def recorg(pl, df_boll):
sign = 0
df = df_boll
if 0: df = pd.DataFrame
#计算参数
#输出原始图片
#closes, boll_up, boll_mid, boll_low = df[['c', 'upper','middle', colname.boll_lower]].to_list()
closes = df['c'].values
boll_up = df[colname.boll_upper].values
boll_mid = pd.Series(df[colname.boll_middle]).values
boll_low = pd.Series(df[colname.boll_lower]).values
#转zz
zz_boll_up = stock.ZigZag(boll_up)
zz_boll_mid = stock.ZigZag(boll_mid)
zz_boll_low = stock.ZigZag(boll_low)
zz_close = stock.ZigZag(closes, percent=5)
zz_close_short = stock.ZigZag(closes, percent=.5)
adx = df[colname.adx].values[-1]
bollw = stock.BOLLW(boll_up, boll_low, closes)
#数值判断
obj = boll_params()
obj.h1 = closes[-1] - boll_mid[-1]
obj.h2 = closes[-1] - boll_low[-1]
boll_y = get_y_boll(closes[-1], boll_low, boll_mid)
techs = calc_property()
y1, y2 = stock.analyzeZZ(zz_close)
techs.close_zz_0 = "%.3f" % y1
techs.close_zz_1 = "%.3f" % y2
y1, y2 = stock.analyzeZZ(zz_boll_low)
techs.boll_low_zz_0 = "%.3f" % y1
techs.boll_low_zz_1 = "%.3f" % y2
techs.boll_w = "%.3f" % bollw[-1]
techs.adx = "%.3f" % adx
techs.boll_x = get_x_boll(closes[-1], boll_low)
techs.boll_y = boll_y
techs.boll_y = "%.2f%%" % (techs.boll_y * 100)
#集中度计算, 使用黄金分割或者是三分之一
techs.jzd = kurtosis.calc_kurtosis(
stock.GuiYiHua(closes[-int(len(closes) * (1 - 0.618)):]))
sign = False
n = 1 # 如果是日线，n=10
#第二次机会
if sign_observation.assemble(0, obj.h1 < 0, adx > 25, bollw[-1] > 0.02 * n,
boll_y < 0.02 * n, techs.boll_x > 10):
techs.choice = 2
sign = True
if sign_observation.assemble(
1,
obj.h1 < 0,
adx > 10,
bollw[-1] > 0.03 * n,
boll_y < 0.5 * n,
techs.boll_x > 10,
techs.jzd[1] * 1000 < 2, #标准差与集中度相关， 偏度与峰度与集中度没有发现关联
#abs(techs.jzd[-2]) < 1,
#abs(techs.jzd[-1]) < 0.5,
1):
techs.choice = 3
sign = True
#第一次机会
if sign_observation.assemble(0, obj.h1 < 0, adx > 25, bollw[-1] > 0.02 * n,
float(techs.boll_low_zz_1) < -0.01 * n,
techs.boll_x > 1, boll_y > 0.8, 1):
sign = True
techs.choice = 1
#输出判断
if sign:
pl: publish.Publish
pl.insertHtml("<tr><td>")
ui.drawBoll(pl, closes, boll_up, boll_mid, boll_low)
pl.insertHtml("</td><td>")
ui.DrawZZ(pl, zz_boll_up, is_append=ui.draw_style.head)
ui.DrawZZ(pl, zz_boll_mid, is_append=ui.draw_style.mid)
ui.DrawZZ(pl, zz_close, c='b', is_append=ui.draw_style.mid)
ui.DrawZZ(pl, zz_close_short, c='g', is_append=ui.draw_style.mid)
ui.DrawZZ(pl, zz_boll_low, is_append=ui.draw_style.end)
pl.insertHtml("</td><td>")
vals = agl.get_print_object(techs)
pl.insertHtml(vals)
pl.insertHtml("<br>%.2f,%.3f,%.2f,%.2f" % techs.jzd)
pl.insertHtml("</td>")
#pl.insertHtml("<td>")
#pl.insertHtml("</td>")
pl.insertHtml("</tr>")
return sign
def Run(self):
account = self._getAccount()
code = self.data.get_code()
hisdat = self.data.get_hisdat(code)
#hisdat = hisdat.dropna()
#closes = hisdat['c'].dropna()
closes = hisdat['c']
if len(closes) < 60:
return
index = len(closes)
#计算技术指标
#four = stock.FOUR(closes)
four = self.getCalcTech('four', index)
#boll_up, boll_mid, boll_low ,boll_w= stock.TDX_BOLL2(closes)
