def OLSV2A(code):
df=wt.get_h_data(code,start='2001-01-01')
#df=get_hist_csv(code)
df=df.sort_index()
df.index=pd.to_datetime(df.index)
df=df.drop_duplicates(keep=False)
print ('\n,Caculating the analysing 1/4 statistics:\n')
print (df.describe(),'\n')
if code[0] in ['6','9']:
ticker='000001'
elif code[0] in ['0','2','3']:
ticker='399001'
dfi=wt.get_h_data(ticker,index=True,start='2001-01-01')
dfi=dfi.sort_index()
dfi.index=pd.to_datetime(dfi.index)
df['cpct']=df['close'].pct_change()*100
df['lpct']=df['cpct'].shift(1)
df['llpct']=df['lpct']**2
df['sina']=np.sin(df['lpct'])
df['vpct']=df['volume'].pct_change()*100
dfi['indpct']=dfi['close'].pct_change()*100
try:
rets=pd.concat([df['cpct'],df['lpct'],df['llpct'],df['sina'],dfi['indpct'],df['vpct'],df['turnover']],axis=1)
rets=rets.dropna(how='any')
X=np.array(rets.iloc[:,1:])
X=sm.add_constant(X)
except:
rets=pd.concat([df['cpct'],df['lpct'],df['llpct'],df['sina'],dfi['indpct'],df['vpct']],axis=1)
rets=rets.dropna(how='any')
X=np.array(rets.iloc[:,1:])
X=sm.add_constant(X)
Y=np.array(rets.iloc[:,0])
model=sm.OLS(Y,X)
results=model.fit()
print (results.summary())
print ("The params for the model:",results.params)
print ("The std for the model:",results.bse)
#df['predict']=''
#df.iloc[1:,df.columns.get_loc('predict')]=results.predict()
#dd=df[['close','cpct','predict']]
return results
def _get_index_data(code):
"""
code only in ['000001','399001']
"""
if code[0] == '0':
ticker = "YAHOO/SS_" + code
else:
ticker = 'YAHOO/SZ_' + code
fn = './Quandl/' + ticker + '.csv'
if not os.path.exists(fn):
print('\nDownloading the data %s:' % code)
df = wt.get_h_data(code, index=True, start='2000-01-01', end=today1)
df.sort_index(ascending=True, inplace=True)
df.rename(columns={
'open': 'Open',
'close': 'Close',
'low': 'Low',
'high': 'High',
'volume': 'Volume',
'amount': 'Amount',
'date': 'Date'
},
inplace=True)
df.to_csv(fn)
else:
dftem = pd.read_csv(fn)
tem = dftem.iloc[-1]['Date']
if tem < bftoday:
print('\nUpdating data from %s for %s:' % (tem, code))
t = time.strptime(tem, "%Y-%m-%d")
y, m, d = t[0:3]
tt = datetime.datetime(y, m, d)
bd = tt + datetime.timedelta(days=1)
if bd.weekday() == 5:
bd = bd + datetime.timedelta(days=2)
print('It is Sat')
#elif bd.weekday()==6:
# bd = bd+datetime.timedelta(days=2)
# print 'It is Sun'
bday = bd.strftime('%Y-%m-%d')
all_data1 = pd.DataFrame()
all_data = wt.get_h_data(code, autype=None, start=bday, end=today)
all_data1 = all_data1.append(all_data)
if all_data1.empty == False:
print(all_data1.head(1))
all_data1.sort_index(ascending=True, inplace=True)
all_data1.to_csv(fn, header=None, mode='a')
return
def get_h_hdf5(code):
"""
获取历史复权数据，分为前复权和后复权数据，接口提供股票上市以来所有历史数据，默认为前复权。如果不设定开始和结束日期，则返回近一年的复权数据，从性能上考虑，推荐设定开始日期和结束日期，而且最好不要超过三年以上，获取全部历史数据，请分年段分步获取，取到数据后，请及时在本地存储。
"""
h5path = './testh5df/stockdata.h5'
if os.path.exists(h5path):
h5 = pd.HDFStore(h5path, 'a', complevel=4, complib='blosc')
else:
h5 = pd.HDFStore(h5path, 'w', complevel=4, complib='blosc')
if code[0] == '0' or code[0] == '3' or code[0] == '2':
label = 'M/sz' + code
elif code[0] == '6' or code[0] == '9':
label = 'M/ss' + code
try:
dd = h5[label]
tem = str(dd.index[-1])[0:10]
if tem != today:
if datetime.datetime.today().isoweekday() in [1, 2, 3, 4, 5]:
#print 'Updating the data from%s for %s:'%(tem,code)
t = time.strptime(tem, '%Y-%m-%d')
y, m, d = t[0:3]
tt = datetime.datetime(y, m, d)
bd = tt + datetime.timedelta(days=1)
bday = bd.strftime('%Y-%m-%d')
df1 = wt.get_h_data(code, start=bday, end=today)
#df1=df1.sort_index(ascending=True,inplace=True)
df = dd.append(df1)
#df=df.sort_index(ascending=True)
h5.append(label, df, data_columns=df.columns)
except:
df = wt.get_h_data(code)
#df=df.sort_index(ascending=True)
h5.append(label, df, data_columns=df.columns)
#finally:
# h5.close()
df.index = pd.to_datetime(df.index)
return df
def get_history_data_mp(code):
h5path = './testh5df/stockdata_history.h5'
if os.path.exists(h5path):
h5 = pd.HDFStore(h5path, 'a', complevel=4, complib='blosc')
else:
h5 = pd.HDFStore(h5path, 'w', complevel=4, complib='blosc')
try:
ddf = h5[code]
#dftem=datetime.datetime.strftime('%Y-%m-%d',ddf.index[-1])
dftem = str(ddf.index[-1])[0:10]
tem = dftem
if tem != bftoday:
print('\nUpdating data from %s for %s:' % (tem, code))
t = time.strptime(tem, "%Y-%m-%d")
y, m, d = t[0:3]
tt = datetime.datetime(y, m, d)
bd = tt + datetime.timedelta(days=1)
#if bd.weekday()==5:
# bd = bd+datetime.timedelta(days=2)
# print 'It is Sat'
bday = bd.strftime('%Y-%m-%d')
all_data1 = pd.DataFrame()
all_data = wt.get_h_data(code, autype=None, start=bday, end=today)
all_data1 = all_data1.append(all_data)
if all_data1.empty == False:
#print all_data1.head(1)
#all_data1.sort_index(ascending=True,inplace=True)
df = ddf.append(all_data1)
h5[code] = df
return df
except Exception as e:
print(e)
print('\nDownloading the data %s:' % code)
df = wt.get_h_data(code, autype=None, start='2000-01-01', end=today)
#df.sort_index(ascending=True,inplace=True)
h5[code] = df
#h5.close()
return df
def get_open_h_hdf5(code, h5):
"""
获取个股全部历史交易数据
"""
if code[0] == '0' or code[0] == '3' or code[0] == '2':
label = 'M/sz' + code
elif code[0] == '6' or code[0] == '9':
label = 'M/ss' + code
try:
df = h5[label]
tem = str(df.index[-1])[0:10]
if tem < today:
#if datetime.datetime.today().isoweekday() in [1,2,3,4,5]:
t = time.strptime(tem, '%Y-%m-%d')
y, m, d = t[0:3]
tt = datetime.datetime(y, m, d)
bd = tt + datetime.timedelta(days=1)
bday = bd.strftime('%Y-%m-%d')
df1 = wt.get_h_data(code, start=bday, end=today)
df = df.append(df1)
#df.index=pd.to_datetime(df.index)
#df=df.sort_index(ascending=True)
#print(df)
h5[label] = df
except:
tem = wf.get_stock_basics()
date = tem.loc[code]['timeToMarket']
t = time.strptime(str(date), '%Y%m%d')
startt = time.strftime('%Y-%m-%d', t)
df = wt.get_h_data(code, start=startt, end=today)
if df is not None:
#df.index=pd.to_datetime(df.index)
#df=df.sort_index(ascending=True)
h5[label] = df
finally:
pass
return df
def get_h_csv(code, index=False, autype='qfq'):
"""
获取历史复权数据，分为前复权和后复权数据，接口提供股票上市以来所有历史数据，
默认为前复权。如果不设定开始和结束日期，则返回近一年的复权数据，从性能上考虑，
推荐设定开始日期和结束日期，而且最好不要超过三年以上，获取全部历史数据，
请分年段分步获取，取到数据后，请及时在本地存储。
index:False(提取非指数的数据;True(提取指数的数据)
autype：主要是提取复权类型，None 表示不复权；qfq 前复权；hfq 后复权
"""
if index:
if code[0] == '0':
tick = 'SS_' + code
else:
tick = 'SZ_' + code
fn = './stockdata/data/history/' + tick + '.csv'
if not os.path.exists(fn):
df = wt.get_h_data(code, index=True, start='1995-01-01', end=today)
df = df.sort_index(ascending=True)
df.to_csv(fn)
else:
df = pd.read_csv(fn, index_col='date')
tem = str(df.index[-1])[0:10]
#dftem=dftem.set_index('date')
if tem < bftoday:
print('\nUpdating data from %s for %s:' % (tem, code))
t = time.strptime(tem, "%Y-%m-%d")
y, m, d = t[0:3]
tt = datetime.datetime(y, m, d)
bd = tt + datetime.timedelta(days=1)
if bd.weekday() == 5:
bd = bd + datetime.timedelta(days=2)
bday = bd.strftime('%Y-%m-%d')
all_data = wt.get_h_data(code,
index=True,
start=bday,
end=today)
if all_data is not None:
all_data = all_data.sort_index(ascending=True)
all_data.to_csv(fn, header=None, mode='a')
df = df.append(all_data)
df.index = pd.to_datetime(df.index)
else:
h5path = './stockdata/data/history/' + code + '.csv'
if not os.path.exists(h5path):
tem = wf.get_stock_basics()
date = tem.loc[code]['timeToMarket']
t = time.strptime(str(date), '%Y%m%d')
startt = time.strftime('%Y-%m-%d', t)
df = wt.get_h_data(code, autype=autype, start=startt, end=today)
df.to_csv(h5path)
else:
df = pd.read_csv(h5path, index_col='date')
tem = str(df.index[-1])[0:10]
if tem < today:
#if datetime.datetime.today().isoweekday() in [1,2,3,4,5]:
t = time.strptime(tem, '%Y-%m-%d')
y, m, d = t[0:3]
tt = datetime.datetime(y, m, d)
bd = tt + datetime.timedelta(days=1)
bday = bd.strftime('%Y-%m-%d')
df1 = wt.get_h_data(code, autype=autype, start=bday, end=today)
if df1 is not None:
df1 = df1.sort_index(ascending=True)
df = df.append(df1)
df1.to_csv(h5path, mode='a', header=None)
df.index = pd.to_datetime(df.index)
return df