diff --git a/katrain/core/ai.py b/katrain/core/ai.py
index 323df1e..e9704d2 100644
--- a/katrain/core/ai.py
+++ b/katrain/core/ai.py
@@ -188,10 +188,29 @@ def generate_ai_move(game: Game, ai_mode: str, ai_settings: Dict) -> Tuple[Move,
                 if ai_mode != AI_RANK:
                     n_moves = int(ai_settings["pick_frac"] * len(legal_policy_moves) + ai_settings["pick_n"])
                 else:
-                    orig_calib_avemodrank = 0.063015 + 0.7624 * board_squares/(10**(-0.05737*ai_settings["kyu_rank"]+1.9482))
+                    orig_calib_avemodrank = 0.063015 + 0.7624 * board_squares / (
+                        10 ** (-0.05737 * ai_settings["kyu_rank"] + 1.9482)
+                    )
                     norm_leg_moves = len(legal_policy_moves) / board_squares
-                    modified_calib_avemodrank = (0.3931+0.6559*norm_leg_moves*math.exp(-1*(3.002*norm_leg_moves*norm_leg_moves-norm_leg_moves-0.034889*ai_settings["kyu_rank"]-0.5097)**2)-0.01093*ai_settings["kyu_rank"]) * orig_calib_avemodrank
-                    n_moves = int(round(board_squares * norm_leg_moves/(1.31165*(modified_calib_avemodrank+1)-0.082653)))
+                    modified_calib_avemodrank = (
+                        0.3931
+                        + 0.6559
+                        * norm_leg_moves
+                        * math.exp(
+                            -1
+                            * (
+                                3.002 * norm_leg_moves * norm_leg_moves
+                                - norm_leg_moves
+                                - 0.034889 * ai_settings["kyu_rank"]
+                                - 0.5097
+                            )
+                            ** 2
+                        )
+                        - 0.01093 * ai_settings["kyu_rank"]
+                    ) * orig_calib_avemodrank
+                    n_moves = int(
+                        round(board_squares * norm_leg_moves / (1.31165 * (modified_calib_avemodrank + 1) - 0.082653))
+                    )
 
                 if ai_mode in [AI_INFLUENCE, AI_TERRITORY, AI_LOCAL, AI_TENUKI]:
                     if cn.depth > ai_settings["endgame"] * board_squares:
diff --git a/katrain/gui/widgets/graph.py b/katrain/gui/widgets/graph.py
index 8c27147..63e1807 100644
--- a/katrain/gui/widgets/graph.py
+++ b/katrain/gui/widgets/graph.py
@@ -163,8 +163,10 @@ def averagemod(data):
         (int(lendata * 0.8) + 1) - int(lendata * 0.2)
     )  # average without the best and worst 20% of ranks
 
+
 def gauss(data):
-    return math.exp(-1*(data)**2)
+    return math.exp(-1 * (data) ** 2)
+
 
 class RankGraph(Graph):
     black_rank_points = ListProperty([])
@@ -195,15 +197,21 @@ class RankGraph(Graph):
         num_legal, rank, value = zip(*non_obvious_moves)
         rank = list(rank)
         for (i, item) in enumerate(rank):
-            if item > num_legal[i]*0.09:
-                rank[i] = num_legal[i]*0.09
+            if item > num_legal[i] * 0.09:
+                rank[i] = num_legal[i] * 0.09
         rank = tuple(rank)
         averagemod_rank = averagemod(rank)
         averagemod_len_legal = averagemod(num_legal)
-        norm_avemod_len_legal = (averagemod_len_legal/num_intersec)
-        rank_kyu = -0.97222*math.log(averagemod_rank)/(0.24634+averagemod_rank*gauss(3.3208*(norm_avemod_len_legal)))+12.703*(norm_avemod_len_legal)+11.198*math.log(averagemod_rank)+12.28*gauss(2.379*(norm_avemod_len_legal))-16.544
-        if rank_kyu<-5:
-            rank_kyu=-5
+        norm_avemod_len_legal = averagemod_len_legal / num_intersec
+        rank_kyu = (
+            -0.97222 * math.log(averagemod_rank) / (0.24634 + averagemod_rank * gauss(3.3208 * (norm_avemod_len_legal)))
+            + 12.703 * (norm_avemod_len_legal)
+            + 11.198 * math.log(averagemod_rank)
+            + 12.28 * gauss(2.379 * (norm_avemod_len_legal))
+            - 16.544
+        )
+        if rank_kyu < -4:
+            rank_kyu = -4
         return 1 - rank_kyu  # dan rank
 
     @staticmethod