def createEarnHoldBarsImg():
from earnmi_demo.strategy_demo.holdbars.HoldBarAnanysic import macd, kdj
sw = SWImpl()
lists = sw.getSW2List()
start = datetime(2014, 5, 1)
end = datetime(2020, 8, 17)
chart = Chart()
indictor = macd()
computeCount = 0
for code in lists:
bars = sw.getSW2Daily(code, start, end)
# print(f"bar.size = {bars.__len__()}")
chart.run(bars, indictor)
holdBarList = indictor.getHoldBars()
for holdBar in holdBarList:
barList = holdBar.bars
if (HoldBarUtils.isEarnBarList(barList)):
tiemStr = holdBar.start_time.strftime('%Y%m%d')
chart.show(barList, savefig=f'imgs\\earn-{code}-{tiemStr}.png')
# print(f"code:{code},cost_pct = %.2f%%" % (total_cost_pct*100))
computeCount = computeCount + 1
if computeCount > 10:
break
def computeAll():
from earnmi.data.SWImpl import SWImpl
from earnmi.chart.Chart import Chart, IndicatorItem, Signal
sw = SWImpl()
lists = sw.getSW2List()
start = datetime(2018, 5, 1)
end = datetime(2020, 8, 17)
dataSet = {}
class DataItem(object):
pass
for code in lists:
#for code in lists:
barList = sw.getSW2Daily(code, start, end)
indicator = Indicator()
for bar in barList:
##先识别形态
rets = KPattern.matchIndicator(indicator)
size = len(rets)
if size > 0:
"""有形态识别出来
"""
for item in rets:
name = item.name
value = item.value
dataItem = None
if dataSet.__contains__(name):
dataItem = dataSet[name]
else:
dataItem = DataItem()
dataItem.values = [] ##形态被识别的值。
dataItem.pcts = [] ##识别之后第二天的盈利情况
dataSet[name] = dataItem
##第二天的收益
pct = (bar.close_price - bar.open_price) / bar.open_price
##收录当前形态
dataItem.values.append(value)
dataItem.pcts.append(pct)
pass
indicator.update_bar(bar)
##打印当前形态
print(f"总共识别出{len(dataSet)}个形态")
for key, dataItem in dataSet.items():
values = np.array(dataItem.values)
pcts = np.array(dataItem.pcts) * 100
print(
f"{key}： len={len(dataItem.values)},values:{values.mean()},pcts:%.2f%%,pcts_std=%.2f"
% (pcts.mean(), pcts.std()))
def collectKPattherAndShowChart():
from earnmi.data.SWImpl import SWImpl
from vnpy.trader.constant import Exchange
from vnpy.trader.constant import Interval
sw = SWImpl()
lists = sw.getSW2List()
start = datetime(2014, 5, 1)
end = datetime(2020, 8, 17)
bars = []
limitSize = 0
chart = Chart()
for code in lists:
# for code in lists:
barList = sw.getSW2Daily(code, start, end)
indicator = Indicator(40)
preBar = None
yestodayIsMatch = False
for i in range(0, len(barList)):
bar = barList[i]
indicator.update_bar(bar)
patternValue = KPattern.encode1KAgo1(indicator)
todayIsMatch = 9 == patternValue
if todayIsMatch:
if indicator.count > 20:
chart.show(
indicator.makeBars(),
savefig=f"imgs/collectKPattherAndShowChart_{limitSize}"
)
limitSize += 1
if (limitSize > 50):
break
pass
if yestodayIsMatch:
pass
preBar = bar
yestodayIsMatch = todayIsMatch
if (limitSize > 50):
break
pass
class SWDataSource(BarDataSource):
def __init__(self, start: datetime, end: datetime):
self.index = 0
from earnmi.data.SWImpl import SWImpl
self.sw = SWImpl()
self.start = start
self.end = end
def nextBars(self) -> Tuple[Sequence['BarData'], str]:
# if self.index > 2:
# return None,None
sw_code_list = self.sw.getSW2List()
