ggplot(aes(x = "", y = percentage, fill = TaggedPitchType)) +
      geom_bar(width = 1, stat = "identity") +
      coord_polar("y", start = 0) +
      facet_grid(BatterSide ~ count, labeller = label_value) +
      theme_minimal() +
      labs(title = paste(player, "- Pitch Usage by Count"), fill = "Pitch Type") +
      scale_fill_manual(values = pitch_colors) +
      theme(plot.title = element_text(hjust = 0.5, size = 12),
            axis.text = element_blank(),
            axis.title = element_blank(),
            axis.ticks = element_blank(),
            strip.text = element_text(size = 12),
            legend.position = "bottom"
      )
  }
  
  # Seam orientation functions
  filtered_seam_data <- reactive({
    req(input$seam_pitcher)
    
    base_data <- ccbl_data %>%
      filter(!is.na(SpinAxis3dSeamOrientationBallAngleHorizontalAmb1),
             !is.na(SpinAxis3dSeamOrientationBallAngleVerticalAmb1),
             Pitcher == input$seam_pitcher)
    
    # Apply pitch type filter if not "all"
    if(input$filter_pitch_type != "all") {
      base_data <- base_data %>%
        filter(TaggedPitchType == input$filter_pitch_type)
    }
    
    base_data
  })
  
  seam_curve_data <- reactive({
    if (!input$show_seam_curve) return(NULL)
    
    # Defined seam control points
    seam_points <- data.frame(
      lon = c(-157, -90, 0, 90, 157, 90, 0, -90, -157),
      lat = c(0, 45, 22.5, 45, 0, -45, -22.5, -45, 0)
    )
    
    # Create parametric interpolation
    t <- seq_along(seam_points$lon)
    interp_n <- 300
    t_interp <- seq(min(t), max(t), length.out = interp_n)
    
    lon_spline <- spline(t, seam_points$lon, xout = t_interp)$y
    lat_spline <- spline(t, seam_points$lat, xout = t_interp)$y
    
    data.frame(lon = lon_spline, lat = lat_spline)
  })
  
  # Release point visualization function
  create_release_plot <- function(game_data, pitcher_name, league_type) {
    # Filter the game data for the selected pitcher
    pitcher_data <- game_data %>%
      filter(Pitcher == pitcher_name) %>%
      filter(!is.na(RelSide), !is.na(RelHeight), !is.na(TaggedPitchType)) %>%
      # Group fastball types together
      mutate(
        pitch_group = case_when(
          TaggedPitchType %in% c("Fastball", "FourSeamFastBall", "FourSeamFastB", "Four-Seam", "4-Seam") ~ "Fastball",
          TaggedPitchType %in% c("OneSeamFastBall", "TwoSeamFastBall", "Sinker", "Two-Seam", "One-Seam") ~ "Sinker", 
          TaggedPitchType %in% c("ChangeUp", "Changeup") ~ "Changeup",
          TRUE ~ TaggedPitchType
        )
      )
    
    if(nrow(pitcher_data) == 0) {
      return(ggplot() + 
               annotate("text", x = 0, y = 4, label = "No release point data\navailable for this pitcher", size = 4) +
               theme_void())
    }
    
    # Get unique pitch groups present in data
    pitch_groups <- unique(pitcher_data$pitch_group)
    
    # Filter colors to only include pitch types present in data
    pitch_colors_used <- TMcolors[names(TMcolors) %in% pitch_groups]
    if(length(pitch_colors_used) == 0) {
      pitch_colors_used <- rainbow(length(pitch_groups))
      names(pitch_colors_used) <- pitch_groups
    }
    
    # Create the release point plot
    ggplot(pitcher_data, aes(x = RelSide, y = RelHeight, fill = pitch_group)) +
      # Add mound and rubber
      annotate("rect", xmin = -5, xmax = 5, ymin = 0, ymax = 0.83, fill = "#632b11") +
      annotate("rect", xmin = -0.5, xmax = 0.5, ymin = 0.8, ymax = 0.95,
               fill = "white", color = "black", linewidth = 0.5) +
      geom_point(size = 3, alpha = 1.5, shape = 21, color = "black", stroke = 0.5) +
      xlim(-5, 5) + ylim(0, 8) +
      annotate("text", x = -4, y = 7.5, label = "← 1B", face = "bold", color = "black", size = 6, hjust = 0.5) +
      annotate("text", x = 4, y = 7.5, label = "3B →", face = "bold", color = "black", size = 6, hjust = 0.5) +
      labs(x = "Rel Side", y = "Rel Height", title = paste(pitcher_name, "Release Heights")) +
      scale_fill_manual(values = pitch_colors_used) +
      theme_minimal() +
      theme(
        plot.title = element_text(hjust = 0.5, face = "bold", color = "black", size = 15),
        axis.title = element_text(size = 10),
        axis.text = element_text(size = 10),
        legend.position = "bottom",
        legend.text = element_text(color = "black", size = 12),
        legend.title = element_blank(),
        panel.border = element_rect(color = "black", fill = NA, size = 0.5),
        legend.key.size = unit(0.2, "cm"),
        legend.margin = margin(0, 0, 0, 0)
      )
  }
  
