明确策略开发方向

.idea/codeStyles/codeStyleConfig.xml

@@ -0,0 +1,5 @@
+<component name="ProjectCodeStyleConfiguration">
+  <state>
+    <option name="PREFERRED_PROJECT_CODE_STYLE" value="Default" />
+  </state>
+</component>

selector.py

@@ -1,20 +0,0 @@
-import pandas as pd
-
-
-class Selector:
-    def select(self, codes: [str], df: pd.DataFrame) -> [str]:
-        return codes
-
-
-class Score:
-    def score(self, codes: [str], df: pd.DataFrame) -> [(str, int)]:
-        return list(map(lambda code: (code, 0), codes))
-
-
-class PeriodSelector(Selector):
-    def __init__(self, period: int = 5):
-        self.__period = period
-
-    def select(self, codes: [str], df: pd.DataFrame) -> [str]:
-        size_df = df.groupby("code").size()
-        return list(filter(lambda code: size_df[code] > self.__period, codes))

strategy.py

@@ -0,0 +1,44 @@
+import pandas as pd
+
+
+class Selector:
+    def select(self, codes: [str], df: pd.DataFrame) -> [str]:
+        return codes
+
+
+class Strategy:
+    def __init__(self, selectors: [Selector]):
+        self.selectors = selectors
+
+    def select(self, codes: [str], df: pd.DataFrame) -> [str]:
+        return list(map(lambda code: self.selectors.select(code, df), codes))
+
+
+class PeriodSelector(Selector):
+    def __init__(self, period: int = 5):
+        self.__period = period
+
+    def select(self, codes: [str], df: pd.DataFrame) -> [str]:
+        size_df = df.groupby("code").size()
+        return list(filter(lambda code: size_df[code] > self.__period, codes))
+
+
+class PyramidSelector(Selector):
+    def select(self, codes: [str], df: pd.DataFrame) -> [str]:
+        target_df = df[df["code"].isin(codes)]
+        target_df["score"] = 0
+        group_df = target_df.groupby("code")
+
+        target_df["prev_total_stockholder_interest"] = group_df["total_stockholder_interest"].shift(1)
+        target_df["roe"] = target_df["net_income"] / ((target_df["prev_total_stockholder_interest"] + target_df["total_stockholder_interest"]) / 2)
+        target_df["average_roe"] = target_df["roe"].mean()
+        target_df[target_df["average_roe"] >= 35] = target_df["score"] + 550
+        target_df[(target_df["average_roe"] < 35) & (target_df["average_roe"] >= 30)] = target_df["score"] + 500
+        target_df[(target_df["average_roe"] < 30) & (target_df["average_roe"] >= 25)] = target_df["score"] + 450
+        target_df[(target_df["average_roe"] < 25) & (target_df["average_roe"] >= 15)] = target_df["score"] + 300
+        target_df[(target_df["average_roe"] < 15) & (target_df["average_roe"] >= 10)] = target_df["score"] + 250
+
+        target_df["prev_total_assets"] = group_df["total_assets"].shift(1)
+        target_df["roa"] = target_df["net_income"] / ((target_df["prev_total_assets"] + target_df["total_assets"]) / 2)
+
+        return super().select(codes, df)

财报筛选/选股测试.ipynb

@@ -67,7 +67,7 @@
    },
    "cell_type": "code",
    "source": [
-    "from selector import PeriodSelector\n",
+    "from strategy import PeriodSelector\n",
     "\n",
     "filter_df = finance_df[(2010 < finance_df[\"year\"]) & (finance_df[\"year\"] < 2025)]\n",
     "codes = filter_df[\"code\"].unique().tolist()\n",