boll_up, boll_mid, boll_low, boll_w = self.getCalcTech('boll', index)
#assert(agl.array_equal(boll_up, boll_up2))
#adx = stock.TDX_ADX(hisdat['h'], hisdat['l'], closes)
adx = self.getCalcTech('adx', index)
#assert(agl.array_equal(adx, adx2))
#收集使用过的技术指标，为了报告使用
self.tech = closes, four, boll_up, boll_mid, boll_low, boll_w, adx
four = four[-1]
price = hisdat.iloc[-1]['c']
#print self.getCurTime(), four, price, boll_low[-1], boll_w[-1]
#基于价值回归及成本控制，加上技术指标择时
#缺点， 资金利用率低， 选股必须选好， 否则碰到一直下跌的情况就会造成亏损
#优点, 成本在相对安全区域
#当成本亏base_ratio%时加倍买入
df_stock_list = account.StockList()
df_stock_list = df_stock_list[df_stock_list['证券代码'] == code]
#第一次买
if so.assemble(
four < self.base_four[0],
len(df_stock_list) == 0,
#self.first_price == 0,
#adx[-1] > 60,
#price<boll_low[-1] ,
#boll_w[-1]>2,
#syl < 50,
):
if hasattr(self, 'base_num_ratio'):
total_money = ac.AccountMgr(self._getAccount(), price,
code).init_money()
self.base_num = ac.ShouShu(total_money * self.base_num_ratio /
price)
self._getAccount().Order(0, code, price, self.base_num)
if self.first_price == 0:
self.first_price = price
if len(df_stock_list) > 0:
chengben = float(df_stock_list['参考盈亏成本价'].loc[0])
changwei = int(df_stock_list['库存数量'].loc[0])
yinkui = calcYinKui(price, self.first_price)
#agl.Print( '盈亏:', chengben, yinkui, changwei, '频繁')
buchagn_ratio = self._Fire(yinkui)
if buchagn_ratio != 0: #补仓
df_zhijing = account.ZhiJing()
canuse_money = float(df_zhijing['可用'][-1])
total_money = canuse_money + changwei * price
use_money = total_money * buchagn_ratio
num = ac.ShouShu(use_money / price)
if num > 0:
account.Order(0, code, price, num)
#当赚base_ratio%时加倍卖出
chengben = float(df_stock_list['买入均价'].loc[0])
if four > self.base_four[1]: #第一次卖
account.Order(1, code, price, self.base_num)
#print '--',chengben, sell_avg_price.get()*(1+self.base_ratio)
if price > self.first_price * (1 + abs(yinkui)):
num = int(df_stock_list['可卖数量'].loc[0])
num = num * 0.1
num = ac.ShouShu(num)
if num > 0:
account.Order(1, code, price, num)
def Run(self):
"""
"""
#self._log('Strategy_Boll_Pre')
#以下为交易测试
code = self.data.get_code() #当前策略处理的股票
self.code = code
if not self.is_backtesting and not self.AllowCode(code):
return
df_hisdat = self.data.get_hisdat(code) #日k线
df_five_hisdat = self.data.get_hisdat(code, dtype='5min') #5分钟k线
df_fenshi = self.data.get_fenshi(code) #日分时
if len(df_fenshi) == 0:
self.data.log(code + u"未取到分时数据")
return
account = self._getAccount() #获取交易账户
price = float(df_fenshi.tail(1)['p']) #当前股价
closes = df_hisdat['c']
yestoday_close = closes[-2] #昨日收盘价
zhangfu = stock.ZhangFu(price, yestoday_close)
self._log(
'price=%.2f %s %s' %
(price, str(df_fenshi.index[-1]), str(df_five_hisdat.iloc[-1])))
account_mgr = ac.AccountMgr(account, price, code)
trade_num = ac.ShouShu(account_mgr.init_money() *
self.trade_num_use_money_percent / price)
trade_num = max(100, trade_num)
# 信号计算
four = stock.FOUR(closes)
four = four[-1]
upper, middle, lower = stock.TDX_BOLL(df_five_hisdat['c'])
highs, lows, closes = df_five_hisdat['h'], df_five_hisdat[
'l'], df_five_hisdat['c']
adx = stock.TDX_ADX(highs, lows, closes)
self._log('boll : %.2f,%.2f,%.2f' % (upper[-1], middle[-1], lower[-1]))
boll_w = abs(upper[-1] - lower[-1]) / middle[-1] * 100
#50个周期内最高值
is_high = abs(price - max(df_five_hisdat[-1000:]['h'])) / price < 0.005
boll_poss = [
upper[-1],
(upper[-1] - middle[-1]) / 2 + middle[-1],
middle[-1],
(middle[-1] - lower[-1]) / 2 + lower[-1],
lower[-1],
]
self._log('boll_poss: %.2f, %.2f boll_w=%.2f adx=%.2f' %
(boll_poss[0], boll_poss[1], boll_w, adx[-1]))
#上一个成交的价位
pre_price = account_mgr.last_chengjiao_price()
pre_pre_price = account_mgr.last_chengjiao_price(index=-2)
sell_count = account_mgr.queryTradeCount(1)
buy_count = account_mgr.queryTradeCount(0)
#买入均价
adx = adx[-1]
boll_up_ratio = 0.02
#信号判断
num = 0
if so.assemble(
price > boll_poss[1],
price > pre_price * (1 + self.trade_ratio),
#price > boll_poss[2],
#price > self.max_buy_price*(1+self.trade_ratio),
#boll_w > 3.5,
#adx > 60,
#sell_count < 2,
#pr.horizontal(df_five_hisdat),
0,
):
num = -trade_num
self.trade_status = self.enum.boll_up_mid
#if self.order(1, code, price, num):
#self._log(agl.utf8_to_ascii('一档卖出%s, %.2f, %d'%(code, price, num)))
if so.assemble(
price > boll_poss[0],
price > pre_price * (1 + self.trade_ratio),
#price > self.max_buy_price*(1+self.trade_ratio),
#boll_w > 3,
adx > 60,
is_high,
#four > self.trade_four[1],
#sell_count < 2,
#self.trade_status == self.enum.nothing,
#0,
):
num = -trade_num * 3
self.trade_status = self.enum.boll_up
#if self.order(1, code, price, num):
#self._log(agl.utf8_to_ascii('二档卖出%s, %.2f, %d'%(code, price, num)))
if so.assemble(
price < boll_poss[-2] * (1 + boll_up_ratio),
price < pre_price * (1 - self.trade_ratio),
#price < boll_poss[2],
#price < self.min_sell_price*(1-0.03),
#boll_w > 3.5,
#adx>60,
#buy_count < 2,
#pr.horizontal(df_five_hisdat),
0,
):
num = trade_num
self.trade_status = self.enum.boll_down_mid
#if boll_w > 6:
#num *= self.trade_num_ratio
#if self.order(0, code, price, num):
#self._log(agl.utf8_to_ascii('一档买入%s, %.2f, %d'%(code, price, num)))
if so.assemble(
price < boll_poss[-1],
price < pre_price * (1 - self.trade_ratio),
#price < self.min_sell_price*(1-0.03),
#boll_w > 3,
#buy_count < 2,
#self.trade_status == self.enum.nothing,
#adx>70,
#four < self.trade_four[0],
#0,
):
num = trade_num * 3
#num = account_mgr.last_chengjiao_num()
self.trade_status = self.enum.boll_down
#if self.order(0, code, price, num):
#self._log(agl.utf8_to_ascii('二档买入%s, %.2f, %d'%(code, price, num)))
#成本区间
if so.assemble(
price < pre_price * (1 - 0.05),
four < -0.1,
self.trade_status == self.enum.boll_up,
0,
):
num = trade_num * self.trade_num_ratio
self.trade_status = self.enum.nothing
if so.assemble(
price > pre_price * (1 + 0.05),
four > 0.1,
self.trade_status == self.enum.boll_down,
0,
):
num = -trade_num * self.trade_num_ratio
self.trade_status = self.enum.nothing
#zz顶抛出后回补
if so.assemble(
price < pre_price * (1 - 0.02),
#sell_count >= 2,
self.trade_status == self.enum.zz_up,
0,
):
#上次zz卖出的数量
num = account_mgr.last_chengjiao_num()
self.trade_status = self.enum.zz_hui_bu
if so.assemble(
price > pre_price * (1 + 0.02),
#sell_count >= 2,
self.trade_status == self.enum.zz_down,
0,
):
#上次zz卖出的数量
num = account_mgr.last_chengjiao_num()
self.trade_status = self.enum.zz_hui_bu
#计算分时zz
zz_sign = 0
closes = df_five_hisdat['c'][-200:].values
zz = stock.ZigZag(closes)
if len(zz) > 2:
zz_result = stock.analyzeZZ(zz)
zz_line_ratio = zz_result[1] / zz_result[0] #线段比率
#扑捉大涨回头的信号
if abs(zz_result[0]) > 0.05 and abs(zz_line_ratio) > 0.05 and abs(
zz_line_ratio) < 0.2 and abs(zz_result[0]) > 0.04:
zz_sign = agl.where(zz_result[1] > 0, 1, -1)
if num != 0:
bSell = agl.where(num > 0, 0, 1)
num = abs(num)
#if bSell:
#num = self._compensate(num, bSell, code)
#基本上每天的振幅都在1个点以上
if abs(stock.ZhangFu(price, yestoday_close)) > 0.01:
self.order(bSell, code, price, num)
zz_pre_price = myredis.createRedisVal('policy_basesign_zz_pre_price',
price)
if so.assemble(
zz_sign != 0,
0,
):
#print self.price, getZZPrePrice(self.price),abs(self.price-getZZPrePrice(self.price))/self.price
num = trade_num * 12
bSell = agl.where(zz_sign > 0, 0, 1)
num = self._compensate(num, bSell, code)
bCanOrder = False
if so.assemble(
bSell,
price > zz_pre_price.get() * (1 + 0.03),
#price > pre_price*(1+self.trade_ratio),
):
bCanOrder = True
self.trade_status = self.enum.zz_up
if so.assemble(
(not bSell),
price < zz_pre_price.get() * (1 - 0.03),
#price < pre_price*(1-self.trade_ratio)
):
bCanOrder = True
self.trade_status = self.enum.zz_down
if bCanOrder:
self._getAccount().Order(bSell, code, price, num)
zz_pre_price.set(price)
#信号发生时语音播报, 并通知界面回显
if not self.is_backtesting and (price > boll_poss[1]
or price < boll_poss[-2]):
codename = stock.GetCodeName(code)
s = '%s, %.2f' % (codename, price)
self.data.show(codename) #通知界面显示
self.data.speak2(s) #语音播报
#tick report
if self.is_backtesting and self.is_tick_report:
self._getAccount().TickReport(df_five_hisdat, 'win')
return
def Run(self):
"""
"""
#self._log('Strategy_Boll_Pre')
#以下为交易测试
code = self.data.get_code() #当前策略处理的股票
self.code = code
if not self.is_backtesting and not self.AllowCode(code):
return
self._log(self.getCurTime())
df_hisdat = self.data.get_hisdat(code) #日k线
df_five_hisdat = self.data.get_hisdat(code, dtype='5min') #5分钟k线
if len(df_five_hisdat) <= 30:
return
account = self._getAccount() #获取交易账户
price = float(df_five_hisdat.iloc[-1]['c']) #当前股价
closes = df_hisdat['c']
yestoday_close = closes[-2] #昨日收盘价
account_mgr = ac.AccountMgr(account, price, code)
trade_num = ac.ShouShu(account_mgr.init_money() * self.trade_num[0] /
price)
# 指标计算
index = len(df_five_hisdat)
four = self.getCalcTech('four', index)
upper, middle, lower = self.getCalcTech('boll', index)
df_five_hisdat.is_copy = False
df_five_hisdat['boll_up'] = upper
df_five_hisdat['boll_mid'] = middle
df_five_hisdat['boll_lower'] = lower
adx = self.getCalcTech('adx', index)
