def process(self):
fo = open('shrink_code', 'w')
oversold = open('oversold', 'w')
week_shrink = open('week_shrink','w')
#raw_data = TS.memchaced_data(ts.get_stock_basics,'get_stock_basics')
raw_data = ts.get_stock_basics()
raw_data['earn_ratio'] = raw_data['esp'] / raw_data['bvps']
for code in raw_data.index:
if code in bad_container:
continue
try:
daydata = ts.get_k_data(code, ktype='D')
ratio = judge_oversold(daydata)
if ratio < -0.45:
oversold.write("{0}\t{1}\n".format(code, ratio))
weekdata = ts.get_k_data(code, ktype='W')
if judge_week_shrinkage(code, 0.02):
week_shrink.write(code + '\n')
if ratio > -0.3:
continue
#if raw_data.ix[code]['earn_ratio'] < 0.05:
# continue
totals = raw_data.ix[code]['totals']
close = np.array(daydata['close'])[-1]
if close < 6.5:
continue
if close * totals > 360:
continue
flag,mean = judge_shrinkage(daydata,0.02)
if ratio > -0.3:
continue
if flag != -1:
fo.write("{0}\t{1}\t{2:.1%}\n".format(flag,code, mean))
except Exception, e:
print e
def download_kline_by_date_range(code, date_start, date_end):
"""
根据日期范围下载股票行情
:param code:股票代码，即6位数字代码，或者指数代码（sh=上证指数 sz=深圳成指 hs300=沪深300指数 sz50=上证50 zxb=中小板 cyb=创业板）
:param date_start:
:param date_end:
:return:
"""
try:
if len(code)==6:
df_qfq = ts.get_k_data(str(code), start=date_start, end=date_end, autype='qfq') # 前复权
df_hfq = ts.get_k_data(str(code), start=date_start, end=date_end, autype='hfq') # 后复权
else:
df_qfq = ts.get_k_data(str(code), start=date_start, end=date_end)
if len(df_qfq)==0 or (len(code)==6 and len(df_hfq)==0):
return pd.DataFrame()
if len(code)==6:
df_qfq['close_hfq'] = df_hfq['close']
else:
df_qfq['close_hfq'] = df_qfq['close'] # 指数后复权=前复权
print(df_qfq.head())
return df_qfq
except Exception as e:
print(str(e))
return pd.DataFrame()
def tang_process(self):
badarea = set(['黑龙江','辽宁','吉林'])
raw_data = self.foundmental_data
month_writer = open('month_health', 'w')
week_writer = open('week_health', 'w')
day_writer = open('day_health', 'w')
for code in raw_data.index:
earn_ratio = raw_data.ix[code]['earn_ratio']
if raw_data.ix[code]['area'] in badarea:
continue
monthdata = ts.get_k_data(code, ktype='M')
weekdata = ts.get_k_data(code, ktype='W')
daydata = ts.get_k_data(code, ktype='D')
if tang_method(monthdata, -0.8, 1.5):
month_writer.write(code + '\t' + str(earn_ratio) + '\n')
if tang_method(weekdata, -0.5, 0.8):
week_writer.write(code + '\t' + str(earn_ratio) + '\n')
if tang_method(daydata,-0.2, 0.6):
day_writer.write(code + '\t' + str(earn_ratio) + '\n')
print code
sys.stdout.flush()
month_writer.close()
week_writer.close()
day_writer.close()
def baseAPI():
# 通过tushare获取股票信息
df = ts.get_k_data('300580', start='2017-01-12', end='2017-05-26')
# 提取收盘价
closed = df['close'].values
# 获取均线的数据，通过timeperiod参数来分别获取 5,10,20 日均线的数据。
ma5 = talib.SMA(closed, timeperiod=5)
ma10 = talib.SMA(closed, timeperiod=10)
ma20 = talib.SMA(closed, timeperiod=20)
# 打印出来每一个数据
print(closed)
print(ma5)
print(ma10)
print(ma20)
# 通过plog函数可以很方便的绘制出每一条均线
plt.plot(closed)
plt.plot(ma5)
plt.plot(ma10)
plt.plot(ma20)
# 添加网格，可有可无，只是让图像好看点
plt.grid()
# 记得加这一句，不然不会显示图像
plt.show()
def ma_type_test():
# MA_Type: 0=SMA, 1=EMA, 2=WMA, 3=DEMA, 4=TEMA, 5=TRIMA, 6=KAMA, 7=MAMA, 8=T3 (Default=SMA)
df = ts.get_k_data('300580', start='2017-01-12', end='2017-05-26')
closed = df['close'].values
sma = talib.MA(closed, timeperiod=10, matype=0)
ema = talib.MA(closed, timeperiod=10, matype=1)
wma = talib.MA(closed, timeperiod=10, matype=2)
dema = talib.MA(closed, timeperiod=10, matype=3)
tema = talib.MA(closed, timeperiod=10, matype=4)
trima = talib.MA(closed, timeperiod=10, matype=5)
kma = talib.MA(closed, timeperiod=10, matype=6)
mama = talib.MA(closed, timeperiod=10, matype=7)
t3 = talib.MA(closed, timeperiod=10, matype=8)
# ouput=talib.MA(closed,timeperiod=5,matype=0)
print(closed)
plt.ylim([0, 40])
plt.plot(sma, 'r--')
plt.plot(ema, 'g-*')
plt.plot(wma)
plt.plot(dema)
plt.plot(tema)
plt.plot(trima)
plt.plot(kma)
plt.plot(mama)
plt.plot(t3)
plt.grid()
plt.text(7, 30, 'BST')
plt.show()
def prepare_stock_data(stock):
stock_data = ts.get_k_data(stock)
stock_data = stock_data.sort_values("date")
stock_data = stock_data.set_index(["date"])
stock_data = calculate_macd(stock_data)
stock_data = calculate_change_aver(stock_data)
return stock_data
def QA_fetch_get_stock_day(name, startDate='', endDate='', if_fq='01', type_='json'):
if (len(name) != 6):
name = str(name)[0:6]
if str(if_fq) in ['qfq', '01']:
if_fq = 'qfq'
elif str(if_fq) in ['hfq', '02']:
if_fq = 'hfq'
elif str(if_fq) in ['bfq', '00']:
if_fq = 'bfq'
else:
QA_util_log_info('wrong with fq_factor! using qfq')
if_fq = 'qfq'
data = QATs.get_k_data(str(name), startDate, endDate,
ktype='D', autype=if_fq, retry_count=200, pause=0.005).sort_index()
data['date_stamp'] = data['date'].apply(lambda x: QA_util_date_stamp(x))
data['fqtype'] = if_fq
if type_ in ['json']:
data_json = QA_util_to_json_from_pandas(data)
return data_json
elif type_ in ['pd', 'pandas', 'p']:
data['date'] = pd.to_datetime(data['date'])
data = data.set_index('date',drop=False)
data['date'] = data['date'].apply(lambda x: str(x)[0:10])
return data
def _load_data(self, code, dt_start, dt_end):
dts = _process_ts_dt(dt_start)
dte = _process_ts_dt(dt_end)
data = ts.get_k_data(code, start=dts, end=dte)
data.set_index('date',drop=True,inplace=True)
data.index.names = ['datetime']
return data.iloc[::-1]
def k_15min_one_stock(_stock_id, _db):
# log_info("k_15min_one_stock begin")
# get from web(by tushare)
begin = get_micro_second()
try:
df = ts.get_k_data(_stock_id, ktype='15', autype='qfq')
except Exception:
log_error("warn:error: %s get_k_data exception!", _stock_id)
return -4
# calc cost time
log_info("get_k_data [%s] costs %d ms", _stock_id, (get_micro_second()-begin)/1000)
if df is None :
log_error("warn: stock %s is None, next", _stock_id)
return -1
if df.empty:
log_error("warn: stock %s is empty, next", _stock_id)
return -2
df.sort_index(ascending=False, inplace=True)
begin = get_micro_second()
k_15min_one_to_db(_stock_id, df, _db)
log_info("one_to_db costs %d ms", (get_micro_second() - begin)/1000)
# log_info("function k_15min_one_stock end")
return
def getk_handler(args,parser):
for s in args.subnargs:
res = ts.get_k_data(s,ktype=args.ktype,autype=args.autype,index=args.indexmode,start=args.startdate,end=args.enddate)
print('%s for %s %s:%s'%(s,args.ktype,args.startdate,args.enddate))
print('%s'%(res))
sys.exit(0)
return
def calc_open_by_percent(self,code):
cont=100000000
#total_vol=self.bases[self.bases['code']==code]['totals'].values[0]
acutal_vol=self.bases[self.bases['code']==code]['outstanding'].values[0]
all_vol= acutal_vol*cont
df1=ts.get_k_data(code)
i=1
while 1:
s=df1.ix[i]
if s['high']!=s['low']:
#date=s['date']
break
i=i+1
j=i-1
date_end=df1.ix[j]['date']
date_start=df1.ix[0]['date']
df3=df1[(df1['date']>=date_start) & (df1['date']<=date_end)]
v_total_break=df3['volume'].sum()
l=len(df3)
print(l)
print(v_total_break)
rate=v_total_break*100*100.00/all_vol #手和股 注意
print(round(rate,6))
return rate,l
