def get_zz_quote(code, end_date=None, count=None, start_date=None):
'''获取中证指数行情,返回panel结构'''
if isinstance(code, str):
code = [code]
code.sort()
code = [x[:6] for x in code]
days = get_trade_days(start_date, end_date, count)
code_df = jy.run_query(
query(jy.SecuMain.InnerCode,
jy.SecuMain.SecuCode, jy.SecuMain.ChiName).filter(
jy.SecuMain.SecuCode.in_(code)).order_by(
jy.SecuMain.SecuCode))
df = jy.run_query(
query(jy.QT_CSIIndexQuote).filter(
jy.QT_CSIIndexQuote.IndexCode.in_(code_df.InnerCode),
jy.QT_CSIIndexQuote.TradingDay.in_(days),
))
df2 = pd.merge(code_df, df, left_on='InnerCode',
right_on='IndexCode').set_index(['TradingDay', 'SecuCode'])
df2.drop([
'InnerCode', 'IndexCode', 'ID', 'UpdateTime', 'JSID', 'OpenInterest',
'SettleValue', 'IndexCSIType'
],
axis=1,
inplace=True)
return df2.to_panel()
def get_SW_index(SW_index = 801010,start_date = '2017-01-31',end_date = '2018-01-31'):
index_list = ['PrevClosePrice','OpenPrice','HighPrice','LowPrice','ClosePrice','TurnoverVolume','TurnoverValue','TurnoverDeals','ChangePCT','UpdateTime']
jydf = jy.run_query(query(jy.SecuMain).filter(jy.SecuMain.SecuCode==str(SW_index)))
link=jydf[jydf.SecuCode==str(SW_index)]
rows=jydf[jydf.SecuCode==str(SW_index)].index.tolist()
result=link['InnerCode'][rows]
df = jy.run_query(query(jy.QT_SYWGIndexQuote).filter(jy.QT_SYWGIndexQuote.InnerCode==str(result[0]),\
jy.QT_SYWGIndexQuote.TradingDay>=start_date,\
jy.QT_SYWGIndexQuote.TradingDay<=end_date
))
df.index = df['TradingDay']
df = df[index_list]
return df
def future_basic_info(self, future):
# 这个后续改成从本地的meta数据中拿metadata数据的方法，暂时先这么使用
if "9999" in future or "8888" in future:
match = re.match(r"(?P<underlying_symbol>[A-Z]{1,})", future)
if not match:
raise ValueError("未知期货标的：{}".format(future))
else:
future = get_dominant_future(
match.groupdict()["underlying_symbol"])
code = future.split(".")[0]
q = query(jy.Fut_ContractMain).filter(
jy.Fut_ContractMain.ContractCode == code)
result = jy.run_query(query_object=q).to_dict("record")
if result:
result = result.pop()
min_point = re.match(
"(?P<value>^[0-9]+([.]{1}[0-9]+){0,1})",
result["LittlestChangeUnit"]).groupdict(nan)["value"]
return {
"ContractUnit":
result["CMValue"],
"PriceScale":
float(str(min_point)[:-1] +
"1") if float(min_point) < 1 else 1,
"MinPoint":
float(min_point)
}
else:
cu = nan
current_code = code[:-4].upper()
if trade_units.get(current_code):
cu = trade_units.get(current_code)
return {"ContractUnit": cu, "PriceScale": nan, "MinPoint": nan}
def __secu_to_jy(cls, codes, category=1):
df = pd.DataFrame()
# 因jy每次最多返回3000条数据，所以要多次查询
# 偏移值
offset = 0
while True:
q = query(
# 内部代码
jy.SecuMain.InnerCode, ).filter(
# 去除聚宽代码后缀
jy.SecuMain.SecuCode.in_(codes),
# 限定查询股票
jy.SecuMain.SecuCategory == category
# 偏移
).offset(offset)
# 查询
temp_df = jy.run_query(q)
# 无数据时退出
if len(temp_df) == 0:
break
# 追加数据
df = df.append(temp_df)
# 偏移值每次递增3000
offset += 3000
# 返回代码list
return df.InnerCode.tolist()
def on_strategy_end(context):
trade_list = ['399006']
incode = jy.run_query(
query(
jy.SecuMain.SecuCode,
jy.SecuMain.InnerCode,
).filter(jy.SecuMain.SecuCode.in_(trade_list), ))
q = query(
jy.LC_Dividend.InnerCode, # 证券内部编码
jy.LC_Dividend.ExDiviDate, # 除权除息日
).filter(jy.LC_Dividend.InnerCode.in_(incode['InnerCode']), ).order_by(
jy.LC_Dividend.ExDiviDate.desc())