self._log('boll : %.2f,%.2f,%.2f' % (upper[-1], middle[-1], lower[-1]))
boll_w = abs(upper[-1] - lower[-1]) / middle[-1] * 100
#zz_up = stock.ZigZag(upper[-60:], percent=.1)
#zz_low = stock.ZigZag(lower[-60:], percent=0.1)
boll_poss = [
upper[-1],
(upper[-1] - middle[-1]) / 2 + middle[-1],
middle[-1],
(middle[-1] - lower[-1]) / 2 + lower[-1],
lower[-1],
]
self.tech = np.array(df_five_hisdat['c']), four, adx
four = four[-1]
adx = adx[-1]
self._log('boll_poss: %.2f, %.2f boll_w=%.2f adx=%.2f' %
(boll_poss[0], boll_poss[1], boll_w, adx))
if np.isnan(four):
return
#形态识别
#shape_is_matched = True
shape_is_matched = False
if price > boll_poss[2]:
base_boll_up = pr.get_boll_up_base()
recognizer = pr.Recognize_boll(base_boll_up, df_five_hisdat)
shape_is_matched = recognizer.is_matched()
else:
base_boll_lower = pr.get_boll_lower_base()
recognizer = pr.Recognize_boll(base_boll_lower, df_five_hisdat)
shape_is_matched = recognizer.is_matched()
#帐户信息
pre_price = account_mgr.last_chengjiao_price(is_sell=-1) #上一个成交的价位
pre_buy_price = account_mgr.last_chengjiao_price(is_sell=0)
if np.isnan(pre_buy_price):
pre_buy_price = pre_price
pre_sell_num = account_mgr.last_chengjiao_num() #上次的成交数量
pre_pre_price = account_mgr.last_chengjiao_price(index=-2)
sell_count = account_mgr.queryTradeCount(1)
buy_count = account_mgr.queryTradeCount(0)
chen_ben = account_mgr.get_BuyAvgPrice() #买入成本
yin_kui = account_mgr.yin_kui() #盈亏成本
canwei = account_mgr.getCurCanWei()
#信号判断
num = 0
#卖
if so.assemble(
#price > boll_poss[1]*1.001,
price > pre_price * (1 + self.trade_ratio),
#price > boll_poss[2],
#price > self.max_buy_price*(1+self.trade_ratio),
#boll_w > 3.5,
#adx > self.trade_adx,
#four > self.trade_four[1],
#sell_count < 2,
#pr.horizontal(df_five_hisdat),
shape_is_matched,
#0,
):
num = -trade_num
#if so.assemble(price > boll_poss[0] ,
#price > pre_price*(1+self.trade_ratio),
##price > self.max_buy_price*(1+self.trade_ratio),
#boll_w > 3.5,∂
##adx>60,
##four > self.trade_four[1],
##sell_count < 2,
##self.trade_status == self.enum.nothing,
##0,
#):
#if pre_sell_num>0:
#num = (pre_sell_num * self.trade_num_ratio)
#else:
#num = ac.ShouShu(account_mgr.getCurCanWei() * 0.5)
#order = 1
#买
if so.assemble(
#price < boll_poss[-2]*0.999,
price < pre_price * (1 - self.trade_ratio),
#price < boll_poss[2],
#price < self.min_sell_price*(1-0.03),
#boll_w > 3.5,
four < self.trade_four[0],
#adx>self.trade_adx,
#buy_count < 2,
#pr.horizontal(df_five_hisdat),
shape_is_matched,
#0,
):
num = trade_num
#if so.assemble(
#price < pre_buy_price*(1-self.trade_ratio),
##price < self.min_sell_price*(1-0.03),
##boll_w > 3.5,
##buy_count < 2,
##self.trade_status == self.enum.nothing,
##adx>70,
#four < self.trade_four[0]-0.02,
##0,
#):
###加仓买入
#num = self._compensate(trade_num, 0, code)
##order = 0
if abs(num) > 0:
order = num < 0 and 1 or 0
self.order(order, code, price, abs(num))
#tick report
if self.is_backtesting and self.is_tick_report:
self._getAccount().TickReport(df_five_hisdat, 'win')
return