def relation(self):
sh_index=ts.get_k_data('000001',index=True,start='2012-01-01')
sh=sh_index['close'].values
print sh
vol_close=sh_index.corr()
print vol_close
'''
def macd(self):
# df=self.fetch_new_ipo(writeable=True)
# all_code=df['code'].values
# all_code=self.get_all_code()
# print all_code
result = []
for each_code in self.all_code:
print each_code
try:
df_x = ts.get_k_data(code=each_code, start='2017-03-01')
# 只找最近一个月的，所以no item的是停牌。
except:
print "Can't get k_data"
continue
if len(df_x) < 11:
# return
print "no item"
continue
ma5 = df_x['close'][-5:].mean()
ma10 = df_x['close'][-10:].mean()
if ma5 > ma10:
# print "m5>m10: ",each_code," ",self.base[self.base['code']==each_code]['name'].values[0], "ma5: ",ma5,' m10: ',ma10
temp = [each_code, self.base[self.base['code'] == each_code]['name'].values[0]]
print temp
result.append(temp)
print result
print "Done"
return result
def calc_open_day(self,code):
cont=100000000
#total_vol=self.bases[self.bases['code']==code]['totals'].values[0]
acutal_vol=self.bases[self.bases['code']==code]['outstanding'].values[0]
all_vol= acutal_vol*cont
#df= ts.get_hist_data(code)
df1=ts.get_k_data(code)
if len(df1)<3:
return None
#print(df1.info())
#df1=df.reset_index()
#print(df1)
start=df1['date'].values[0]
print('Start day:', start)
df2= df1[(df1['close']==df1['low']) & (df1['high']==df1['low'])]
print(self.bases[self.bases['code']==code]['name'].values[0])
end=df2['date'].values[-1]
print("Break day" , end)
df3=df1[(df1['date']>=start) & (df1['date']<=end)]
v_total_break=df3['volume'].sum()
l=len(df3)
print(l)
print(v_total_break)
rate=v_total_break*100*100.00/all_vol #手和股 注意
print(round(rate,6))
return rate,l
def _getFundDaysFromTuShare(self, code, startDate, endDate, fields, name=None):
"""
从tushare获取基金（ETF）日线数据。
# !!!TuShare没有提供换手率，复权因子和成交额，所以只能假设。
# 策略针对ETF的，需要注意。
"""
tuShareCode = code[:-3]
try:
# 以无复权方式从腾讯获取OHCLV，成交量是手（整数化过）
# 此接口支持ETF日线数据
df = ts.get_k_data(tuShareCode, startDate, endDate, autype=None).sort_index()
except Exception as ex:
self._info.print("从TuShare获取{}({})日线数据[{}, {}]失败: {}".format(code, name, startDate, endDate, ex), DyLogData.error)
return None
df['volume'] = df['volume']*100
# !!!TuShare没有提供换手率，复权因子和成交额，所以只能假设。
# 策略针对ETF的，需要注意。
df['turnover'] = 0
df['factor'] = 1
df['amount'] = 0
df.index.name = None
# change to Wind's indicators
df.rename(columns={'date': 'datetime', 'amount': 'amt', 'turnover': 'turn', 'factor': 'adjfactor'}, inplace=True)
# 把日期的HH:MM:SS转成 00:00:00
df['datetime'] = pd.to_datetime(df['datetime'], format='%Y-%m-%d')
# select according @fields
df = df[['datetime'] + fields]
return df
def gen_zz800_market_index_price_and_ratio(start, end):
zz800_market_ratio = ts.get_k_data('000906', start=start, end=end, index=True)
zz800_market_ratio.columns = zz800_market_ratio.columns.str.upper()
file_path = '/data/gen_data/' + start + '_' + end + '_zz800_market_ratio' + '.csv'
zz800_market_ratio['RATIO'] = zz800_market_ratio['CLOSE'].rolling(window=2).apply(lambda x: (x[1] - x[0]) / x[0])
zz800_market_ratio[['DATE', 'CLOSE', 'RATIO']].to_csv(config.rootPath + file_path, index=False, header=True)
def process(self):
week_shrink = open('week_shrink','w')
raw_data = ts.get_stock_basics()
raw_data['earn_ratio'] = raw_data['esp'] / raw_data['bvps']
container = defaultdict(list)
for code in raw_data.index:
if code in bad_container:
continue
weekdata = ts.get_k_data(code, ktype='W')
try:
info = compute_foundation_info(code, weekdata, 36)
if info['ratio'] > 0.2:
continue
if not info['cross_flag'] and not info['nearcross_flag']:
continue
if info['macdmean'] > 0.3:
continue
assment = info['close'] * raw_data.ix[code]['totals']
if info['week_shrink'] < 2 or info['close'] < 6 or assment > 1800:
continue
for feature in features:
feature_value = info[feature]
container[feature].append(feature_value)
except Exception,e:
print "ERROR:{0}".format(code)
def k_day_one_stock(_stock_id, _db):
# log_info("k_day_one_stock begin")
# get max-date from table, as start date
max_date = k_day_get_max_date(_stock_id, _db)
end_date = get_date_by(0)
# log_debug("[%s, %s]", max_date, end_date)
# qfq
if max_date is None:
start_date = '2016-01-01'
start_date = '2017-07-01'
start_date = '2014-01-01'
start_date = '2017-01-03'
log_debug("it's first time: [%s]", _stock_id)
else:
start_date = str(max_date)
# log_debug("a old friend: [%s]", _stock_id)
log_debug("[%s, %s]", start_date, end_date)
# get from web(by tushare)
begin = get_micro_second()
try:
df = ts.get_k_data(_stock_id, autype='qfq', start=start_date, end=end_date) # reject by tencent since 2018-7-16
# df = ts.get_h_data(_stock_id, autype='qfq', start=start_date, end=end_date, retry_count=5, pause=6)
# df = ts.get_h_data(_stock_id, start='2016-08-20', end='2016-10-30')
# df = ts.get_h_data(_stock_id, autype='qfq')
except Exception:
log_error("warn:error: %s get_k_data exception!", _stock_id)
return -4
# calc cost time
log_info("get_k_data [%s] costs %d us", _stock_id, get_micro_second()-begin)
if df is None :
log_error("warn: stock %s is None, next", _stock_id)
return -1
if df.empty:
log_error("warn: stock %s is empty, next", _stock_id)
return -2
df.sort_index(ascending=True, inplace=True)
# df = df.sort_index(ascending=True)
# df = df.reindex(index=range(0, len(df)))
# log_debug("df: \n%s", df)
begin = get_micro_second()
k_day_one_to_db(_stock_id, df, max_date, _db)
log_info("one_to_db costs %d us", get_micro_second() - begin)
# log_info("function k_day_one_stock end")
return
def profit(code,start,end):
# conn=ts.get_apis()
try:
# df=ts.bar(code,conn=conn,start_date=start,end_date=end)
df=ts.get_k_data(code,start=start,end=end)
except Exception,e:
print e
return None
def price_moment():
df = ts.get_k_data('300580', start='2017-01-12', end='2017-05-26')
closed = df['close'].values
ouput = talib.MOM(closed, timeperiod=5)
print(closed)
print(ouput)
plt.plot(ouput)
plt.grid()
plt.show()
def make_k_data():
df = ts.get_k_data('600000', start='2015-01-01')
datastr = ''
for idx in df.index:
rowstr = '[\'%s\',%s,%s,%s,%s]' % (df.ix[idx]['date'], df.ix[idx]['open'],
df.ix[idx]['close'], df.ix[idx]['low'],
df.ix[idx]['high'])
datastr += rowstr + ',\n'
print(datastr)
def return_risk(stocks,startdate='2005-01-01'):
close=pd.DataFrame()
for stock in stocks.values():
close[stock]=ts.get_k_data(stock,ktype='D',
autype='qfq', start=startdate)['close']
tech_rets = close.pct_change()[1:]
rets = tech_rets.dropna()
ret_mean=rets.mean()*100
ret_std=rets.std()*100
return ret_mean,ret_std
def amplitude_compute(code):
'''
计算一些指数的日均振幅信息
'''
rawdata = ts.get_k_data(code,start='2011-10-01', end='2014-05-31')
rawdata['delta'] = rawdata['high'] - rawdata['low']
length = len(rawdata.index) - 1
base = rawdata[rawdata.index < length]['close']
delta = rawdata[rawdata.index > 0]['delta']
ave_percent = np.mean(delta / base)
return ave_percent
def profit(self,start,end,code):
start_price=ts.get_k_data(start=start,end=start,code=code)
if len(start_price)==0:
print("Not open for start day")
#采用日期操作，往前移动上一个开盘日.
s_price=start_price['close'].values[0]
print("Start price: ",s_price)
end_price=ts.get_k_data(start=end,end=end,code=code)
if len(end_price)==0:
print("Not open for end day")
e_price=end_price['close'].values[0]
print("End price: ",e_price)
earn_profit=(e_price-s_price)/s_price*100
print("Profit: ",)
print(round(earn_profit,2))
def download_stock_history(data):
symbol = data["symbol"]
is_index = data["is_index"]
df = ts.get_k_data(symbol, index=is_index, start="2000-01-01")
if df.empty:
return
path = os.path.join(stock_dir, symbol)
exsymbol = stock.utils.symbol_util.symbol_to_exsymbol(symbol, is_index)
redis_store = get_store("redis_store")
file_store = get_store("file_store")
redis_store.save(exsymbol, df)
file_store.save(exsymbol, df)
def computeYield(code, market, date):
ds = ts.get_k_data(code, date)
hs300 = ts.get_hist_data(market, date)
close = ds['close'].values[:len(ds)]
hs300_close = hs300['close'].values[:len(ds)]
ri = yieldOf(close)
rm = yieldOf(hs300_close)
# print("股票" + code + "收益率", ri)
# print("市场收益率", rm)
covi = np.cov(ri, rm)
print(covi)
varm = np.var(rm)
# print("协方差", covi)
csvwriter.writerow([code, covi[0][1] / varm])
def data_recieve(name, start, data_type='normal', ktype='D'):
if data_type not in {'normal', 'carg', 'grow'}:
raise ValueError('Invalid data_type "%s"' % data_type)
data = ts.get_k_data(name, start=start, ktype=ktype)
data_var = data.drop(data.columns, axis=1)
date = data.pop('date').values
code = data.pop('code')
#print('获得原始数据：\t %s' % code[0])
print('起始时间：\t %s' % str(date[0]))
print('结束时间：\t %s' % str(date[-1]))
print('数据个数：\t %s\n' % len(date))
print data.head(5)
if data_type == 'normal':
for i in data.columns:
data_var[i] = data[i]
if data_type == 'carg':
for i in data.columns:
data_var[i] = (data[i] - data[i].shift(1)) / data[i].shift(1)
if data_type == 'grow':
for i in data.columns:
data_var[i] = data[i] / data[i].shift(1)
open = data_var.open.values
high = data_var.high.values
close = data_var.close.values
low = data_var.low.values
volume = data_var.volume.values
new_data = data.drop(data.columns, axis=1)
new_data.index = date
new_data['open'] = open
new_data['high'] = high
new_data['close'] = close
new_data['low'] = low
new_data['volume'] = volume
new_data = new_data.dropna()
print('\n查看最新数据：\n')
print new_data.head(5)
return new_data
def check_k_data():
each_code = '300333'
# 如果当天还没收盘，就获取昨天的收盘
df_x = ts.get_k_data(code=each_code, start='2017-03-01')
print(df_x)
if len(df_x) < 11:
print("Error")
exit()
print(df_x)
ma5 = df_x['close'][-5:].mean()
ma10 = df_x['close'][-10:].mean()
print(ma5)
print(ma10)
print(df_x['volume'])
def work_one(_stock_id, _row, _db):
rs = False
content = ""
begin = get_micro_second()
try:
df = ts.get_k_data(_stock_id, ktype='15', autype='qfq')
except Exception:
log_error("warn:error: %s get_k_data exception!", _stock_id)
return -4
# calc cost time
log_info("get_k_data [%s](%d) costs %d us", _stock_id, len(df), get_micro_second()-begin)
if df is None :
log_error("warn: stock %s is None, next", _stock_id)
return -1
if df.empty:
log_error("warn: stock %s is empty, next", _stock_id)
return -2
rt_u15m_format_ref(_stock_id, df)
# log_debug("+++++++++++++++++\n%s", df)
rv = rt_u15m_case1(_stock_id, df, _db)
if rv == 0:
rs = True
log_info("nice1: %s single pivot", _stock_id)
return rs
log_debug("-----------------------------------------")
"""
rv = rt_u15m_case2(_stock_id, df, _db)
if rv == 0:
rs = True
log_info("nice2: %s six-cross", _stock_id)
return rs
log_debug("-----------------------------------------")
"""
return rs
def downloadEquityDailyBarts(self, symbol):
"""
下载股票的日行情，symbol是股票代码
"""
print u'开始下载%s日行情' %symbol
# 查询数据库中已有数据的最后日期
cl = self.dbClient[DAILY_DB_NAME][symbol]
cx = cl.find(sort=[('datetime', pymongo.DESCENDING)])
if cx.count():
last = cx[0]
else:
last = ''
# 开始下载数据
import tushare as ts
if last:
start = last['date'][:4]+'-'+last['date'][4:6]+'-'+last['date'][6:]
data = ts.get_k_data(symbol,start)
if not data.empty:
# 创建datetime索引
self.dbClient[DAILY_DB_NAME][symbol].ensure_index([('datetime', pymongo.ASCENDING)],
unique=True)
for index, d in data.iterrows():
bar = VtBarData()
bar.vtSymbol = symbol
bar.symbol = symbol
try:
bar.open = d.get('open')
bar.high = d.get('high')
bar.low = d.get('low')
bar.close = d.get('close')
bar.date = d.get('date').replace('-', '')
bar.time = ''
bar.datetime = datetime.strptime(bar.date, '%Y%m%d')
bar.volume = d.get('volume')
except KeyError:
print d
flt = {'datetime': bar.datetime}
self.dbClient[DAILY_DB_NAME][symbol].update_one(flt, {'$set':bar.__dict__}, upsert=True)
print u'%s下载完成' %symbol
else:
print u'找不到合约%s' %symbol
import numpy as np
import pandas as pd
from pandas import Series, DataFrame
import json
import time
import sys
from sqlalchemy import create_engine
with open('/home/code/shares/block/industry.json', 'r') as f:
datas = json.load(f)
startTime = '2001-01-01'
endTime = time.strftime('%Y-%m-%d', time.localtime(time.time()))
engine = create_engine('mysql://*****:*****@127.0.0.1/shares?charset=utf8')
for data in datas:
code = data.get('code')
singles = ts.get_k_data(code, start=startTime, end=endTime)
if singles['open'] + singles['close'] == 23.393:
singles['one_point'] = 1
elif singles['open'] + singles['close'] == 23.915:
singles['one_point'] = 2
else:
singles['one_point'] = 3
print(singles)
sys.exit()
# for single in singles['date']
# print(single)
# die()
# singles.to_sql(code+'_history', engine, if_exists='append')
def download_d(self, code, start='2000-01-01', end='2023-01-01'):
"""docstring for download"""
print(f"Downloading {code}...")