  # Pitch break visualization function
  create_pitch_break_plot <- function(game_data, pitcher_name, league_type) {
    
    # Filter the game data for the selected pitcher
    pitcher_data <- game_data %>%
      filter(Pitcher == pitcher_name) %>%
      filter(!is.na(HorzBreak), !is.na(InducedVertBreak), !is.na(TaggedPitchType)) %>%
      # Group fastball types together
      mutate(
        pitch_group = case_when(
          TaggedPitchType %in% c("Fastball", "FourSeamFastBall", "FourSeamFastB", "Four-Seam", "4-Seam") ~ "Fastball",
          TaggedPitchType %in% c("OneSeamFastBall", "TwoSeamFastBall", "Sinker", "Two-Seam", "One-Seam") ~ "Sinker", 
          TaggedPitchType %in% c("ChangeUp", "Changeup") ~ "Changeup",
          TRUE ~ TaggedPitchType
        )
      )
    
    p_throws <- "R"  # Default to right-handed
    
    # Apply handedness adjustment to horizontal break
    if (p_throws == "R") {
      pitcher_data$adj_HorzBreak <- pitcher_data$HorzBreak * -1
    } else {
      pitcher_data$adj_HorzBreak <- pitcher_data$HorzBreak
    }
    
    # Pitch colors
    pitch_colors <- c(
      "Fastball" = "dodgerblue2", "FourSeamFastBall" = "dodgerblue2", "FourSeamFastB" = "dodgerblue2",
      "OneSeamFastBall" = "yellow", "TwoSeamFastBall" = "yellow", "Sinker" = "yellow",
      "Cutter" = "orange", "Changeup" = "#01FF70", "ChangeUp" = "#01FF70",
      "Slider" = "tomato", "Sweeper" = "gold3", "Curveball" = "#B10DC9",
      "Splitter" = "#01FF70", "Knuckleball" = "lightblue",
      "Other" = "gray60", "Undefined" = "gray40"
    )
    
    # Get unique pitch groups present in data
    pitch_groups <- unique(pitcher_data$pitch_group)
    
    # Filter colors to only include pitch types present in data
    pitch_colors_filtered <- pitch_colors[names(pitch_colors) %in% pitch_groups]
    if(length(pitch_colors_filtered) == 0) {
      pitch_colors_filtered <- rainbow(length(pitch_groups))
      names(pitch_colors_filtered) <- pitch_groups
    }
    
    if(nrow(pitcher_data) == 0) {
      return(ggplot() + 
               annotate("text", x = 0, y = 0, label = "No pitch break data available for this pitcher", size = 6) +
               theme_void())
    }
    
    # Create the base plot
    p <- ggplot(pitcher_data, aes(x = adj_HorzBreak, y = InducedVertBreak)) +
      geom_hline(yintercept = 0, color = "#808080", alpha = 0.5, linetype = "dashed", size = 0.8) +
      geom_vline(xintercept = 0, color = "#808080", alpha = 0.5, linetype = "dashed", size = 0.8) +
      geom_point(aes(fill = pitch_group), shape = 21, size = 4, color = "black", stroke = 0.5, alpha = 1) +
      scale_fill_manual(values = pitch_colors_filtered) +
      scale_x_continuous(limits = c(-25, 25), breaks = seq(-20, 20, 10)) +
      scale_y_continuous(limits = c(-25, 25), breaks = seq(-20, 20, 10)) +
      coord_equal() +
      labs(x = "Horizontal Break (in)", y = "Induced Vertical Break (in)", title = "Pitch Breaks") +
      theme_minimal() +
      theme(
        legend.position = "none",
        plot.title = element_text(hjust = 0.5, face = "bold", color = "black", size = 16),
        aspect.ratio = 1,
        plot.background = element_rect(fill = "white", color = NA),
        panel.background = element_rect(fill = "white", color = NA),
        axis.text = element_text(color = "black", size = 10),
        axis.title = element_text(color = "black", size = 12),
        panel.border = element_rect(color = "black", fill = NA, size = 1)
      )
    
    # Add handedness-specific annotations
    if (p_throws == "R") {
      p <- p + 
        annotate("text", x = -24.2, y = -24.2, label = "← Glove Side", hjust = 0, vjust = 0, fontface = "italic", size = 3) +
        annotate("text", x = 24.2, y = -24.2, label = "Arm Side →", hjust = 1, vjust = 0, fontface = "italic", size = 3)
    }
    
    return(p)
  }
  
  # Original pitch movement visualization function
  create_pitch_movement_plot <- function(game_data, pitcher_name, league_type) {
    
    # Filter the game data for the selected pitcher
    pitcher_data <- game_data %>%
      filter(Pitcher == pitcher_name) %>%
      filter(!is.na(HorzBreak), !is.na(InducedVertBreak), !is.na(TaggedPitchType)) %>%
      # Group fastball types together
      mutate(
        pitch_group = case_when(
          TaggedPitchType %in% c("Fastball", "FourSeamFastBall", "FourSeamFastB", "Four-Seam", "4-Seam") ~ "Fastball",
          TaggedPitchType %in% c("OneSeamFastBall", "TwoSeamFastBall", "Sinker", "Two-Seam", "One-Seam") ~ "Sinker", 
          TaggedPitchType %in% c("ChangeUp", "Changeup") ~ "Changeup",
          TRUE ~ TaggedPitchType
        )
      )
    
    # Calculate average locations, usage, and velocity for each pitch group
    avg_locations <- pitcher_data %>%
      group_by(pitch_group) %>%
      summarise(
        avg_HB = mean(HorzBreak, na.rm = TRUE),
        avg_IVB = mean(InducedVertBreak, na.rm = TRUE),
        avg_velo = round(mean(RelSpeed, na.rm = TRUE), 1),
        count = n(),
        .groups = 'drop'
      ) %>%
      mutate(
        usage = paste0(round(count / sum(count) * 100, 1), "%")
      )
    
