Refactor exposure_spread function to streamline comparable player selection logic, ensuring accurate filtering based on salary, median projections, and position overlap, while maintaining team removal functionality.

global_func/exposure_spread.py

@@ -233,26 +233,6 @@ def exposure_spread(working_frame, exposure_player, exposure_target, exposure_st
         player_pos_list = player_positions.split('/')
         return any(pos in target_positions for pos in player_pos_list)
 
-    comparable_players = projections_df[
-        (projections_df['salary'] >= comp_salary_low) &
-        (projections_df['salary'] <= comp_salary_high) &
-        (projections_df['median'] >= comp_projection_low) &
-        (projections_df['median'] <= comp_projection_high) &
-        (projections_df['position'].apply(lambda x: has_position_overlap(x, comp_player_position)))
-    ]
-
-    if exposure_target == 0:
-        comparable_players = comparable_players[comparable_players['player_names'] != exposure_player]
-    
-    if remove_teams is not None:
-        remove_mask = comparable_players.apply(
-                lambda row: not any(team in list(row) for team in remove_teams), axis=1
-            )
-        comparable_players = comparable_players[remove_mask]
-
-    # Create a list of comparable players
-    comparable_player_list = comparable_players['player_names'].tolist()
-
     # find the exposure rate of the player in the working frame
     player_mask = working_frame[working_frame.columns].apply(
                                     lambda row: exposure_player in list(row), axis=1
@@ -267,7 +247,6 @@ def exposure_spread(working_frame, exposure_player, exposure_target, exposure_st
     if comparable_stack != 0:
         player_rows = player_rows[player_rows['Stack'] != comparable_stack]
     
-
     change_counter = 0
 
     random_row_indices = list(player_rows.index)
@@ -276,6 +255,24 @@ def exposure_spread(working_frame, exposure_player, exposure_target, exposure_st
     # for each row to the the number of lineups to remove, replace with random choice from comparable player list
     for row in random_row_indices:
         if change_counter < math.ceil(lineups_to_remove):
+            comparable_players = projections_df[
+                (projections_df['salary'] >= comp_salary_low) &
+                (projections_df['salary'] <= comp_salary_high + (50000 - working_frame['salary'][row])) &
+                (projections_df['median'] >= comp_projection_low) &
+                (projections_df['position'].apply(lambda x: has_position_overlap(x, comp_player_position)))
+            ]
+
+            if exposure_target == 0:
+                comparable_players = comparable_players[comparable_players['player_names'] != exposure_player]
+            
+            if remove_teams is not None:
+                remove_mask = comparable_players.apply(
+                        lambda row: not any(team in list(row) for team in remove_teams), axis=1
+                    )
+                comparable_players = comparable_players[remove_mask]
+
+            # Create a list of comparable players
+            comparable_player_list = comparable_players['player_names'].tolist()
             insert_player = random.choice(comparable_player_list)
             # Find which column contains the exposure_player
             row_data = working_frame.iloc[row]