diff --git a/scripts/simulation.py b/scripts/simulation.py
index f103d0d..d6de19f 100644
--- a/scripts/simulation.py
+++ b/scripts/simulation.py
@@ -1,3 +1,4 @@
+import uuid
 from binding.python_c_ffi import Board
 from binding.python_c_ffi import Move
 from binding.python_c_ffi import ChessFFI
@@ -6,11 +7,12 @@ from binding.python_c_ffi import print_board
 from binding.python_c_ffi import sq_to_coord
 from scripts.evaluation import RandomEval
 from scripts.evaluation import NegaMaxEval
+from scripts.format import LongPGNFormatter
 
 
 YMD_HM = "%Y-%m-%d-%H-%M"
 MAX_MOVES = 256
-DATA_PATH = "./data"
+DATA_PATH = "./data/games"
 
 
 class Engine:
@@ -19,11 +21,12 @@ class Engine:
         self.board = Board()
         self.max_plys = max_plys
         self.moves = []
+        self.fens = []
         self._load_attack_cache()
         self._seed_engine()
 
         random_strat = RandomEval(chess_ffi=self.chess_ffi)
-        nega_strat = NegaMaxEval(chess_ffi=self.chess_ffi, depth=2, cp_window=30)
+
         if not strat_white:
             self.strat_white = random_strat
         else:
@@ -34,21 +37,33 @@ class Engine:
         else:
             self.strat_black = strat_black
 
+        self.strategies = {
+            "white": {
+                "name": self.strat_white.NAME,
+                "params": self.strat_white.get_params()
+            },
+            "black": {
+                "name": self.strat_black.NAME,
+                "params": self.strat_black.get_params()
+            },
+        }
 
-    def run(self, fen):
+
+    def run(self, fen, save=True):
         self.chess_ffi.load_fen(self.board, fen)
         self._clear_moves()
         plys = 0
 
+        self.fens.append(fen)
         while plys <= self.max_plys:
             print_board(self.board)
 
             moves_buf = (Move * MAX_MOVES)()
             n_legal = self.chess_ffi.get_legal_moves(self.board, moves_buf)
 
-            game_over, ending = self._is_game_over(n_legal)
+            game_over, result = self._is_game_over(n_legal)
             if game_over:
-                print(ending)
+                print(result)
                 break
 
             legal = [moves_buf[i] for i in range(n_legal)]
@@ -57,7 +72,6 @@ class Engine:
             else:
                 best_move = self.strat_black.get_best_move(self.board, legal)
 
-
             new_board = Board()
             if not self.chess_ffi.apply_move_on_copy(self.board, new_board, best_move):
                 print("ERROR: apply_move_on_copy failed")
@@ -66,22 +80,31 @@ class Engine:
             self.board = new_board
             plys += 1
 
+            fen_string = self.chess_ffi.board_to_fen(self.board)
+            self.fens.append(fen_string)
+            
             move = self.to_uci(best_move)
             self.moves.append(move)
 
             if self.board.halfmove_clock >= 100:
                 print("draw (50-move rule)")
+                result = "1/2-1/2"
                 break
 
-        for move in self.moves:
-            # print(move)
-            pass
-
-
         print("done")
-        print(f"winner: {self.side_tag(not self.board.side_to_move)}")
         print(len(self.moves))
 
+        if save:
+            print("saving game to disk.")
+            id = uuid.uuid4().hex
+
+            formatter = LongPGNFormatter(
+                save_path=DATA_PATH,
+                strategies=self.strategies,
+                result=result
+            )
+            formatter.save_game(id, self.moves, self.fens)
+
 
     def side_tag(self, side):
         return 'w' if side == WHITE else 'b'
@@ -90,35 +113,32 @@ class Engine:
     def load_fen(self, fen):
         self.chess_ffi.load_fen(self.board, fen)
 
-    
-    def _clear_moves(self):
-        self.moves = []
-
 
     def to_uci(self, move):
         fr = sq_to_coord(getattr(move, "from"))
         to = sq_to_coord(move.to)
-        promo = getattr(move, "promo", 0) or ""
-        
-        if promo:
-            MAP = {
-                1:"n",
-                2:"b",
-                3:"r",
-                4:"q",
-                "n":"n",
-                "b":"b",
-                "r":"r",
-                "q":"q"
-            }
-            promo = MAP.get(promo, "").lower()
-        return f"{fr}{to}{promo}"
+
+        # Only add a promotion letter if promo is set
+        promo_letter = ""
+        promo_val = int(getattr(move, "promo", 0) or 0)
+        if promo_val:
+            # Normalize piece id to type 0..5 (P,N,B,R,Q,K)
+            pt = promo_val % 6
+            # Map N=1, B=2, R=3, Q=4
+            letter_map = {1: "n", 2: "b", 3: "r", 4: "q"}
+            promo_letter = letter_map.get(pt, "")
+
+        return f"{fr}{to}{promo_letter}"
 
 
     def _load_attack_cache(self):
         self.chess_ffi.init_attack_cache()
 
 
+    def _clear_moves(self):
+        self.moves = []
+
+
     def _seed_engine(self):
         import ctypes as C, time
 
@@ -135,15 +155,28 @@ class Engine:
     def _is_game_over(self, legal_moves):
         if legal_moves == 0:
             if self.chess_ffi.in_check(self.board, self.board.side_to_move):
-                ending = "checkmate"
+                # Checkmate.
+                if self.board.side_to_move == WHITE:
+                    result = "0-1"
+                else:
+                    result = "1-0"
             else:
-                ending = "stalemate"
-            return True, ending
+                # Stalemate.
+                result = "1/2-1/2"
+            return True, result
         return False, None
     
 
 if __name__ == "__main__":
-    engine = Engine()
+    chess_ffi = ChessFFI()
+    nega_strat_1 = NegaMaxEval(chess_ffi=chess_ffi, depth=2, cp_window=5)
+    nega_strat_2 = NegaMaxEval(chess_ffi=chess_ffi, depth=1, cp_window=5)
+
+
+    engine = Engine(
+        strat_white=nega_strat_1,
+        strat_black=nega_strat_2,
+    )
 
     fen = "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1"
-    engine.run(fen)
\ No newline at end of file
+    engine.run(fen, save=True)
\ No newline at end of file