    # Determine league name for title
    league_name <- case_when(
      league_type == "ccbl" ~ "Cape Cod League",
      league_type == "ncaa2025" ~ "NCAA 2025",
      league_type == "ncaa2024" ~ "NCAA 2024",
      TRUE ~ "Unknown League"
    )
    
    # Pitch colors for movement chart
    pitch_colors <- c(
      "Fastball" = "dodgerblue2", "FourSeamFastBall" = "dodgerblue2", "FourSeamFastB" = "dodgerblue2",
      "OneSeamFastBall" = "yellow", "TwoSeamFastBall" = "yellow", "Sinker" = "yellow",
      "Cutter" = "orange", "Changeup" = "#01FF70", "ChangeUp" = "#01FF70",
      "Slider" = "tomato", "Sweeper" = "gold3", "Curveball" = "#B10DC9",
      "Splitter" = "#01FF70", "Knuckleball" = "lightblue",
      "Other" = "gray60", "Undefined" = "gray40"
    )
    
    # Get unique pitch groups present in data
    pitch_groups <- unique(pitcher_data$pitch_group)
    
    # Filter colors to only include pitch types present in data
    pitch_colors_filtered <- pitch_colors[names(pitch_colors) %in% pitch_groups]
    if(length(pitch_colors_filtered) == 0) {
      pitch_colors_filtered <- rainbow(length(pitch_groups))
      names(pitch_colors_filtered) <- pitch_groups
    }
    
    if(nrow(pitcher_data) == 0 || nrow(avg_locations) == 0) {
      return(ggplot() + 
               annotate("text", x = 0, y = 0, label = "No pitch movement data available for this pitcher", size = 6) +
               theme_void())
    }
    
    ggplot(pitcher_data, aes(x = HorzBreak, y = InducedVertBreak)) +
      geom_vline(xintercept = 0, color = "black", size = 0.8, linetype = 1) +
      geom_hline(yintercept = 0, color = "black", size = 0.8, linetype = 1) +
      geom_point(aes(fill = pitch_group), shape = 21, size = 4, color = "black", stroke = 0.4, alpha = 1) +
      geom_point(data = avg_locations, aes(x = avg_HB, y = avg_IVB, fill = pitch_group),
                 color = "black", size = 8, stroke = 2, shape = 21, alpha = 0.95) +
      geom_label(data = avg_locations, aes(x = avg_HB, y = avg_IVB, label = paste0(avg_velo, " mph")),
                 color = "black", fill = "white", size = 5, fontface = "bold", alpha = 0.9) +
      scale_fill_manual(values = pitch_colors_filtered) +
      scale_color_manual(values = pitch_colors_filtered) +
      labs(x = "HB", y = "IVB", title = paste(pitcher_name, "- Pitch Movement -", league_name)) +
      xlim(-25, 25) + ylim(-25, 25) +
      theme_void() +
      theme(
        legend.position = "right",
        plot.title = element_text(hjust = 0.5, face = "bold", color = "black", size = 20),
        aspect.ratio = 1,
        plot.background = element_rect(fill = "white", color = NA),
        panel.background = element_rect(fill = "white", color = NA),
        axis.text = element_text(color = "white", size = 10),
        axis.title = element_text(color = "white", size = 12),
        legend.background = element_rect(fill = "white"),
        legend.text = element_text(color = "black", size = 10),
        legend.title = element_blank(),
        panel.border = element_rect(color = "black", fill = NA, size = 2)
      )
  }
  
  # Reactive data for all tabs - UPDATED to use preprocessing with Stuff+
  current_data <- reactive({
    if (input$league == "ccbl") {
      preprocess_data(ccbl_data, "ccbl")
    } else if (input$league == "ncaa2025") {
      preprocess_data(TM25, "ncaa2025")
    } else if (input$league == "ncaa2024") {
      preprocess_data(NCAA, "ncaa2024")
    }
  })
  
  current_movement_data <- reactive({
    if (input$movement_league == "ccbl") {
      preprocess_data(ccbl_data, "ccbl")
    } else if (input$movement_league == "ncaa2025") {
      preprocess_data(TM25, "ncaa2025")
    } else if (input$movement_league == "ncaa2024") {
      preprocess_data(NCAA, "ncaa2024")
    }
  })
  
  current_location_data <- reactive({
    switch(input$location_league,
           "ccbl" = preprocess_data(ccbl_data, "ccbl"),
           "ncaa2025" = preprocess_data(TM25, "ncaa2025"),
           "ncaa2024" = preprocess_data(NCAA, "ncaa2024"))
  })
  
  summary_stats <- reactive({
    create_summary_stats(current_data())
  })
  
  # Update pitcher choices based on league selection for main tab
  observe({
    data <- current_data()
    pitcher_choices <- sort(unique(data$Pitcher))
    pitcher_choices <- pitcher_choices[!is.na(pitcher_choices)]
    
    updateSelectInput(session, "pitcher",
                      choices = pitcher_choices,
                      selected = pitcher_choices[1]
    )
  })
  