if self.index < len(sw_code_list):
code = sw_code_list[self.index]
self.index += 1
return self.sw.getSW2Daily(code, self.start, self.end), code
return None, None
def buildAndSaveModel(start: datetime, end: datetime, patternList=[]):
sw = SWImpl()
from earnmi_demo.strategy_demo.kbars.analysis_KPattern_skip1_predit2 import \
Generate_Feature_KPattern_skip1_predit2
for kPattern in patternList:
generateTrainData = Generate_Feature_KPattern_skip1_predit2(
kPatters=[kPattern])
sw.collect(start, end, generateTrainData)
featureData = generateTrainData.getPandasData()
filleName = f"models/predict_sw_top_k_{kPattern}.m"
print(f"k线形态[{kPattern}]的模型能力:")
model = PredictModel()
model.setFeature(featureData)
model.printCrossScoreTest()
model.saveToFile(filleName)
def crateHoldbarsImg():
sw = SWImpl()
lists = sw.getSW2List()
start = datetime(2019, 5, 1)
end = datetime(2020, 8, 17)
chart = Chart()
for code in lists:
bars = sw.getSW2Daily(code, start, end)
# print(f"bar.size = {bars.__len__()}")
indictor = kdj()
chart.run(bars, indictor)
holdBarList = indictor.getHoldBars()
barList = utils.to_bars(holdBarList)
chart.show(barList, item=indictor, savefig=f'imgs\\{code}.png')
#print(f"code:{code},cost_pct = %.2f%%" % (total_cost_pct*100))
break
def __init__(self):
self.lasted15Bar = np.array([
None, None, None, None, None, None, None, None, None, None, None,
None, None, None, None
])
self.lasted3BarKdj = np.array([None, None, None])
self.lasted3BarMacd = np.array([None, None, None])
self.lasted3BarArron = np.array([None, None])
self.kdjEncoder = FloatEncoder([15, 30, 45, 60, 75, 90])
self.mDateOccurCountMap = {} ##统计产生收集个数的次数
self.sw = SWImpl()
class QuantStrategy(CoreEngineStrategy):
def __init__(self):
self.sw = SWImpl()
def generatePredictOrder(self, engine: CoreEngine, predict: PredictData,debugPrams:{}=None) -> PredictOrder:
if debugPrams is None:
debugPrams = {}
quantData = engine.queryQuantData(predict.dimen)
code = predict.collectData.occurBars[-1].symbol
name = self.sw.getSw2Name(code)
order = PredictOrder(dimen=predict.dimen, code=code, name=name)
start_price = engine.getEngineModel().getYBasePrice(predict.collectData)
_min, _max = quantData.getSellFloatEncoder().parseEncode(quantData.sellRange[0].encode)
order.suggestSellPrice = start_price * (1 + (_min + _max) / 2 / 100)
_min, _max = quantData.getBuyFloatEncoder().parseEncode(quantData.buyRange[0].encode)
order.suggestBuyPrice = start_price * (1 + (_min + _max) / 2 / 100)
order.power_rate = quantData.getPowerRate();
##for backTest
self.checkIfBuyPrice(order,predict.collectData.occurBars[-1].close_price,debugPrams)
return order
def checkIfBuyPrice(self,order: PredictOrder,targetPrice:float,debugPrams:{}=None):
if order.status != PredictOrderStatus.TRACE:
return
quantData = engine.queryQuantData(order.dimen)
if quantData.getPowerRate() > 0.9 and order.suggestBuyPrice >= targetPrice:
order.status = PredictOrderStatus.HOLD
order.buyPrice = targetPrice
def updatePredictOrder(self, order: PredictOrder, bar: BarData, isTodayLastBar: bool,debugPrams:{}):