df = jy.run_query(q).fillna(value=0, method=None, axis=0)
df2 = pd.merge(incode, df, on='InnerCode')
print(df2)
# 策略运行结束时调用(可选)
print('回测结束')
def get_market_news(news_type):
#今天日期
today = pd.datetime.today().strftime('%Y-%m-%d')
#三个时间段
morning_time = today + ' 09:30:00'
afternoon_time = today + ' 13:00:00'
evening_time = today + ' 22:00:00'
#查询聚源数据库ED_GilBroadcast
q = query(
#具体时间
jy.ED_GilBroadcast.InfoPublTime,
#标题
jy.ED_GilBroadcast.InfoTitle,
#正文
jy.ED_GilBroadcast.Content,
).filter(
#按照发布日期过虑
jy.ED_GilBroadcast.InfoPublDate == today).order_by(
#按照发布时间排序
jy.ED_GilBroadcast.InfoPublTime.desc())
df = jy.run_query(q)
#筛选、组织数据
title = '市场快讯（%s DGCIdea）' % (today)
if news_type == 'morning':
df = df[(df.InfoPublTime <= morning_time)]
title = '早报：%s' % title
elif news_type == 'afternoon':
df = df[(df.InfoPublTime > morning_time)
& (df.InfoPublTime <= afternoon_time)]
title = '午报：%s' % title
elif news_type == 'evening':
df = df[(df.InfoPublTime > afternoon_time)
& (df.InfoPublTime <= evening_time)]
title = '晚报：%s' % title
#生成html
if len(df) == 0:
return ('', '')
else:
news = ''
for i in range(len(df)):
news = news + '<hr />'
news = news + '<p>'
news = news + df.iloc[i].InfoTitle + '<br/>'
news = news + df.iloc[i].InfoPublTime.strftime(
'%Y-%m-%d %H:%M:%S') + '<br/>'
news = news + df.iloc[i].Content.decode('gbk').replace(
'\n', '<br/>') + '<br/>'
news = news + '<br/></p>'
return (title, news)
def __get_jy_divid(code, end_date):
"""
指定日期股息率,jy版本
code：股票代码,list or str
end_date：截至日期
返回：股息率、市值
"""
# 判断代码是str还是list
if not type(code) is list:
code = [code]
# 聚宽代码转换为jy内部代码
InnerCodes = Code.stk_to_jy(code)
# 获取所有成份股派息总额
# 数据表
df = pd.DataFrame()
# 因jy每次最多返回3000条数据，所以要多次查询
# 偏移值
offset = 0
while True:
# 查询语句
q = query(
# 内部代码
jy.LC_Dividend.InnerCode,
# 派息日期
jy.LC_Dividend.ToAccountDate,
# 派息总额
jy.LC_Dividend.TotalCashDiviComRMB,
).filter(
# 已分红
jy.LC_Dividend.IfDividend == 1,
# 获取指定日期前所有分红
jy.LC_Dividend.ToAccountDate <= end_date,
jy.LC_Dividend.InnerCode.in_(InnerCodes)
# 偏移
).offset(offset)
# 查询
temp_df = jy.run_query(q)
if len(temp_df) == 0:
break
# 追加数据
df = df.append(temp_df)
# 偏移值每次递增3000
offset += 3000
if len(df) == 0:
div = float('NaN')
else:
# 生成排序字段
df['sort'] = df['InnerCode'].astype(
'str') + df['ToAccountDate'].astype('str').str[0:10]
df = df.sort('sort')
# 只保留最后一次派息数据
df = df.drop_duplicates('InnerCode', take_last=True) # keep='last'
# 返回合计的派息数
div = df['TotalCashDiviComRMB'].sum()
# 获取指数总市值
q = query(
# 市值
valuation.market_cap).filter(valuation.code.in_(code))
# 获取各成份股市值(亿元)
df = get_fundamentals(q, end_date)
# 返回合计的成份股总市值（亿元）
cap = df['market_cap'].sum()
try:
# 返回股息率
return div / cap / 100000000 * 100.0, cap
except:
return float('NaN'), float('NaN')
def fun_get_Dividend_yield(self, stocks, statsDate):
# 按照派息日计算，计算过去 12个月的派息率(TTM)
start_date = statsDate - dt.timedelta(365)
end_date = statsDate
year = statsDate.year
#将当前股票池转换为聚源的6位股票池
stocks_symbol = []
for s in stocks:
stocks_symbol.append(s[0:6])
#查找聚源的内部编码
df_code = jy.run_query(
query(
jy.SecuMain.InnerCode, #证券内部编码
jy.SecuMain.SecuCode #证券代码
))
df_code = df_code[df_code.SecuCode.isin(stocks_symbol)]
list_InnerCode = list(set(list(df_code['InnerCode'])))
# 按派息日，查找过去12个月的分红记录
#查找有分红的（受派息日约束）；
df = jy.run_query(
query(
jy.LC_Dividend.InnerCode, #证券内部编码
jy.LC_Dividend.