  # Update pitcher choices based on league selection for movement tab
  observe({
    data <- current_movement_data()
    pitcher_choices <- sort(unique(data$Pitcher))
    pitcher_choices <- pitcher_choices[!is.na(pitcher_choices)]
    
    updateSelectInput(session, "movement_pitcher",
                      choices = pitcher_choices,
                      selected = pitcher_choices[1]
    )
  })
  
  # Update pitcher choices for location tab
  observe({
    data <- current_location_data()
    pitchers <- sort(unique(data$Pitcher[!is.na(data$Pitcher)]))
    updateSelectInput(session, "location_pitcher", choices = pitchers, selected = pitchers[1])
  })
  
  # Update pitcher choices for seam orientation tab (CCBL only)
  observe({
    seam_pitcher_choices <- ccbl_data %>%
      filter(!is.na(SpinAxis3dSeamOrientationBallAngleHorizontalAmb1),
             !is.na(SpinAxis3dSeamOrientationBallAngleVerticalAmb1),
             !is.na(Pitcher)) %>%
      pull(Pitcher) %>%
      unique() %>%
      sort()
    
    updateSelectInput(session, "seam_pitcher",
                      choices = seam_pitcher_choices,
                      selected = seam_pitcher_choices[1]
    )
  })
  
  # Update pitch type filter based on selected pitcher
  observe({
    req(input$seam_pitcher)
    
    pitcher_pitch_types <- ccbl_data %>%
      filter(!is.na(SpinAxis3dSeamOrientationBallAngleHorizontalAmb1),
             !is.na(SpinAxis3dSeamOrientationBallAngleVerticalAmb1),
             Pitcher == input$seam_pitcher,
             !is.na(TaggedPitchType)) %>%
      pull(TaggedPitchType) %>%
      unique() %>%
      sort()
    
    pitch_choices <- c("All Pitches" = "all")
    if(length(pitcher_pitch_types) > 0) {
      pitch_choices <- c(pitch_choices, setNames(pitcher_pitch_types, pitcher_pitch_types))
    }
    
    updateSelectInput(session, "filter_pitch_type",
                      choices = pitch_choices,
                      selected = "all"
    )
  })
  
  # Pitcher percentiles function - UPDATED to include Stuff+
  pitcher_percentiles <- function(pitcher_name, data, summary_df, league_type) {
    # Check if AutoHitType column exists for GB% calculation
    if(!"AutoHitType" %in% names(data)) {
      gb_calc <- quote(100 * mean(Angle[PitchCall == "InPlay"] < 10, na.rm = TRUE))
    } else {
      gb_calc <- quote(100 * (sum((AutoHitType == "GroundBall" & PitchCall == "InPlay"), na.rm = TRUE) / 
                                sum(PitchCall == "InPlay", na.rm = TRUE)))
    }
    
    # Get individual pitcher stats
    pitcher_split <- data %>%
      filter(Pitcher == pitcher_name) %>%
      summarise(
        `FB Velo` = mean(RelSpeed[TaggedPitchType == "Fastball"], na.rm = TRUE),
        `FB IVB` = mean(InducedVertBreak[TaggedPitchType == "Fastball"], na.rm = TRUE),
        `Avg EV` = mean(ExitSpeed, na.rm = TRUE),
        `Hard-Hit%` = 100 * mean(is_hard_hit, na.rm = TRUE),
        `GB%` = !!gb_calc,
        `Zone%` = round(mean(in_zone, na.rm = TRUE) * 100, 1),
        `Chase%` = 100 * mean(chase, na.rm = TRUE),
        `Whiff%` = 100 * mean(is_whiff, na.rm = TRUE),
        `Z-Whiff%` = 100 * mean(in_zone_whiff, na.rm = TRUE),
        `K%` = 100 * mean(is_k, na.rm = TRUE),
        `BB%` = 100 * mean(is_walk, na.rm = TRUE),
        Extension = mean(Extension, na.rm = TRUE),
        `Stuff+` = mean(StuffPlus, na.rm = TRUE),  # ADD STUFF+ HERE TOO
        .groups = "drop"
      )
    
    if(nrow(pitcher_split) == 0) {
      return(ggplot() + annotate("text", x = 0.5, y = 0.5, label = "No data available for this pitcher") + theme_void())
    }
    
    # Convert to long format for easier processing
    pitcher_long <- pitcher_split %>%
      pivot_longer(cols = everything(), names_to = "Metric", values_to = "Value") %>%
      filter(!is.na(Value) & !is.infinite(Value))
    
    if(nrow(pitcher_long) == 0) {
      return(ggplot() + annotate("text", x = 0.5, y = 0.5, label = "No valid metrics available for this pitcher") + theme_void())
    }
    
    # Calculate percentiles for each metric
    pitcher_long$Percentile <- sapply(1:nrow(pitcher_long), function(i) {
      metric_name <- pitcher_long$Metric[i]
      pitcher_value <- pitcher_long$Value[i]
      
      if(metric_name %in% names(summary_df)) {
        league_values <- summary_df[[metric_name]]
        league_values <- league_values[!is.na(league_values) & !is.infinite(league_values)]
        
        if(length(league_values) > 1) {
          # Higher is better for these metrics
          if (metric_name %in% c("FB Velo", "FB IVB", "K%", "Whiff%", "Z-Whiff%", "Extension", "GB%", "Zone%", "Chase%", "Stuff+")) {
            percentile <- round(mean(league_values <= pitcher_value, na.rm = TRUE) * 100)
          } else {
            # Lower is better for these metrics
            percentile <- round(mean(league_values >= pitcher_value, na.rm = TRUE) * 100)
          }
          return(max(1, min(99, percentile)))
        }
      }
      return(50)
    })
    