df = ts.get_k_data(code, start=start, end=end)
df.to_csv(f"{self.DATA_FOLDER}/{code}.csv", index=False)
warnings.simplefilter('ignore') # 忽略警告信息
# 解决中文显示问题
plt.rcParams['font.sans-serif'] = ['KaiTi'] # 指定默认字体
plt.rcParams['axes.unicode_minus'] = False # 解决保存图像是负号'-'显示为方块的问题
plt.style.use('seaborn')
mpl.rcParams['font.family'] = 'serif'
np.sign(-0.02) # 重要函数
date = '2014-01-20'
type(date)
data = pd.to_datetime('2014-01-20')
type(data) # 转换成为时间数据
# 数据准备 & 回测准备
data = ts.get_k_data('hs300', start='2010-01-01',
end='2017-06-30')[['date', 'close']]
data.rename(columns={'close': 'price'}, inplace=True)
data.set_index('date', inplace=True)
# 策略开发思路
data['returns'] = np.log(data['price'] / data['price'].shift(1))
data['position'] = np.sign(data['returns'])
# 关键语句,np.sign()很多地方用到；向量化
data['strategy'] = data['position'].shift(1) * data['returns']
# 计算Momentum策略收益；避免未来函数
# 策略可视化
data[['returns', 'strategy'
]].cumsum().apply(np.exp).plot(figsize=(10, 6)) # 计算出策略的最终的累计收益
plt.show()
def generateline(stocknumber,Type,startdate,enddate,interval):
startdata = startdate.encode("ascii").replace("/","-").replace("\n","") #convert to tushare readable date
enddata = enddate.encode("ascii").replace("/","-").replace("\n","")
#print startdata
#print enddata
current_time = time.strftime("%Y/%m/%d")
if Type == "分笔".decode("utf-8"):
if startdate!=current_time:
array = ts.get_tick_data(stocknumber, date = startdata)#分笔
if array is None:
return
array = array.sort_values("time")
date = array["time"].tolist()
amount = array["amount"].tolist()
atype = array["type"].tolist()
price = array["price"].tolist()
flag = ["bar" for i in date]
for idx,val in enumerate(atype):#if卖盘，交易变成负数
if val == "卖盘":
amount[idx] = -amount[idx]
if val == "中性盘":#if中性盘，则忽略. Might have a problem with this part??
amount[idx] = 0
returnarray = zip(date,amount,flag,price)
return returnarray
else:
array = ts.get_today_ticks(stocknumber)#Tushare里今日分笔和历史分笔需要分别对待
if array is None:
return
array = array.sort_values("time")
date = array["time"].tolist()
amount = array["amount"].tolist()
atype = array["type"].tolist()
flag = ["bar" for i in date]
for idx, val in enumerate(atype):
if val == "卖盘".decode("utf-8"):
amount[idx] = -amount[idx]
if val == "中性盘".decode("utf-8"):
amount[idx] = 0
returnarray = zip(date, amount, flag)
return returnarray
if Type=="季度饼图".decode("utf-8"):
datestr = startdate.split("/")
thisyear = datestr[0]
df2 = ts.top10_holders(code=stocknumber, gdtype="1")
test = df2[1]["quarter"].tolist()
df_ready = df2[1]
idxlist = []
for idx, val in enumerate(test):
a = val.split("-")
if a[0] == thisyear:
# print a[0],idx
idxlist.append(idx)
thing = df_ready.loc[idxlist]
thing = thing.sort_values(["quarter", "name"])
# print a[0],id
name = thing["name"].tolist()
value = thing["hold"].tolist()
quarter = thing["quarter"].tolist()
namearray = [name[i:i + 10] for i in xrange(0, len(name), 10)]
valuearray = [value[j:j + 10] for j in xrange(0, len(value), 10)]
quarterarray = [quarter[k:k + 10] for k in xrange(0, len(quarter), 10)]
flag = ["pie" for i in namearray]
num = [len(value) for k in namearray]
returnarray = zip(namearray,valuearray,quarterarray,flag,num)
return returnarray
if interval!="qfq" and interval!="hfq":
if interval=="1min" or interval=="5min" or interval=="15min" or interval=="30min" or interval=="60min":
df = ts.get_tick_data(stocknumber, date=startdata)
df.sort_values("time")
a = startdata + " " + df["time"]
df["time"] = a
df["time"] = pd.to_datetime(a)
df = df.set_index("time")
price_df = df["price"].resample(interval).ohlc()
price_df = price_df.dropna()
vols = df["volume"].resample(interval).sum() #relevant data processing algorithm taken from Jimmy, Creator of Tushare
vols = vols.dropna()
vol_df = pd.DataFrame(vols, columns=["volume"])
amounts = df["amount"].resample(interval).sum()
amounts = amounts.dropna()
amount_df = pd.DataFrame(amounts, columns=["amount"])
newdf = price_df.merge(vol_df, left_index=True, right_index=True).merge(amount_df, left_index=True,
right_index=True)
if Type != "Kline":
Type1 = firstletter(Type).encode("ascii")
target = newdf[Type1].tolist()
date = newdf.index.format()
returnarray = zip(date, target)
return returnarray
else:
Date = newdf.index.format()
Open = newdf["open"].tolist()
Close = newdf["close"].tolist()
High = newdf["high"].tolist()
Low = newdf["low"].tolist()
Candlestick = zip(*[Date, Open, Close, Low, High])
return Candlestick
#正常历史k线
if Type == "差价图".decode("utf-8"):
global array300
if array300 is None:
array300 = ts.get_k_data("399300", start=startdata, end=enddata, ktype=interval)
array = ts.get_k_data(stocknumber, start=startdata, end=enddata, ktype=interval, autype='qfq')
if array is None or array300 is None:
return None
target = []
benchmark_list = []
try:
close_list = array["close"].tolist()
date = array["date"].tolist()
date300 = array300["date"].tolist()
close300 = array300["close"].tolist()
if len(close_list) is 0:
return None
data_price_300_map = dict()
benchmark300 = 100 / (close300[0])
benchmark = 100 / (close_list[0])
for index, day in zip(range(0, len(date300)), date300):
data_price_300_map[day] = close300[index]
for index, close in zip(range(1, len(close_list)), close_list):
price300 = data_price_300_map[date[index]] * benchmark300
target.append(round(close * benchmark - price300, 3))
benchmark_list.append(0)
returnarray = zip(date, target, benchmark_list)
if checkValid(returnarray) is False:
return None
return returnarray
except Exception as e:
print e
return None
elif Type == "价格指数".decode("utf-8"):
global array300
if array300 is None:
array300 = ts.get_k_data("399300", start=startdata, end=enddata, ktype=interval)
array = ts.get_k_data(stocknumber, start=startdata, end=enddata, ktype=interval, autype='qfq')
if array is None or array300 is None:
return None
price_list = []
price300_list = []
try :
close_list = array["close"].tolist()
date = array["date"].tolist()
close300_list = array300["close"].tolist()
if len(close_list) is 0:
return None
benchmark300 = 100 / (close300_list[0])
benchmark = 100 / (close_list[0])
for close in close_list:
price_list.append(close * benchmark)
for close300 in close300_list:
price300_list.append(close300 * benchmark300)
returnarray = zip(date, price_list, price300_list)
if checkValid(returnarray) is False:
return None
return returnarray
except Exception as e:
print e
return None
elif Type!="Kline":
array = ts.get_k_data(stocknumber, start=startdata, end=enddata, ktype=interval)
if array is None:
return
Type1 = firstletter(Type).encode("ascii")
target = array[Type1].tolist()
date = array["date"].tolist()
returnarray = zip(date,target)
return returnarray
else:
array = ts.get_k_data(stocknumber, start=startdata, end=enddata, ktype=interval)
if array is None:
return
Date = array["date"].tolist()
Open = array["open"].tolist()
Close = array["close"].tolist()
High = array["high"].tolist()
Low = array["low"].tolist()
Candlestick = zip(*[Date,Open,Close,Low,High])
return Candlestick
else:
if Type!="Kline": # 复权
array = ts.get_h_data(stocknumber, start = startdata, end = enddata, autype= interval)
if array is None:
return
Type1 = firstletter(Type).encode("ascii")
array = array.sort_index()
target = array[Type1].tolist()
date = array.index.format()
returnarray = zip(date, target)
return returnarray
else :
array = ts.get_h_data(stocknumber, start=startdata, end=enddata, autype=interval)
if array is None:
return
array = array.sort_index()
Date = array.index.format()
Open = array["open"].tolist()
Close = array["close"].tolist()
High = array["high"].tolist()
Low = array["low"].tolist()
Candlestick = zip(*[Date, Open, Close, Low, High])
return Candlestick
def show_trend(stock_id, stock_name,
end_date=(datetime.datetime.now() + datetime.timedelta(days=-1)).strftime('%Y%m%d'), working_days=200,
step=10):
end_date = datetime.datetime.strptime(end_date, '%Y%m%d')
start_date = end_date
while get_working_days(start_date, end_date) < working_days:
start_date = start_date + datetime.timedelta(days=-1)
start_date = start_date + datetime.timedelta(days=-(step - 1))
k_dict = {}
k_index = ts.get_k_data(stock_id, start_date.strftime('%Y-%m-%d'), end_date.strftime('%Y-%m-%d'))
for i in range(0, len(k_index)):
k_dict[list(k_index['date'])[i]] = list(k_index['close'])[i]
trend_dict = wssrdb.get_trend(stock_id, start_date, end_date)
date_list = sorted(list(k_dict.keys()) + [item for item in trend_dict.keys() if item not in k_dict.keys()])
k_list = []
trend_list = []
for r in date_list:
if r in k_dict.keys():
k_list.append(k_dict[r])
elif len(k_list) > 0:
k_list.append(k_list[-1])
else:
k_list.append(0.0)
if r in trend_dict.keys():
trend_list.append(trend_dict[r])
else:
trend_list.append(0)
#####################################################
if step > 1: # 大于1天
date_list_2 = []
k_list_2 = []
trend_list_2 = []
k_avg = 0.0
trend_sum = 0
for i in range(0, len(date_list)):
if (i + 1) % step > 0:
k_avg += k_list[i]
trend_sum += trend_list[i]
else:
date_list_2.append(date_list[i])
k_list_2.append(k_avg / step)
trend_list_2.append(trend_sum)
k_avg = 0.0
trend_sum = 0
date_list = date_list_2
k_list = k_list_2
trend_list = trend_list_2
#####################################################
date_list = [d.replace("-", "\n") for d in date_list]
fig, pl1 = pl.subplots()
pl1.plot(date_list, k_list, label=u"收盘价", color="blue", linewidth=2.5)
pl1.set_ylabel(u"收盘价", fontsize=18, color="blue")
for label in pl1.get_yticklabels():
label.set_color("blue")
pl2 = pl1.twinx()
pl2.plot(date_list, trend_list, label=u"研报推荐", color="red", linewidth=2.5)
pl2.set_ylabel(u"研报推荐", fontsize=18, color="red")
for label in pl2.get_yticklabels():
label.set_color("red")
pl2.set_xlabel(u'时间', fontsize=18, color="black")
pl.title(stock_name + "[" + stock_id + "]")
pl.legend()
pl.show()
import tushare as ts
import json
import time
from sqlalchemy import create_engine
todayTime = time.strftime('%Y-%m-%d', time.localtime(time.time()))
single = ts.get_k_data('000026', start='2018-05-25', end='2018-05-25')
engine = create_engine('mysql://*****:*****@127.0.0.1/shares?charset=utf8')
#print(single)
single.to_sql('000026_history', engine, if_exists='append')
import talib as ta
import tushare as ts
import matplotlib.pyplot as plt
from matplotlib import rc
rc('mathtext', default='regular')
import seaborn as sns
sns.set_style('white')
from matplotlib import dates
import matplotlib as mpl
%matplotlib inline
# myfont = mpl.font_manager.FontProperties(
# fname=r"c:\windows\fonts\simsun.ttc", size=14)
plt.rcParams["figure.figsize"] = (20, 10)
dw = ts.get_k_data("600600")
dw = dw[300:]
dw.index = range(len(dw))
close = dw.close.values
dw["rsi"] = ta.RSI(close, timeperiod=14)
#dw[["close","rsi"]].plot()
fig = plt.figure(figsize=(20, 10))
fig.set_tight_layout(True)
ax1 = fig.add_subplot(111)
#fig.bar(dw.index, dw.volume, align='center', width=1.0)
ax1.plot(dw.index, dw.close, '-', color='g')
ax2 = ax1.twinx()
ax2.plot(dw.index, dw.rsi, '-', color='r')
ax2.plot(dw.index, [70] * len(dw), '-', color='r')
ax2.plot(dw.index, [30] * len(dw), '-', color='r')
def _download_data(self, code, index):
start_day_date = date_to_str(self.test_date - timedelta(40))
start_week_date = date_to_str(self.test_date - timedelta(25) * 7)
#取日线行情:
data_day = ts.get_k_data(code,
ktype='d',
autype='qfq',
index=index,
start=start_day_date,
end=self.buy_date_str)
#检查够不够20个交易日
if (data_day.index.size < 20):
data_day = ts.get_k_data(
code,
ktype='d',
autype='qfq',
index=index,
start=date_to_str(self.test_date - timedelta(180)),
end=self.buy_date_str) #考虑有停牌的情况，祖国不够20个，再往前追加3个月
#检查是否仍然空。如果还空，要么停牌超过半年，要么服务器出错。
if (data_day.empty):
print("数据无效：停牌超过半年，或者服务器出错：" + code + ": " + self.buy_date_str)
self.data_valid = False
return None, None
# 检查buydate是否存在
if (data_day.iloc[-1, 0] != self.buy_date_str):
print("数据无效：目标买入日不是交易日：" + code + ": " + self.buy_date_str)
self.data_valid = False
return None, None
if (data_day.index.size < 20):
print("数据无效：股票上市日子不够，计算不出日MA20：" + code + ": " + self.buy_date_str)
self.no_week_ma = True
self.data_valid = False
return None, None
#取周线行情:
data_week = ts.get_k_data(code,
ktype='w',
autype='qfq',
index=index,
start=start_week_date,
end=self.buy_date_str)
#检查够不够20个交易日
if (data_week.index.size < 20):
data_week = ts.get_k_data(code,
ktype='w',
autype='qfq',
index=index,
start=date_to_str(self.test_date -
timedelta(50) * 7),
end=self.buy_date_str)
if (data_week.index.size < 20):
print("数据无效：股票上市日子不够，计算不出周MA20：" + code + ": " + self.buy_date_str)
self.no_week_ma = True
self.data_valid = False
return None, None
#当日price
price = [data_day.iloc[-1, 2]]
#日均值price
price += compute_ma(data_day['close'])
#周均值price
price += compute_ma(data_week['close'])
if (self._check_data(price) is False):
print("数据无效：股价太小" + code + ": " + self.buy_date_str)
self.data_valid = False
return None, None
#当日volume
volume = [data_day.iloc[-1, 5]]
#日均值volume
volume += compute_ma(data_day['volume'])
#周均值volume
volume += compute_ma(data_week['volume'])
if (self._check_data(volume) is False):
print("数据无效：成交量太小" + code + ": " + self.buy_date_str)
self.data_valid = False
return None, None
return price, volume
first, last = lookback_trade_cal(last, LOOKBACKDAYS)
########## Log记录信息##########
logger.info('开始日期: ' + first)
logger.info('结束日期: ' + last)
################################
# 通过tushare获取前lookback_days 个交易日数据,并筛选满足收盘价序列的股票
stockList = totList
# stockList = ['300706']
targetDict = {}
fd = open('./log.txt', 'w')
i = 0
for stock in stockList:
data = ts.get_k_data(code=stock, start=first, end=last)
logger.info("start analysis code=%s" % stock)
records = len(data)
if (records == 0):
continue
logger.info("len=%d" % records)
print("start analysis code=%s" % stock)
#print data
#if len(data) == LOOKBACKDAYS:
c_data = data.close.values
v_data = data.volume.values
open_data = data.open.values
high_data = data.high.values
low_data = data.low.values
#!/usr/bin/python
# coding: UTF-8
import tushare as ts
import pandas as pd
import matplotlib.pyplot as plt
import numpy as np
import talib
df = ts.get_k_data('600600')
df['MA10_rolling'] = pd.rolling(df['close']).mean(10)
close = [float(x) for x in df['close']] # 调用talib计算10日移动平均线的值
df['MA10_talib'] = talib.MA(np.array(close), timeperiod=10)
df.tail(12)
import tushare as ts
df = ts.get_k_data('sz50', start='2010-01-01', end='2017-12-31')
df.to_csv('./test/sz50.csv')
@author: Lu.yipiao
'''
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import tensorflow as tf
import tushare as ts
import pandas as pd
base_path="./"
#——————————————————导入数据——————————————————————
f=open('dataset_1.csv')
get_k_data = ts.get_k_data("000001", start='1990-12-19')
df = pd.DataFrame(get_k_data);
data=np.array(df['close']) #获取最高价序列
df=pd.read_csv(f) #读入股票数据
data=np.array(df['close']) #获取最高价序列
data=data[::-1] #反转，使数据按照日期先后顺序排列
print(data)
#以折线图展示data
# plt.figure()
# plt.plot(data)
# plt.show()
normalize_data=(data-np.mean(data))/np.std(data) #标准化
normalize_data=normalize_data[:,np.newaxis] #增加维度
def getData(code):
df = ts.get_k_data(code=code)
hd = df.head(1)
q.put(hd)
pbar.update(1)
return hd
# -*- coding:utf-8 -*-
import matplotlib as mpl
import tushare as ts
import matplotlib.pyplot as plt
import mpl_finance as mpf
from matplotlib.pylab import date2num
import datetime
import matplotlib.dates as mdates
import numpy as np
wdyx = ts.get_k_data('000001', '2017-01-01')
#wdyx.info()
print(wdyx[:3])
def date_to_num(dates):
num_time = []
for date in dates:
date_time = datetime.datetime.strptime(date, '%Y-%m-%d')
#print("date_time",date_time)
num_date = date2num(date_time)
#print("num_date", num_date)
num_time.append(num_date)
return num_time
# dataframe转换为二维数组
mat_wdyx = wdyx.values
#print(mat_wdyx[:,0])
def dmpStkDaysFromTS(self,
folder: str,
sdate: str = "1990-01-01",
edate: str = "",
codes: list = None,
bNeedAdj: bool = True,
isIndex: bool = False):
'''
从tushare导出日线数据，慢！！！\n
@sdate 开始日期，格式如2020-06-09\n
@edate 结束日期，格式同上\n
@codes 代码列表，为空时自动获取\n
@bNeedAdj 是否除权，默认为True\n
@isIndex 是否指数数据，默认为False\n
'''
if edate == "":
edate = datetime.datetime.now().strftime('%Y-%m-%d')
if not os.path.exists(folder + 'day/'):
create_dirs(folder + 'day/')
if not os.path.exists(folder + 'min5/'):
create_dirs(folder + 'min5/')
if codes is None:
if isIndex:
codes = self.__indice__.keys()
else:
codes = self.__stocks__.keys()
total = len(codes)
count = 0
for code in codes:
count += 1
print("正在拉取%s的日线[%d/%d]……" % (code, count, total))
exchg = ""
if code[:2] == "SH":
exchg = "SSE"
else:
exchg = "SZSE"
thisFolder = folder + 'day/' + exchg + "/"
if not os.path.exists(thisFolder):
create_dirs(thisFolder)
thisFolder = folder + 'min5/' + exchg + "/"
if not os.path.exists(thisFolder):
create_dirs(thisFolder)
df_bars = None
try:
# get_h_data貌似也可以
df_bars = ts.get_k_data(code[2:],
start=sdate,
end=edate,
ktype='D',
autype='qfq' if bNeedAdj else None,
index=isIndex)
csvpath = (folder + 'day/%s/%s%s.csv') % (
exchg, code[2:], "Q" if bNeedAdj and not isIndex else "")
isEmpty = is_file_empty(csvpath)
f = open(csvpath, 'a')
if isEmpty:
f.write("date, time, open, high, low, close, volumn\n")
for idx, row in df_bars.iterrows():
f.write(
datetime.datetime.strptime(str(
row["date"]), '%Y-%m-%d').strftime("%Y/%m/%d") +
", ")
f.write("0, ")
f.write(str(row["open"]) + ", ")
f.write(str(row["high"]) + ", ")
f.write(str(row["low"]) + ", ")
f.write(str(row["close"]) + ", ")
f.write(str(row["volume"]) + "\n")
f.close()
except:
print("正在拉取%s的日线异常" % (code))
print("正在拉取%s的5分钟线[%d/%d]……" % (code, count, total))
df_bars = None
try:
df_bars = ts.get_k_data(code[2:],
sdate,
edate,
ktype='5',
autype='qfq' if bNeedAdj else None)
csvpath = (folder + 'min5/%s/%s%s.csv') % (
exchg, code[2:], "Q" if bNeedAdj and not isIndex else "")
f = open(csvpath, 'w')
f.write("date, time, open, high, low, close, volumn\n")
for idx, row in df_bars.iterrows():
curDt = datetime.datetime.strptime(str(row["date"]),
'%Y-%m-%d %H:%M')
f.write(curDt.strftime("%Y/%m/%d") + ", ")
f.write(curDt.strftime("%H:%M") + ", ")
f.write(str(row["open"]) + ", ")
f.write(str(row["high"]) + ", ")
f.write(str(row["low"]) + ", ")
f.write(str(row["close"]) + ", ")
f.write(str(row["volume"]) + "\n")
f.close()
except:
print("正在拉取%s的5分钟线异常" % (code))
def get_data(gid, start=None, end=None):
gdata = tushare.get_k_data(gid, start, end)
return gdata
from scipy import stats
import statsmodels.api as sm
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import arch
import tushare as ts
IndexData = ts.get_k_data(code='sh', start='2014-01-01', end='2016-08-01');
IndexData.index = pd.to_datetime(IndexData.date);
close = IndexData.close;
rate = (close - close.shift(1)) / close.shift(1)
data = pd.DataFrame();
data['rate'] = rate;
data = data.dropna();
data1 = np.array(data['rate'])
