modified

alphamind/data/dbmodel/models.py

@@ -1074,19 +1074,18 @@ class Uqer(Base):
 class SecurityMaster(Base):
     __tablename__ = 'security_master'
 
-    Code = Column(Integer, primary_key=True)
-    exchangeCD = Column(Text())
+    exchangeCD = Column(String(4))
     ListSectorCD = Column(BigInteger)
-    ListSector = Column(Text())
+    ListSector = Column(String(6))
     transCurrCD = Column(Text())
-    secShortName = Column(Text())
+    secShortName = Column(String(10))
     secFullName = Column(Text())
-    listStatusCD = Column(Text())
+    listStatusCD = Column(String(2))
     listDate = Column(DateTime)
     delistDate = Column(DateTime)
-    equTypeCD = Column(Text())
-    equType = Column(Text())
-    exCountryCD = Column(Text())
+    equTypeCD = Column(String(4))
+    equType = Column(String(10))
+    exCountryCD = Column(String(3))
     partyID = Column(BigInteger)
     totalShares = Column(Float(53))
     nonrestFloatShares = Column(Float(53))
@@ -1095,6 +1094,7 @@ class SecurityMaster(Base):
     primeOperating = Column(Text())
     endDate = Column(DateTime)
     TShEquity = Column(Float(53))
+    Code = Column(Integer, primary_key=True)
 
 
 if __name__ == '__main__':

alphamind/examples/factor_analysis_example.py

@@ -11,33 +11,50 @@ from matplotlib import pyplot as plt
 from PyFin.api import makeSchedule
 from alphamind.api import *
 
+
+strategies = {
+    'prod': {
+        'factors': ['RVOL', 'EPS', 'DROEAfterNonRecurring', 'DivP', 'CFinc1', 'BDTO'],
+        'weights': [0.05, 0.3, 0.35, 0.075, 0.15, 0.05]
+    },
+    'candidate': {
+        'factors': ['RVOL', 'EPS', 'CFinc1', 'BDTO', 'VAL', 'GREV', 'ROEDiluted'],
+        'weights': [0.02, 0.2, 0.15, 0.05, 0.2, 0.2, 0.2]
+    }
+}
+
+
 engine = SqlEngine("mssql+pymssql://licheng:A12345678!@10.63.6.220/alpha")
 universe = Universe('custom', ['zz500'])
+benchmark_code = 905
 neutralize_risk = ['SIZE'] + industry_styles
 constraint_risk = ['SIZE'] + industry_styles
 n_bins = 5
-factors = ['ROEDiluted']
-factor_weights = np.array([1.])
-
-dates = makeSchedule('2017-01-01',
+dates = makeSchedule('2012-01-14',
                      '2017-08-14',
                      tenor='1w',
                      calendar='china.sse')
 
 final_res = np.zeros((len(dates), n_bins))
 
-rets = []
 
-for i, date in enumerate(dates):
-    print(date)
+total_data_dict = {}
+
+for strategy in strategies:
+    factors = strategies[strategy]['factors']
+    factor_weights = strategies[strategy]['weights']
+
+    rets = []
+    for i, date in enumerate(dates):
         ref_date = date.strftime('%Y-%m-%d')
 
         codes = engine.fetch_codes(ref_date, universe)
 
-    data = engine.fetch_data(ref_date, factors, codes, 905)
+        data = engine.fetch_data(ref_date, factors, codes, benchmark_code)
         returns = engine.fetch_dx_return(ref_date, codes, horizon=4)
 
         total_data = pd.merge(data['factor'], returns, on=['Code']).dropna()
+        print(date, ': ', len(total_data))
         risk_exp = total_data[neutralize_risk].values.astype(float)
         industry = total_data.industry_code.values
         dx_return = total_data.dx.values
@@ -66,14 +83,17 @@ for i, date in enumerate(dates):
                                             benchmark=benchmark,
                                             is_tradable=total_data.isOpen.values.astype(bool),
                                             method='risk_neutral',
-                                        constraints=constraint)
-    except:
-        print(date, ' is error!')
+                                            constraints=constraint,
+                                            use_rank=100)
+        except Exception as e:
+            print(e)
             rets.append(0.)
         else:
-        rets.append(analysis.er[-1])
+            rets.append(analysis.er[-1] / benchmark.sum())
+
+    total_data_dict[strategy] = rets
+
 
-ret_series = pd.Series(rets, dates)
-ret_series.cumsum().plot(figsize=(12, 6))
-plt.title('{0} weekly reblance'.format(factors))
-plt.show()
+ret_df = pd.DataFrame(total_data_dict, index=dates)
+ret_df.cumsum().plot(figsize=(12, 6))
+plt.savefig("backtest_big_universe_20170814.png")

alphamind/examples/quantile_analysis_example.py

@@ -12,21 +12,26 @@ from PyFin.api import makeSchedule
 from alphamind.api import *
 
 engine = SqlEngine("mssql+pymssql://licheng:A12345678!@10.63.6.220/alpha")
-universe = Universe('custom', ['zz500'])
+#engine = SqlEngine('postgresql+psycopg2://postgres:A12345678!@10.63.6.220/alpha')
+universe = Universe('custom', ['pm500_mirror'])
 neutralize_risk = ['SIZE'] + industry_styles
 n_bins = 5
-factors = ['ROEDiluted']
+
 factor_weights = np.array([1.])
 
-dates = makeSchedule('2017-01-01',
+dates = makeSchedule('2016-08-14',
                      '2017-08-14',
                      tenor='1w',
                      calendar='china.sse')
 
-final_res = np.zeros((len(dates), n_bins))
+prod_factors = ['EARNYILD', 'ROAEBIT']
+
+for factor in prod_factors:
 
-for i, date in enumerate(dates):
-    print(date)
+    factors = [factor]
+    final_res = np.zeros((len(dates), n_bins))
+
+    for i, date in enumerate(dates):
         ref_date = date.strftime('%Y-%m-%d')
 
         codes = engine.fetch_codes(ref_date, universe)
@@ -35,6 +40,7 @@ for i, date in enumerate(dates):
         returns = engine.fetch_dx_return(ref_date, codes, horizon=4)
 
         total_data = pd.merge(data['factor'], returns, on=['Code']).dropna()
+        print(date, ': ', len(total_data))
         risk_exp = total_data[neutralize_risk].values.astype(float)
         dx_return = total_data.dx.values
         benchmark = total_data.weight.values
@@ -47,13 +53,15 @@ for i, date in enumerate(dates):
                                     risk_exp=risk_exp,
                                     n_bins=n_bins,
                                     benchmark=benchmark)
-    except:
-        print(date, ' is error!')
+        except Exception as e:
+            print(e)
             res = np.zeros(n_bins)
 
-    final_res[i] = res
+        final_res[i] = res / benchmark.sum()
+
+    df = pd.DataFrame(final_res, index=dates)
+    df.cumsum().plot(figsize=(12, 6))
+    plt.title('{0} weekly re-balance'.format(factors[0]))
+    plt.savefig('{0}_big_universe_20170814.png'.format(factors[0]))
 
-df = pd.DataFrame(final_res, index=dates)
-df.cumsum().plot(figsize=(12, 6))
-plt.title('{0} weekly reblance'.format(factors))
 plt.show()
\ No newline at end of file