    # Filter valid metrics
    metrics <- pitcher_long %>%
      filter(!is.na(Percentile) & !is.na(Value))
    
    if(nrow(metrics) == 0) {
      return(ggplot() + annotate("text", x = 0.5, y = 0.5, label = "No valid metrics available") + theme_void())
    }
    
    # Updated custom order to include Stuff+
    custom_order <- c("FB Velo", "FB IVB", "Avg EV", "Hard-Hit%", "GB%", "Zone%", "Chase%", "Whiff%", "Z-Whiff%", "K%", "BB%", "Extension", "Stuff+")
    
    metrics$Metric <- factor(metrics$Metric, levels = custom_order)
    metrics <- metrics %>% arrange(Metric) %>% filter(!is.na(Metric))
    
    if(nrow(metrics) == 0) {
      return(ggplot() + annotate("text", x = 0.5, y = 0.5, label = "No valid metrics available") + theme_void())
    }
    
    metrics$color <- gradient_n_pal(c("#3661ad","#90A4AE", "#d82029"))(metrics$Percentile/100)
    
    # Determine league name for subtitle
    league_name <- case_when(
      league_type == "ccbl" ~ "Cape Cod League",
      league_type == "ncaa2025" ~ "NCAA 2025", 
      league_type == "ncaa2024" ~ "NCAA 2024",
      TRUE ~ "Unknown League"
    )
    
    ggplot(metrics, aes(y = factor(Metric, levels = rev(levels(Metric))))) +
      geom_tile(aes(x = 50, width = 100), fill = "#c7dcdc", alpha = 0.3, height = 0.25) +
      geom_tile(aes(x = Percentile/2, width = Percentile, fill = color), height = 0.7) +
      annotate("segment", x = c(10, 50, 90), xend = c(10, 50, 90), y = 0, yend = nrow(metrics) + 0.5, color = c("white"), size = 1.5, alpha = 0.5) +
      geom_text(aes(x = -3, label = Metric), hjust = 1, size = 4) +
      geom_text(aes(x = 103, label = sprintf("%.1f", Value)), hjust = 0, size = 4) +
      geom_point(aes(x = Percentile, color = "white", fill = color), size = 10, shape = 21, stroke = 2) +
      geom_text(aes(x = Percentile, label = Percentile), size = 4, color = "white", fontface = "bold") +
      scale_x_continuous(limits = c(-16, 113), expand = c(0, 0)) +
      scale_fill_identity() + scale_color_identity() +
      labs(title = paste("Pitcher Percentiles:", pitcher_name), subtitle = paste("League:", league_name),
           caption = "Made by Isaac Groffman | Inspired by Baseball Savant") +
      theme_minimal() +
      theme(
        axis.text = element_blank(), axis.title = element_blank(), panel.grid = element_blank(),
        plot.title = element_text(hjust = 0.5, size = 16, face = "bold"),
        plot.subtitle = element_text(hjust = 0.5, size = 12),
        plot.caption = element_text(hjust = 1, size = 10, color = "gray60", margin = margin(t = 10))
      )
  }
  
  # OUTPUT RENDERERS
  
  # Generate percentile plot
  output$percentile_plot <- renderPlot({
    req(input$pitcher)
    data <- current_data()
    summary_df <- summary_stats()
    pitcher_percentiles(input$pitcher, data, summary_df, input$league)
  })
  
  # Generate location plot
  output$location_plot <- renderPlot({
    req(input$location_pitcher)
    data <- current_location_data()
    locations(data, input$location_pitcher)
  })
  
  # Generate count usage plot
  output$count_usage_plot <- renderPlot({
    req(input$location_pitcher)
    data <- current_location_data()
    count_usage(data, input$location_pitcher)
  })
  
  # Generate movement plot
  output$movement_plot <- renderPlot({
    req(input$movement_pitcher)
    data <- current_movement_data()
    
    if (input$plot_type == "movement") {
      create_pitch_movement_plot(data, input$movement_pitcher, input$movement_league)
    } else if (input$plot_type == "breaks") {
      create_pitch_break_plot(data, input$movement_pitcher, input$movement_league)
    }
  })
  
  # Generate release point plot
  output$release_plot <- renderPlot({
    req(input$movement_pitcher)
    data <- current_movement_data()
    create_release_plot(data, input$movement_pitcher, input$movement_league)
  })
  
  # Generate seam orientation plot
  output$seam_plot <- renderPlot({
    req(input$seam_pitcher)
    data <- filtered_seam_data()
    
    if (nrow(data) == 0) {
      return(ggplot() + 
               annotate("text", x = 0, y = 0, label = "No seam orientation data available for this pitcher", size = 6) +
               theme_void())
    }
    
    # Get world map data for Mollweide background
    world_map <- map_data("world")
    
    p <- ggplot(data, aes(x = SpinAxis3dSeamOrientationBallAngleHorizontalAmb1, 
                          y = SpinAxis3dSeamOrientationBallAngleVerticalAmb1, 
                          color = TaggedPitchType)) +
      
      # Add world map as background
      geom_polygon(data = world_map, aes(x = long, y = lat, group = group), 
                   fill = "lightgray", color = "white", alpha = 0.3, size = 0.1, inherit.aes = FALSE) +
      
      scale_color_manual(values = TMcolors, na.value = "gray60") +
      geom_point(size = input$point_size, alpha = input$point_alpha, stroke = 0.5, color = "black", shape = 21, aes(fill = TaggedPitchType)) +
      geom_point(aes(x = SpinAxis3dSeamOrientationBallAngleHorizontalAmb2, 
                     y = SpinAxis3dSeamOrientationBallAngleVerticalAmb2, fill = TaggedPitchType), 
                 size = input$point_size, alpha = input$point_alpha, stroke = 0.5, color = "black", shape = 21) +
      geom_point(aes(x = SpinAxis3dSeamOrientationBallAngleHorizontalAmb3, 
                     y = SpinAxis3dSeamOrientationBallAngleVerticalAmb3, fill = TaggedPitchType), 
                 size = input$point_size, alpha = input$point_alpha, stroke = 0.5, color = "black", shape = 21) +
      geom_point(aes(x = SpinAxis3dSeamOrientationBallAngleHorizontalAmb4, 
                     y = SpinAxis3dSeamOrientationBallAngleVerticalAmb4, fill = TaggedPitchType), 
                 size = input$point_size, alpha = input$point_alpha, stroke = 0.5, color = "black", shape = 21) +
      scale_fill_manual(values = TMcolors, na.value = "gray60") +
      coord_map("mollweide") +
      scale_x_continuous(limits = c(-180, 180), breaks = seq(-180, 180, 45)) +
      scale_y_continuous(limits = c(-90, 90), breaks = seq(-90, 90, 30)) +
      
      # Add seam orientation guide annotations with boxes
      annotate("label", x = 0, y = 67.5, label = "2S", color = "blue", size = 6, fontface = "bold", 
               fill = "white", label.padding = unit(0.3, "lines"), label.r = unit(0.15, "lines")) +
      annotate("label", x = 95, y = 52, label = "1S", color = "red", size = 6, fontface = "bold",
               fill = "white", label.padding = unit(0.3, "lines"), label.r = unit(0.15, "lines")) +
      annotate("label", x = 90, y = 0, label = "2S", color = "blue", size = 6, fontface = "bold",
               fill = "white", label.padding = unit(0.3, "lines"), label.r = unit(0.15, "lines")) +
      annotate("label", x = 95, y = -52, label = "1S", color = "red", size = 6, fontface = "bold",
               fill = "white", label.padding = unit(0.3, "lines"), label.r = unit(0.15, "lines")) +
      annotate("label", x = 0, y = -67.5, label = "2S", color = "blue", size = 6, fontface = "bold",
               fill = "white", label.padding = unit(0.3, "lines"), label.r = unit(0.15, "lines")) +
      annotate("label", x = -95, y = -52, label = "1S", color = "red", size = 6, fontface = "bold",
               fill = "white", label.padding = unit(0.3, "lines"), label.r = unit(0.15, "lines")) +
      annotate("label", x = -90, y = 0, label = "2S", color = "blue", size = 6, fontface = "bold",
               fill = "white", label.padding = unit(0.3, "lines"), label.r = unit(0.15, "lines")) +
      annotate("label", x = -95, y = 52, label = "1S", color = "red", size = 6, fontface = "bold",
               fill = "white", label.padding = unit(0.3, "lines"), label.r = unit(0.15, "lines")) +
      annotate("label", x = 180, y = 0, label = "4S", color = "darkgreen", size = 6, fontface = "bold",
               fill = "white", label.padding = unit(0.3, "lines"), label.r = unit(0.15, "lines")) +
      annotate("label", x = -180, y = 0, label = "4S", color = "darkgreen", size = 6, fontface = "bold",
               fill = "white", label.padding = unit(0.3, "lines"), label.r = unit(0.15, "lines")) +
      annotate("label", x = 0, y = 0, label = "4S", color = "darkgreen", size = 6, fontface = "bold",
               fill = "white", label.padding = unit(0.3, "lines"), label.r = unit(0.15, "lines")) +
      
      labs(
        title = paste0("Seam Orientation: ", input$seam_pitcher, 
                       if(input$filter_pitch_type != "all") paste0(" (", input$filter_pitch_type, ")") else ""),
        subtitle = "Using TrackMan 3D Spin (4S = 4-Seam, 2S = 2-Seam, 1S = 1-Seam)",
        x = "Horizontal Angle (°)",
        y = "Vertical Angle (°)",
        color = "Pitch Type"
      ) +
      theme_minimal() +
      theme(
        plot.title = element_text(size = 16, face = "bold"),
        plot.subtitle = element_text(size = 12),
        axis.title = element_text(size = 12),
        legend.title = element_text(size = 11),
        legend.text = element_text(size = 10)
      )
    
    # Add seam curve if selected
    if (input$show_seam_curve) {
      curve_data <- seam_curve_data()
      if (!is.null(curve_data)) {
        p <- p + geom_path(data = curve_data, aes(x = lon, y = lat), 
                           color = "red", linewidth = 1.5, inherit.aes = FALSE)
      }
    }
    
    p
  })
  
