In [2]:
import numpy as np
import pandas as pd
import datetime as dt
pd.set_option('display.max_rows', 16)
import matplotlib.pyplot as plt
%matplotlib inline
plt.rcParams['figure.figsize'] = (16.0, 9.0)
import seaborn as sns
import statsmodels.api as sm
from sklearn.linear_model import LinearRegression
import gc
In [3]:
import matplotlib.ticker as ticker
In [4]:
plt.rcParams['figure.figsize'] = (16.0, 9.0)
Information Coefficient Analysis¶
In [5]:
ret_df = pd.read_pickle('./output_data/factor_exposure/all_exposure.pkl')
In [6]:
ret_df
Out[6]:
In [7]:
ic_illiq = ret_df.groupby('ret_date')[['exret','illiq']].corr(method='spearman')
ic_illiq.reset_index(inplace=True)
ic_illiq
Out[7]:
In [8]:
ic_illiq = ic_illiq[ic_illiq['level_1'] != 'illiq'].drop(['exret','level_1'],axis=1)
In [9]:
ic_illiq
Out[9]:
In [10]:
ic_illiq.set_index('ret_date',inplace=True)
In [11]:
# ic_illiq.index = ic_illiq.index.to_timestamp()
In [12]:
# fig, ax = plt.subplots()
# plt.bar(ic_illiq.index, ic_illiq['illiq'])
# plt.gca().xaxis.set_major_locator(dates.MonthLocator())
# plt.gca().xaxis.set_major_formatter(dates.DateFormatter("%b\n%Y"))
# # fig.autofmt_xdate()
# plt.show()
ax = ic_illiq.plot.bar(rot=45)
positions = [i if not i%4 else 0 for i in ax.get_xticks()]
positions.append(ax.get_xticks()[-1])
positions
ax.set_xticks(positions)
ax.set_xticklabels([t if not i%4 else "" for i,t in enumerate(ax.get_xticklabels())])
# ax.xaxis.set_major_locator(ticker.MultipleLocator(20))
# ax.xaxis.set_minor_locator(ticker.MultipleLocator(2.5))
plt.show()
In [13]:
ic_size = ret_df.groupby('ret_date')[['exret','size']].corr(method='spearman')
ic_size.reset_index(inplace=True)
ic_size = ic_size[ic_size['level_1'] != 'size'].drop(['exret','level_1'],axis=1)
ic_size.set_index('ret_date',inplace=True)
ax = ic_size.plot.bar(rot=45)
positions = [i if not i%4 else 0 for i in ax.get_xticks()]
positions.append(ax.get_xticks()[-1])
positions
ax.set_xticks(positions)
ax.set_xticklabels([t if not i%4 else "" for i,t in enumerate(ax.get_xticklabels())])
plt.show()
In [14]:
ic_rev = ret_df.groupby('ret_date')[['exret','rev']].corr(method='spearman')
ic_rev.reset_index(inplace=True)
ic_rev = ic_rev[ic_rev['level_1'] != 'rev'].drop(['exret','level_1'],axis=1)
ic_rev.set_index('ret_date',inplace=True)
ax = ic_rev.plot.bar(rot=45)
positions = [i if not i%4 else 0 for i in ax.get_xticks()]
positions.append(ax.get_xticks()[-1])
positions
ax.set_xticks(positions)
ax.set_xticklabels([t if not i%4 else "" for i,t in enumerate(ax.get_xticklabels())])
plt.show()
In [15]:
ic_mom = ret_df.groupby('ret_date')[['exret','mom']].corr(method='spearman')
ic_mom.reset_index(inplace=True)
ic_mom = ic_mom[ic_mom['level_1'] != 'mom'].drop(['exret','level_1'],axis=1)
ic_mom.set_index('ret_date',inplace=True)
ax = ic_mom.plot.bar(rot=45)
positions = [i if not i%4 else 0 for i in ax.get_xticks()]
positions.append(ax.get_xticks()[-1])
positions
ax.set_xticks(positions)
ax.set_xticklabels([t if not i%4 else "" for i,t in enumerate(ax.get_xticklabels())])
plt.show()
In [16]:
ic_ivol = ret_df.groupby('ret_date')[['exret','ivol']].corr(method='spearman')
ic_ivol.reset_index(inplace=True)
ic_ivol = ic_ivol[ic_ivol['level_1'] != 'ivol'].drop(['exret','level_1'],axis=1)
ic_ivol.set_index('ret_date',inplace=True)
ax = ic_ivol.plot.bar(rot=45)
positions = [i if not i%4 else 0 for i in ax.get_xticks()]
positions.append(ax.get_xticks()[-1])
positions
ax.set_xticks(positions)
ax.set_xticklabels([t if not i%4 else "" for i,t in enumerate(ax.get_xticklabels())])
plt.show()
In [17]:
cols = ['beta','size','bm','mom','rev','illiq','ivol']
ic_df = pd.DataFrame()
for col in cols:
temp = ret_df.groupby('ret_date')[['exret',col]].corr(method='spearman')
temp.reset_index(inplace=True)
ic_df[col] = temp[temp['level_1'] != col].drop(['exret','level_1','ret_date'],axis=1)[col].values
In [18]:
ic_df.index = np.sort(ret_df['ret_date'].unique().dropna())
In [19]:
ic_df[cols].rolling(3).mean().plot(subplots=True,grid=True)
Out[19]:
information ratio = IC.mean()/IC.std()
In [20]:
# Information ratio
ic_df.apply(lambda x: x.mean()/x.std())
Out[20]:
Fund performance¶
Fund return data¶
In [21]:
START = '2008-01-01'
END = '2025-03-31'
# Security Id
fund_info = DataAPI.SecIDGet(assetClass="F",pandas="1")
In [22]:
fund_info
Out[22]:
In [23]:
# fund_id = ['110022','009550','001938','009341',
# '000751','004997','009863','001714',
# '001410','004851','005827','110011',
# '260108','163402','320007','163417',
# '001511','161728','161005','161131']
# fund_id =["519212","519191","519185","001678",
# "004685","001437","519198","002049",
# "010761","000398","011891","001470",
# "166301","008488","006195","002601",
# "001959","011403","013142","001763"]
# fund_id = [id_+'.OFCN' for id_ in fund_id]
# fund_id = ['014283', '001614', '016325', '001323', '001322', '001167', '210009', '014273', '290012', '016307', '014277', '210002', '162102', '004890', '620007', '014279', '217021', '010383', '005310', '004685']
fund_id = [
"017102",
"017490",
"009707",
"015641",
"006718",
"001480",
"014915",
"009062",
"001412",
"501046"
]
In [24]:
len(fund_id)
Out[24]:
In [25]:
fund_name = fund_info.loc[fund_info['ticker'].isin(fund_id),['ticker','secShortName']]
fund_name
Out[25]:
In [26]:
print(fund_name.to_string())
In [27]:
# fund_name.loc[~fund_name['secShortName'].str.contains('债')]
In [61]:
# fund_name = fund_name.loc[~fund_name['secShortName'].str.contains('债')].copy()
In [28]:
fund_name.loc[fund_name.duplicated('ticker',keep=False)] # keep='last'
Out[28]:
In [29]:
fund_name = fund_name.drop_duplicates('ticker',keep='last')
fund_name
Out[29]:
In [30]:
# Fund_NAV_Month.csv 从 CSMAR 下载
# 基金市场系列 -> 公募基金 -> 基金表现 -> 基金月净值文件
fund_df = pd.read_csv('./data/Fund_NAV_Month2025.csv',dtype={'Symbol':str})
fund_df
Out[30]:
In [31]:
top_fund_df = fund_df[fund_df['Symbol'].isin(fund_id)].copy()
In [32]:
top_fund_df['Symbol'].nunique()
Out[32]:
In [38]:
list(set(fund_id) - set(top_fund_df['Symbol'].unique()))
Out[38]:
In [39]:
top_fund_df.rename(columns={'Symbol':'ticker','TradingMonth':'ret_date','ReturnNAV':'ret'},inplace=True)
In [40]:
top_fund_df = pd.merge(top_fund_df, fund_name, on='ticker')
In [41]:
top_fund_df = top_fund_df[['ret_date','ticker','secShortName','ret']].copy()
In [42]:
top_fund_df['ret_date'] = pd.to_datetime(top_fund_df['ret_date']).dt.to_period('M')
In [43]:
top_fund_df
Out[43]:
Factor return data¶
In [44]:
# factor_ret = pd.read_pickle('./output_data/factors/factors_all.pkl')
factor_ret = pd.read_pickle('./output_data/factors/factors_all_long_only.pkl')
In [45]:
factor_ret
Out[45]:
In [46]:
reg_df = pd.merge(factor_ret.reset_index(), top_fund_df, on='ret_date')
In [47]:
reg_df['exret'] = reg_df['ret']-reg_df['rf']
In [48]:
reg_df
Out[48]:
In [49]:
# results = {}
# for fund in fund_id:
# df_ = reg_df[reg_df['ticker'] == fund].copy()
# X = df_[['exmktret','SMB','HML','rev','illiq','ivol']]
# y = df_['exret']
# X = sm.add_constant(X)
# reg = sm.OLS(y, X).fit().get_robustcov_results(cov_type='HAC', maxlags=4)
# results[fund] = pd.DataFrame([reg.params,reg.tvalues],columns=['alpha','exmktret','SMB','HML','rev','illiq','ivol'],
# index=['coefs','tvalues'])
results = {}
for fund in fund_id:
df_ = reg_df[reg_df['ticker'] == fund].copy()
if df_.shape[0] == 0:
continue
else:
X = df_[['exmktret','small_only','high_only','rev_long','illiq_long','ivol_long']]
y = df_['exret']
X = sm.add_constant(X)
reg = sm.OLS(y, X).fit().get_robustcov_results(cov_type='HAC', maxlags=4)
results[fund] = pd.DataFrame([reg.params,reg.tvalues],columns=['alpha','exmktret','small_only','high_only','rev_long','illiq_long','ivol_long'],
index=['coefs','tvalues'])
In [50]:
fund_id = list(set(fund_id).intersection(reg_df['ticker'].unique()))
In [51]:
for fund in fund_id:
print(fund_name.loc[fund_name['ticker']==fund,'secShortName'].values)
display(results[fund])
# print('#################'*5)
# for fund in fund_id[10:20]:
# print(fund_name.loc[fund_name['ticker']==fund,'secShortName'].values)
# display(results[fund])
In [ ]: