class DailyStocks(object):
db, db_engine = Database.initSession()
def __init__(self, date):
self.date = date
self.CGO = {}
self.stock_list = list(
pd.read_sql_query("select distinct code from basic_list",
self.db_engine).code)
def get_cgo(self):
for code in self.stock_list:
try:
# 找到当前日期，获得当前日期以前的100个数据
sql = "select * from basic_list where code =" + "'" + code + "'" + "and date < " + "'" + self.date + "'"
data = pd.read_sql_query(sql, self.db_engine)
data = data[-100:]
data.reset_index(drop=True, inplace=True)
# 计算每日均价
np_data = np.array(
data[['close_price', 'ave_price', 'turnover_rate']])
length = np.shape(np_data)[0]
for i in range(length):
for j in range(i + 1, length):
# data.loc[index, 'turnover_weight'] *= (1 - data.loc[j, 'turnover_weight'])
np_data[i][2] *= (1 - np_data[j][2])
num = np.sum(np_data[:, 2])
np_data[:, 2] /= num
RP = np.sum(np_data[:, 2] * np_data[:, 1])
# data.eval("""
# RP = turnover_weight * ave_price
# """, inplace=True)
# RP = data['RP'].sum()
self.CGO[code] = (np_data[-1][0] - RP) / RP
except:
pass
def insertCGO(self):
ss = scoped_session(self.db)
dbs = ss()
data = self.CGO.items()
dbs.execute(Database.cgo_list.__table__.insert(),
[{
"date": self.date,
"code": element[0],
"cgo_factor": float(element[1])
} for element in data])
dbs.commit()
dbs.close()
ss.remove()
def get_top5(date):
db, db_engine = Database.initSession()
cgo_sql = "select * from cgo_list where date =" + "'" + date + "'"
cgo = pd.read_sql_query(cgo_sql, db_engine)
data = cgo.sort_values(by='cgo_factor', ascending=False)
data = data[0 * int(data.shape[0] / 5):1 * int(data.shape[0] / 5)]
stock_list = list(data.code)
return stock_list
class Factors(object):
db, db_engine = Database.initSession()
def __init__(self):
self.data = None
def getToday(self):
self.data = pd.read_excel(path, dtype=object)
self.data[['return']] = self.data[['return']].astype(float)
def insertToday(self):
engine = self.db_engine
max_iteration = math.ceil(self.data.shape[0] / 50000)
for i in range(max_iteration):
if i != max_iteration - 1:
self.data[i * 50000:(i + 1) * 50000].to_sql('basic_list',
con=engine,
if_exists='append',
index=False)
else:
self.data[i * 50000:].to_sql('basic_list',
con=engine,
if_exists='append',
index=False)
# total_threads[st].start()
# running_threads.append(total_threads[st])
# st += 1
# continue
# for t in running_threads:
# if not t.is_alive():
# running_threads.remove(t)
# for t in total_threads:
# t.join()
# f = Factors()
# f.getToday()
#print(f.data)
#获取时间交集，保证全部时间的交集一致
db, db_engine = Database.initSession()
basic = pd.read_excel(path, dtype=object)
code_list = "select distinct code from basic_list"
code_list = list(pd.read_sql_query(code_list, db_engine).code)
error_code = []
for code in code_list:
#date_sql = "select distinct date from basic_list where code =" + "'"+code+"'"
#date_list = list(pd.read_sql_query(date_sql, db_engine).date)
date_list = list(basic[basic['code'] == code].date)
if len(date_list) > 1900:
pass
#print(allin(date_list,date_intersection))