# data['rate'].plot(figsize=(15, 5))
# t = sm.tsa.stattools.adfuller(data1)
# print("p-vaule:", t[1])
# fig = plt.figure(figsize=(20, 5))
# ax1 = fig.add_subplot(111)
# fig = sm.graphics.tsa.plot_acf(data1, lags=20, ax=ax1)
# order = (8, 0)
# model = sm.tsa.ARMA(data1, order).fit()
# print(model)
# at = data1 - model.fittedvalues
# at2 = np.square(at)
# plt.figure(figsize=(10, 6))
# plt.plot(at, label='at')
# plt.legend();
'''当前交易所有股票的行情数据'''
realTimeIndex = ts.get_index()
'''当前大盘的行情数据'''
realtime_tran_data = ts.get_realtime_quotes(target_stock_code)
'''实时分笔数据'''
#ts.get_realtime_quotes(['600848','000980','000981']) #symbols from a list
#ts.get_realtime_quotes('sh')#上证指数
#ts.get_realtime_quotes(['sh','sz','hs300','sz50','zxb','cyb'])#上证指数 深圳成指 沪深300指数 上证50 中小板 创业板
#ts.get_realtime_quotes(['sh','600848'])#或者混搭
realday_deal_data = ts.get_today_ticks(target_stock_code)
'''当日历史分笔'''
big_deal_data = ts.get_sina_dd(target_stock_code, date=transaction_day)
'''获取大单交易数据，默认为大于等于400手'''
#df = ts.get_sina_dd('600848', date='2015-12-24', vol=500) #指定大于等于500手的数据
M_recent_k_data = ts.get_k_data(target_stock_code, ktype='M')
'''最近月k线'''
W_recent_k_data = ts.get_k_data(target_stock_code, ktype='W')
'''最近周k线'''
D_recent_k_data = ts.get_k_data(target_stock_code, ktype='D')
'''最近日k线'''
recent_k60_data = ts.get_k_data(target_stock_code, ktype='60')
'''最近小时线'''
recent_k30_data = ts.get_k_data(target_stock_code, ktype='30')
'''最近30min线'''
recent_k15_data = ts.get_k_data(target_stock_code, ktype='15')
'''最近15min线'''
recent_k5_data = ts.get_k_data(target_stock_code, ktype='5')
'''最近5min线'''
#==============================================================================指数成分==============================================================================
'''获取指数成分'''
# encoding=utf_8
"""
本脚本用于研究stk价格在一定时期内所处的位置
"""
import tushare as ts
stk = ts.get_k_data('300183', start='2012-01-01').reset_index()
win_size = 180
for idx in stk.index:
price_now = stk.loc[idx, 'close']
se = stk.loc[idx - win_size:idx, 'close']
price_expensive = list(filter(lambda x: x < price_now, se))
stk.loc[idx,
'price_scale'] = len(price_expensive) / len(se) * 100.0 # 越高越好
stk.loc[win_size:, :].plot('date', ['close', 'price_scale'], subplots=True)
end = 0
# stk.rooling()
codes = ['000001.SZ', '603999.SH', '600000.SH'] # or set() -> list()
inital_cash = 1e5
begt = dt.datetime(2013, 1, 1)
endt = dt.datetime(2018, 1, 1)
cerebro = bt.Cerebro()
# cerebro.addsizer(bt.sizers.FixedSize, stake=1000) # 默认买入一手
cerebro.broker.setcash(inital_cash) # 初始资金
cerebro.addstrategy(ManmStrategy, codes=codes)
# Add Analyzer
cerebro.addanalyzer(bt.analyzers.Returns, _name='returns')
cerebro.addanalyzer(bt.analyzers.DrawDown, _name='drawdown')
# Feed Data
for each, code in enumerate(codes):
tohlcva = ts.get_k_data(code.split('.')[0],
start=str(begt.date()),
end=str(endt.date()),
autype='qfq')
tohlcva.drop('code', axis=1, inplace=True)
tohlcva['date'] = pd.to_datetime(tohlcva['date'])
tohlcva.set_index('date', inplace=True)
tohlcva = tohlcva[['open', 'high', 'low', 'close', 'volume']]
tohlcva['openinterest'] = 0.0
# print(tohlcva.head(10))
data = bt.feeds.PandasData(dataname=tohlcva,
fromdate=begt,
todate=endt)
cerebro.adddata(data, name=code)
cerebro.run()
cerebro.plot()
def predict_tomorrow(stk_code,
label,
N_STEPS=N_STEPS,
feature_cols=feature_cols,
HIDDEN_SIZE=HIDDEN_SIZE,
NUM_LAYERS=NUM_LAYERS):
"""
:param stk_code: 例子 '300508'
:param label: 例子 'high'
:param N_STEPS:
:param feature_cols:
:param HIDDEN_SIZE:
:param NUM_LAYERS:
:return:
"""
""" ---------------------- 读取json中存储的极值 ---------------------- """
with open(rootPath + '\LSTM\AboutLSTM\stk_max_min.json', 'r') as f:
max_min_info = json.load(f)
""" ---------------------- 获取实时数据 ---------------------- """
data_now = ts.get_k_data(stk_code)[-(N_STEPS + 30):]
# 增加M9 Rank
data_now['m9'] = data_now['close'].rolling(window=9).mean()
data_now['diff_m9'] = data_now.apply(lambda x:
(x['close'] - x['m9']) / x['close'],
axis=1)
data_now['rank'] = data_now.apply(lambda x: relativeRank(
max_min_info[stk_code]['m9_history'], x['diff_m9']),
axis=1)
# rootPath = 'C:/Users\paul\Desktop\软件代码\Git-Clone'
for c in ['close', 'high', 'low', 'open']:
data_now[c] = (data_now[c].values - max_min_info[stk_code]['p_min']
) / (max_min_info[stk_code]['p_max'] -
max_min_info[stk_code]['p_min'])
data_now['volume'] = (
data_now['volume'].values - max_min_info[stk_code]['v_min']) / (
max_min_info[stk_code]['v_max'] - max_min_info[stk_code]['v_min'])
# 进行归一化
input_normal = data_now.loc[:, feature_cols].tail(20).values
tf.reset_default_graph()
""" ---------------------- 创建模型 ---------------------- """
predictions, loss, train_op, X, y = lstm_model(n_steps=N_STEPS,
n_inputs=len(feature_cols),
HIDDEN_SIZE=HIDDEN_SIZE,
NUM_LAYERS=NUM_LAYERS)
# 创建保存器用于模型
saver = tf.train.Saver()
# 初始化
sess = tf.Session()
model_name = stk_code + '_' + label
model_dir = rootPath + '\LSTM\AboutLSTM\modelDir/'
if os.path.exists(model_dir + model_name + '/' + model_name +
'.ckpt.meta'):
saver = tf.train.import_meta_graph(model_dir + model_name + '/' +
model_name + '.ckpt.meta')
saver.restore(sess,
tf.train.latest_checkpoint(model_dir + model_name + '/'))
# graph = tf.get_default_graph()
# 防报错
tf.reset_default_graph()
r_rela = sess.run([predictions], feed_dict={X:
[input_normal]})[0][0][0]
return max_min_info[stk_code]['p_min'] + (
max_min_info[stk_code]['p_max'] -
max_min_info[stk_code]['p_min']) * r_rela
else:
print('加载模型' + model_name + '失败！')
return -1
def get(code):
df = ts.get_k_data(code, start='2017-01-01')
result.append(df)
print(len(result), code)
continue
elif float(stock_info.loc[stock_info.index==code].perundp) < 0:
continue
elif float(stock_info.loc[stock_info.index==code].rev) < -20:
continue
elif float(stock_info.loc[stock_info.index==code].profit) < -20:
continue
# elif float(stock_info.loc[stock_info.index==code].gpr) < 15:
# continue
# elif float(stock_info.loc[stock_info.index==code].npr) < 0:
# continue
try:
if TEST=='True':
stock=ts.get_k_data(code, start=Tstart, end=Tend)
else:
stock=ts.get_k_data(code)
except:
time.sleep(1)
if TEST=='True':
stock=ts.get_k_data(code, start=Tstart, end=Tend)
else:
stock=ts.get_k_data(code)
if len(stock) < 250:
continue
stock.index=stock.iloc[:,0]
stock.index=pd.to_datetime(stock.index,format='%Y-%m-%d')
stock=stock.iloc[:,1:-1]
import tushare as ts
import talib
from pylab import *
from DataSource.Data_Sub import my_pro_bar
from Experiment.Constraint.Constraint import calBSReseau
from SDK.MyTimeOPT import minus_date_str, get_current_date_str