CashDiviRMB, #派现(含税/人民币元)（***说明文档没说明分红单位是每股，还是每十股）
jy.LC_Dividend.DiviBaseBeforeChange, #变更前分红股本基数(股)（***又是不提单位）
).filter(
jy.LC_Dividend.ToAccountDate <= end_date,
jy.LC_Dividend.ToAccountDate >= start_date,
jy.LC_Dividend.IfDividend == 1 #不考虑特殊分红和其他分红
)).fillna(value=0, method=None, axis=0)
df = df[df.InnerCode.isin(list_InnerCode)]
df = df.reset_index(drop=True)
df.index = df['InnerCode']
df_code = df_code[df_code.InnerCode.isin(
list(set(list(df['InnerCode']))))]
df = df.drop(['InnerCode'], axis=1)
df_code.index = df_code['InnerCode']
df = df.join([df_code])
df['SecuCode'] = map(normalize_code, list(df['SecuCode']))
# 获取最新股本
q = query(valuation.code, valuation.capitalization)
df2 = get_fundamentals(q).fillna(value=0)
df2 = df2[df2.code.isin(df.SecuCode)]
df2['SecuCode'] = df2['code']
df2 = df2.drop(['code'], axis=1)
# 合并成一个 dataframe
df = df.merge(df2, on='SecuCode')
df.index = list(df['SecuCode'])
df = df.drop(['SecuCode'], axis=1)
# 转换成 float
df['DiviBaseBeforeChange'] = map(float, df['DiviBaseBeforeChange'])
# 计算股份比值
# *** 因为聚源的 DiviBaseBeforeChange 数据全数为 0，也就是没有记录分红时的股份基数，所以暂时默认都是 1.0
# 实际市场里，因为送股/配股转股的缘故，分红时的股份基数，往往会与当下的股份不一致
#df['CAP_RATIO'] = df['DiviBaseBeforeChange'] / (df['capitalization'] * 10000)
df['CAP_RATIO'] = 1.0
df['CashDiviRMB'] = map(float, df['CashDiviRMB'])
# 计算相对于目前股份而言的分红额度
df['CashDiviRMB'] = df['CashDiviRMB'] * df['CAP_RATIO']
df = df.drop(['DiviBaseBeforeChange', 'capitalization', 'CAP_RATIO'],
axis=1)
#接下来这一步是考虑多次分红的股票，因此需要累加股票的多次分红
df = df.groupby(df.index).sum()
#得到当前股价
Price = history(1,
unit='1d',
field='close',
security_list=list(df.index),
df=True,
skip_paused=False,
fq='pre')
Price = Price.T
df['pre_close'] = Price
#计算股息率 = 股息/股票价格，* 10 是因为取到的是每 10 股分红
df['divpercent'] = df['CashDiviRMB'] / (df['pre_close'] * 10)
df = df.drop(['pre_close', 'CashDiviRMB'], axis=1)
df = df[df.divpercent > 0]
#print df
return df