  # Generate seam orientation metrics table
  output$seam_metrics_table <- DT::renderDataTable({
    req(input$seam_pitcher)
    
    data <- filtered_seam_data()
    
    if (nrow(data) == 0) {
      empty_df <- data.frame(
        `Pitch Type` = character(0),
        Count = numeric(0),
        `Gyro Angle` = character(0),
        `Spin Efficiency` = character(0),
        `Active Spin` = character(0),
        `Total Spin` = character(0),
        `Avg Tilt` = character(0),
        `Spin Axis X` = character(0),
        `Spin Axis Y` = character(0),
        `Spin Axis Z` = character(0),
        `Transverse Angle` = character(0),
        check.names = FALSE
      )
      
      return(DT::datatable(empty_df, 
                           options = list(pageLength = 15, dom = 't'),
                           rownames = FALSE))
    }
    
    seam_metrics <- data %>%
      filter(!is.na(TaggedPitchType)) %>%
      group_by(`Pitch Type` = TaggedPitchType) %>%
      summarise(
        Count = n(),
        `Gyro Angle` = sprintf("%.1f°", mean(SpinAxis3dLongitudinalAngle, na.rm = TRUE)),
        `Spin Efficiency` = sprintf("%.1f%%", mean(SpinAxis3dSpinEfficiency * 100, na.rm = TRUE)),
        `Active Spin` = sprintf("%.0f rpm", mean(SpinAxis3dActiveSpinRate, na.rm = TRUE)),
        `Total Spin` = sprintf("%.0f rpm", mean(SpinRate, na.rm = TRUE)),
        `Avg Tilt` = if_else(
          all(is.na(SpinAxis3dTilt)), 
          "N/A",
          paste0(first(na.omit(SpinAxis3dTilt)))
        ),
        `Spin Axis X` = sprintf("%.3f", mean(SpinAxis3dVectorX, na.rm = TRUE)),
        `Spin Axis Y` = sprintf("%.3f", mean(SpinAxis3dVectorY, na.rm = TRUE)),
        `Spin Axis Z` = sprintf("%.3f", mean(SpinAxis3dVectorZ, na.rm = TRUE)),
        `Transverse Angle` = sprintf("%.1f°", mean(SpinAxis3dTransverseAngle, na.rm = TRUE)),
        .groups = "drop"
      ) %>%
      arrange(desc(Count))
    
    DT::datatable(seam_metrics, 
                  options = list(
                    pageLength = 15, 
                    dom = 't',
                    scrollX = TRUE,
                    columnDefs = list(
                      list(className = 'dt-center', targets = 1:10)
                    )
                  ),
                  rownames = FALSE) %>%
      DT::formatStyle(columns = names(seam_metrics), fontSize = '11px') %>%
      DT::formatStyle(
        columns = "Count",
        backgroundColor = "#f8f9fa"
      ) %>%
      DT::formatStyle(
        columns = c("Gyro Angle", "Spin Efficiency", "Avg Tilt"),
        backgroundColor = "#e8f4f8"
      ) %>%
      DT::formatStyle(
        columns = c("Active Spin", "Total Spin"),
        backgroundColor = "#e6f3ff"
      ) %>%
      DT::formatStyle(
        columns = c("Spin Axis X", "Spin Axis Y", "Spin Axis Z"),
        backgroundColor = "#fff3cd"
      )
  }, server = FALSE)
  
  # Generate pitcher statistics table - UPDATED to include Stuff+
  output$pitcher_stats <- DT::renderDataTable({
    req(input$pitcher)
    
    data <- current_data()
    
    # Check if AutoHitType column exists for GB% calculation in table
    if(!"AutoHitType" %in% names(data)) {
      gb_calc_table <- quote(sprintf("%.1f", 100 * mean(Angle[PitchCall == "InPlay"] < 10, na.rm = TRUE)))
    } else {
      gb_calc_table <- quote(sprintf("%.1f", 100 * (sum((AutoHitType == "GroundBall" & PitchCall == "InPlay"), na.rm = TRUE) / 
                                                      sum(PitchCall == "InPlay", na.rm = TRUE))))
    }
    
    pitcher_stats <- data %>%
      filter(Pitcher == input$pitcher) %>%
      summarise(
        `FB Velo` = sprintf("%.1f", mean(RelSpeed[TaggedPitchType == "Fastball"], na.rm = TRUE)),
        `Avg EV` = sprintf("%.1f", mean(ExitSpeed, na.rm = TRUE)),
        `Barrel%` = sprintf("%.1f", 100 * mean(is_barrel, na.rm = TRUE)),
        `Hard-Hit%` = sprintf("%.1f", 100 * mean(is_hard_hit, na.rm = TRUE)),
        `GB%` = !!gb_calc_table,
        `Zone%` = sprintf("%.1f", round(mean(in_zone, na.rm = TRUE) * 100, 1)),
        `Chase%` = sprintf("%.1f", 100 * mean(chase, na.rm = TRUE)),
        `Whiff%` = sprintf("%.1f", 100 * mean(is_whiff, na.rm = TRUE)),
        `Z-Whiff%` = sprintf("%.1f", 100 * mean(in_zone_whiff, na.rm = TRUE)),
        `K%` = sprintf("%.1f", 100 * mean(is_k, na.rm = TRUE)),
        `BB%` = sprintf("%.1f", 100 * mean(is_walk, na.rm = TRUE)),
        Extension = sprintf("%.1f", mean(Extension, na.rm = TRUE)),
        `Stuff+` = sprintf("%.1f", mean(StuffPlus, na.rm = TRUE)),  # ADD STUFF+ TO TABLE
        .groups = "drop"
      ) %>%
      pivot_longer(cols = everything(), names_to = "Metric", values_to = "Value")
    