from HuiCe.Sub import bs_opt, plot_op_result
if __name__ == '__main__':
# 登录datasource
from DataSource.auth_info import *
stk_code = '000001'
df = ts.get_k_data(stk_code, start='2017-01-12', end='2019-05-26')
record_info = {
'floor_last': 0,
'money_remain': 20000,
'amount_remain': 1500,
'M_last': -1, # 用以记录上次的均线值，在反向操作中（本次操作与上次不同的情况）使用上次均值！
'BS_last': 'init', # 上次是买还是卖 "buy" "false" "init"
'price_last': df.head(1)['close'].values[0], # 上次价格
'BS_trend_now': 'init',
'BS_real': 'NO_OPT', # 本次实际操作
'Price_now': 12,
'last_opt_date': '2018-12-15',
'time_span_from_last': 1, # 当前距离上次操作的时间间隔
'B_continue': 1,
'S_continue': 1,
import matplotlib.pyplot as plt
import tushare as ts
import futuquant as ft
import time
datas = ts.get_k_data('hs300', ktype='D', start='2007-01-01', end='2018-07-01')
#datas = datas[['code','close']].sort_index()
#把列变成时间
datas['date'] = pd.to_datetime(datas['date'])
#把时间对角例,变成索引
datas = datas.set_index('date')
datas['无风险利率'] = (4.0 / 100 + 1)**(1.0 / 250) - 1
datas['无风险收益净值'] = (datas['无风险利率'] + 1).cumprod()
#取每个月的第一个交易日
by_month = datas.resample('M', kind='period').first()
trade_log = pd.DataFrame(index=by_month.index)
trade_log['基金净值'] = by_month['close'] / 1000
#每月月初投入1000元
trade_log['投入资金'] = 1000
#购买份额
trade_log['基金份额'] = trade_log['投入资金'] / trade_log['基金净值']
trade_log['总基金份额'] = trade_log['基金份额'].cumsum()
trade_log['累计投入资金'] = trade_log['投入资金'].cumsum()
trade_log['理财份额'] = trade_log['投入资金'] / by_month['无风险收益净值']
trade_log['理财总份额'] = trade_log['理财份额'].cumsum()
#重新采样转换为日历日
tem = trade_log.resample('D').ffill()
datas = datas.to_period('D')
#计算每个交易日的基金净值和资产
from sqlalchemy import create_engine
from sqlalchemy.types import VARCHAR
import pymysql
pymysql.install_as_MySQLdb()
# 创建数据连接
conn = create_engine('mysql+mysqldb://root:111111@localhost:3306/stock?charset=utf8')
# 股票代码列表
# stock_list = ['002218', '600206', '300349', '002493', '600600', '600667', '603225', '600508', '000729', '002079', '000633']
stock_list = ['600035']
# 定义get_k_data接口返回数据的字段类型
k_data_dtype = {
'date': VARCHAR(64),
'code': VARCHAR(64)
}
# 循环获取股票日K线数据
for stock in stock_list:
df = ts.get_k_data(code=stock, ktype='D', autype='qfq', start='1990-01-01')
table = 'kd_d_' + stock
pd.io.sql.to_sql(frame=df, name=table, con=conn, if_exists='replace', dtype=k_data_dtype)
# 循环获取股票周K线数据
for stock in stock_list:
df = df = ts.get_k_data(code=stock, ktype='W', autype='qfq', start='1990-01-01')
table = 'kd_w_' + stock
pd.io.sql.to_sql(frame=df, name=table, con=conn, if_exists='replace', dtype=k_data_dtype)
def fenxi(request):
if request.method == "GET":
dict = {
"title": '用户注册'
}
return render(request, 'register.html', context=dict)
elif request.method == "POST":
daima = request.POST.get("daima")
if daima == '':
daima = '600519'
dict = {
'daima': daima
}
listX = []
listy = []
data = ts.get_k_data(code=daima, ktype='15')
data.index = data['date']
data.pop('date')
# 基于n日均线计算boll
n = 26
boll = create_boll(data, n)
# 涨跌幅度
boll['zdfu'] = boll['close'].pct_change()
# 涨跌标签：1，-1
boll['zd'] = np.sign(boll['zdfu'])
boll['fen'] = boll['close'] - boll['bm']
# 去掉含空值行
con = boll.notnull().all(axis=1)
boll = boll[con]
lmc = pd.DataFrame(index=boll.index)
lmc['zd'] = boll['zd']
lmc['bm'] = boll['bm']
lmc['close'] = boll['close']
lmc['fengy'] = boll[['fen']].apply(guiyi)
lmc['liang'] = boll['volume']
lmc['liangfu'] = (boll['fen']) * lmc['liang']
lmc['liangfugy'] = lmc[['liangfu']].apply(guiyi)
# 量脉冲限制条件
for i in range(0, len(lmc)):
if (boll['close'][i] > boll['bl'][i]) and boll['zdfu'][i] < 0:
lmc['liangfugy'][i] = lmc['fengy'][i]
if (boll['close'][i] < boll['bm'][i]) and boll['zdfu'][i] > 0:
lmc['liangfugy'][i] = 0
figure = plt.figure(figsize=(14, 7))
ax1 = figure.add_subplot(111)
ax1.plot((boll['close'][-200:]), 'k', alpha=0.8)
ax1.plot(lmc['liangfugy'][-200:] * np.min(boll['close'][-200:])/10 + np.min(boll['close'][-200:]), 'b', alpha=0.5)
ax1.set_xticks(range(0, len(boll['close'][-200:]), 20))
ax1.set_xticklabels(boll['close'][-200:].index[::20], rotation=30)
ax1.grid(b=True, linestyle="--")
ax1.set_title("{}".format(daima), fontsize=20, color="black")
figure.savefig("./static/img_gupiao/{}01.jpg".format(daima), facecolor="w", pad_inches=5)
X, y = fun_y(lmc)
listX.append(X)
listy.append(y)
return render(request, 'fenxi.html', context= dict)
return None, None, None
except Exception:
pass
if __name__ == '__main__':
#init_db()
df = ts.get_hs300s()
#df['code'].values
for code in [
'603630',
]:
try:
#print(code)
data = ts.get_k_data(code, start='2017-01-01')
data["mavol30"] = cal_mavol30(data)
data["ma30"] = cal_ma30(data)
data["ma60"] = cal_ma60(data)
data['pre_close'] = data['close'].shift()
#data['p_change'] = ((data['close'] - data['pre_close']) / data['pre_close'])
data = data.tail(30)
#print(data[['date','close','ma30', 'volume', 'mavol30']])
code, date, volume = abnormal_val(data, code)
#code = cal_singal_stock('002661')
if code is None:
continue
# -*- coding: utf-8 -*-
import tushare as ts
window_size = 5 # 过程演示用，待改为55
code = "000001"
temp = ts.get_k_data(code, start="2018-03-01", ktype="D", autype="qfq")
temp.index = temp.pop("date")
df = temp.loc[:, ["high", "close"]]
df["hhv"] = df["high"].rolling(window_size).max()
df["pre_hhv"] = df["hhv"].shift(1)
# df["signals"] = df["close"] > df["pre_hhv"]
df["signals"] = (df["close"].shift(1) <= df["pre_hhv"].shift(1)) & \
(df["close"] > df["pre_hhv"])
print(df)
results = df[df["signals"]]
print(list(results.index))
net_pct_s 小单净占比(%) 小单净占比 = 小单净额 / 成交额
"""
df_flow = get_money_flow('300183.XSHE', '2018-05-01', '2019-06-10')
df_flow_plot = df_flow.loc[:, [x not in ['sec_code'] for x in df_flow.columns.values]]
df_flow_plot['net_total'] = df_flow.apply(lambda x: np.sum([
x['net_amount_main'],
x['net_amount_xl'],
x['net_amount_l'],
x['net_amount_m'],
x['net_amount_s']]), axis=1)
df_flow_plot['net_m'] = df_flow_plot['net_total'].rolling(window=5).sum()
df_flow_plot['date_str'] = df_flow_plot.apply(lambda x: str(x['date'])[:10], axis=1)
df_flow_plot = df_flow_plot.set_index(keys='date_str')
# 下载tushare数据
df_t = ts.get_k_data('300183', start='2018-05-01')
df_t = df_t.set_index(keys='')
end = 0
"""
df_flow_plot.plot('date', list(df_flow.columns.values).remove('date'), subplots=True)
df_flow_plot.plot('date', ['change_pct', 'net_total'], subplots=True)
"""