    DT::datatable(pitcher_stats, 
                  options = list(pageLength = 15, dom = 't'),
                  rownames = FALSE) %>%
      DT::formatStyle(columns = c("Metric", "Value"), fontSize = '14px')
  })
  
  # Generate movement statistics table - UPDATED to include Stuff+
  output$movement_stats <- DT::renderDataTable({
    req(input$movement_pitcher)
    
    data <- current_movement_data()
    
    movement_stats <- data %>%
      filter(Pitcher == input$movement_pitcher) %>%
      filter(!is.na(HorzBreak), !is.na(InducedVertBreak), !is.na(TaggedPitchType)) %>%
      mutate(
        pitch_group = case_when(
          TaggedPitchType %in% c("Fastball", "FourSeamFastBall", "FourSeamFastB", "Four-Seam", "4-Seam") ~ "Fastball",
          TaggedPitchType %in% c("OneSeamFastBall", "TwoSeamFastBall", "Sinker", "Two-Seam", "One-Seam") ~ "Sinker", 
          TaggedPitchType %in% c("ChangeUp", "Changeup") ~ "Changeup",
          TRUE ~ TaggedPitchType
        )
      ) %>%
      group_by(`Pitch Type` = pitch_group) %>%
      summarise(
        Count = n(),
        `Usage%` = sprintf("%.1f%%", (n() / nrow(filter(data, Pitcher == input$movement_pitcher))) * 100),
        `Ext.` = sprintf("%.1f", mean(Extension, na.rm = TRUE)),
        `Avg Velo` = sprintf("%.1f mph", mean(RelSpeed, na.rm = TRUE)),
        `90th Velo` = sprintf("%.1f mph", quantile(RelSpeed, 0.9, na.rm = TRUE)),
        `Max Velo` = sprintf("%.1f mph", max(RelSpeed, na.rm = TRUE)),
        `Avg IVB` = sprintf("%.1f in", mean(InducedVertBreak, na.rm = TRUE)),
        `Avg HB` = sprintf("%.1f in", mean(HorzBreak, na.rm = TRUE)),
        `Avg Spin` = sprintf("%.0f rpm", mean(SpinRate, na.rm = TRUE)),
        `Rel Height` = sprintf("%.1f", mean(RelHeight, na.rm = TRUE)),
        `Zone%` = sprintf("%.1f%%", round(mean(in_zone, na.rm = TRUE) * 100, 1)),
        `Whiff%` = sprintf("%.1f%%", round(mean(is_whiff, na.rm = TRUE) * 100, 1)),
        `Chase%` = sprintf("%.1f%%", round(mean(chase, na.rm = TRUE) * 100, 1)),
        `Stuff+` = sprintf("%.1f", mean(StuffPlus, na.rm = TRUE)),  # ADD STUFF+ TO MOVEMENT TABLE
        .groups = "drop"
      ) %>%
      arrange(desc(Count))
    
    DT::datatable(movement_stats, 
                  options = list(pageLength = 15, dom = 't'),
                  rownames = FALSE) %>%
      DT::formatStyle(columns = names(movement_stats), fontSize = '12px')
  })
  
  # Generate location statistics table - UPDATED to include Stuff+
  output$location_stats <- DT::renderDataTable({
    req(input$location_pitcher)
    
    data <- current_location_data()
    
    location_stats <- data %>%
      filter(Pitcher == input$location_pitcher) %>%
      filter(!is.na(PlateLocSide), !is.na(PlateLocHeight), !is.na(TaggedPitchType)) %>%
      mutate(
        pitch_group = case_when(
          TaggedPitchType %in% c("Fastball", "FourSeamFastBall", "FourSeamFastB", "Four-Seam", "4-Seam") ~ "Fastball",
          TaggedPitchType %in% c("OneSeamFastBall", "TwoSeamFastBall", "Sinker", "Two-Seam", "One-Seam") ~ "Sinker", 
          TaggedPitchType %in% c("ChangeUp", "Changeup") ~ "Changeup",
          TRUE ~ TaggedPitchType
        )
      ) %>%
      group_by(`Pitch Type` = pitch_group) %>%
      summarise(
        Count = n(),
        `Usage%` = sprintf("%.1f%%", (n() / nrow(filter(data, Pitcher == input$location_pitcher))) * 100),
        `Avg Horizontal` = sprintf("%.2f ft", mean(PlateLocSide, na.rm = TRUE)),
        `Avg Vertical` = sprintf("%.2f ft", mean(PlateLocHeight, na.rm = TRUE)),
        `Zone%` = sprintf("%.1f%%", round(mean(in_zone, na.rm = TRUE) * 100, 1)),
        `Strike%` = sprintf("%.1f%%", round(mean(is_csw, na.rm = TRUE) * 100, 1)),
        `Whiff%` = sprintf("%.1f%%", round(mean(is_whiff, na.rm = TRUE) * 100, 1)),
        `Chase%` = sprintf("%.1f%%", round(mean(chase, na.rm = TRUE) * 100, 1)),
        `Stuff+` = sprintf("%.1f", mean(StuffPlus, na.rm = TRUE)),  # ADD STUFF+ TO LOCATION TABLE
        .groups = "drop"
      ) %>%
      arrange(desc(Count))
    
    DT::datatable(location_stats, 
                  options = list(pageLength = 15, dom = 't'),
                  rownames = FALSE) %>%
      DT::formatStyle(columns = names(location_stats), fontSize = '12px')
  })
}

# Run the application 
shinyApp(ui = ui, server = server)