if (order.status == PredictOrderStatus.HOLD):
if bar.high_price >= order.suggestSellPrice:
order.sellPrice = order.suggestSellPrice
order.status = PredictOrderStatus.CROSS
return
if order.holdDay >= 1:
order.sellPrice = bar.close_price
order.status = PredictOrderStatus.CROSS
return
order.holdDay += 1
self.checkIfBuyPrice(order,bar.low_price,debugPrams)
from datetime import datetime
from earnmi.data.MarketImpl import MarketImpl
from earnmi.data.SWImpl import SWImpl
import time
sw = SWImpl()
swCode = sw.getSW2List()[0]
stockslist = sw.getSW2Stocks(swCode)
print(f"stockslist:{len(stockslist)}")
startDate = datetime(2015, 4, 1)
endDate = datetime(2020, 5, 1)
market = MarketImpl()
market.setToday(endDate)
market.addNotice('002852')
bars = market.getHistory().getKbars('002852', 300 * 3)
for code in stockslist:
market.addNotice(code)
print(f"getBars:{code}")
bars = market.getHistory().getKbars(code, 300 * 3)
#time.sleep(2)
print(f"getBars size:{len(bars)}")
dif, dea, macd_bar = indicator.macd(fast_period=12,
slow_period=26,
signal_period=9,
array=True)
##金叉出现
if (macd_bar[-1] >= 0 and macd_bar[-2] <= 0):
if not signal.hasBuy:
signal.buy = True
##死叉出现
if (macd_bar[-1] <= 0 and macd_bar[-2] >= 0):
if signal.hasBuy:
signal.sell = True
return values
sw = SWImpl()
lists = sw.getSW2List()
chart = Chart()
start = datetime(2014, 5, 1)
end = datetime(2020, 8, 17)
dict = {}
dict[0] = []
dict[1] = []
dict[2] = []
dict[3] = []
dict[4] = []
indictor = macd()
for code in lists:
if len(sw.getSW2Stocks(code)) < 10:
def __init__(self):
self.lasted3Bar = np.array([None, None, None])
self.lasted3BarKdj = np.array([None, None, None])
self.kdjEncoder = FloatEncoder([15, 30, 45, 60, 75, 90])
self.sw = SWImpl()
def __init__(self):
self.sw = SWImpl()
self.mDateHoldCountMap = {} ##统计产生收集个数的次数
def printKPatterMoreDetail(kPatters=[
6, 3, 17, 81, 7, 5, 4, 82, 159, 16, 28, 83, 15, 84, 18, 27, 93, 104, 158,
92, 160, 236, 157, 94, 85, 80, 14, 8, 161, 9, 29, 170, 26, 19, 38, 2, 79
]):
from earnmi.data.SWImpl import SWImpl
from vnpy.trader.constant import Exchange
from vnpy.trader.constant import Interval
sw = SWImpl()
lists = sw.getSW2List()
start = datetime(2014, 5, 1)
end = datetime(2020, 8, 17)
pct_split = [-7, -5, -3, -1.5, -0.5, 0.5, 1.5, 3, 5, 7]
pct_split = [-7, -5, -3, -1.0, 0, 1, 3, 5, 7]
pct_split = [-0.5, 0.5]
pctEncoder = FloatEncoder(pct_split)
kPattersMap = {}
for value in kPatters:
kPattersMap[value] = True
class InnerData(object):
kValue: int ##
sell_disbute = np.zeros(pctEncoder.mask()) ##卖方力量分布情况
buy_disbute = np.zeros(pctEncoder.mask()) #买方力量分布情况
pass
dataSet = {}
occurDayMap = {}
allTrayDay = 1
for code in lists:
# for code in lists:
barList = sw.getSW2Daily(code, start, end)
indicator = Indicator(40)
preBar = None
previousIsMatch = False
previousPatternVaule = None
allTrayDay = max(allTrayDay, len(barList))
for i in range(0, len(barList)):
bar = barList[i]
indicator.update_bar(bar)
patternValue = KPattern.encode1KAgo1(indicator)
todayIsMatch = False
if not patternValue is None:
todayIsMatch = kPattersMap.__contains__(patternValue)
if todayIsMatch:
dayKey = bar.datetime.year * 13 * 35 + bar.datetime.month * 13 + bar.datetime.day
occurDayMap[dayKey] = True
pass
if previousIsMatch:
innerData: InnerData = dataSet.get(previousIsMatch)
if innerData is None:
innerData = InnerData()
innerData.kValue = previousIsMatch
dataSet[previousPatternVaule] = innerData
sell_pct = 100 * ((bar.high_price + bar.close_price) / 2 -
preBar.close_price) / preBar.close_price
buy_pct = 100 * ((bar.low_price + bar.close_price) / 2 -
preBar.close_price) / preBar.close_price
innerData.buy_disbute[pctEncoder.encode(buy_pct)] += 1
innerData.sell_disbute[pctEncoder.encode(sell_pct)] += 1
pass
preBar = bar
previousIsMatch = todayIsMatch
previousPatternVaule = patternValue
print(f"所有交易日中，有意义的k线形态出现占比：%.2f%%" %
(100 * len(occurDayMap) / allTrayDay))
for kValue, dataItem in dataSet.items():
total_count1 = 0
total_count2 = 0
for cnt in dataItem.sell_disbute:
total_count1 += cnt
for cnt in dataItem.buy_disbute:
total_count2 += cnt
assert total_count1 == total_count2
assert total_count1 > 0
print(f"\n\nk:%6d, " % (kValue))
print(f" 卖方价格分布：")
for encode in range(0, len(dataItem.sell_disbute)):
occurtRate = 100 * dataItem.sell_disbute[encode] / total_count1
print(f" {pctEncoder.descriptEncdoe(encode)}：%.2f%%" %
(occurtRate))
print(f" 买方价格分布：")
for encode in range(0, len(dataItem.buy_disbute)):
occurtRate = 100 * dataItem.buy_disbute[encode] / total_count1
print(f" {pctEncoder.descriptEncdoe(encode)}：%.2f%%" %
(occurtRate))
pass
class MyPattherCollector(KPatternCollector):
# 收集指定的k线
collectKPatternOnly = True
KPattern: [] = [712]
pct_split = [-7, -5, -3, -1.5, -0.5, 0.5, 1.5, 3, 5, 7]
# pct_split = [-7, -5, -3, -1.0, 0, 1, 3, 5, 7]
# pct_split = [-0.5,0.5]
pctEncoder = FloatEncoder(pct_split)
def __init__(self):
self.kPattersMap = {}
self.sw = SWImpl()
self.dataSet = []
for value in self.KPattern:
self.kPattersMap[value] = True
def onStart(self, code: str) -> bool:
self.indicator = Indicator(40)
self.traceCode = code
self.traceName = self.sw.getSw2Name(code)
return True
"""
检查是否追踪某个k线形态，是的话，创建一个最终值对象。
"""
def checkIfTrace(self, newBar: BarData) -> TraceIn3DayItem:
self.indicator.update_bar(newBar)
kEncodeValue = KPattern.encode2KAgo1(self.indicator)
if kEncodeValue is None:
return None
if self.collectKPatternOnly and self.kPattersMap.get(
kEncodeValue) is None:
return None
return TraceIn3DayItem(kEncodeValue, newBar)
def onTraceFinish(self, traceItem: TraceIn3DayItem):
if traceItem.isWanted():
##收集数据。
data = []
data.append(self.traceCode)
data.append(self.traceName)
data.append(traceItem.kPattern)
data.append(traceItem.first_buy_pct)
data.append(traceItem.fisrt_sell_pct)
data.append(traceItem.current_sell_pct)
data.append(traceItem.current_buy_pct)
self.dataSet.append(data)
def onDestroy(self):
import pandas as pd
cloumns = [
"code", "name", "kPattern", "buy_price", "sell_price",
"label_sell_price", "label_buy_price"
]
wxl = pd.DataFrame(self.dataSet, columns=cloumns)
writer = pd.ExcelWriter('files/sw_train_data_sample.xlsx')
wxl.to_excel(writer, sheet_name="sample", index=False)
writer.save()
writer.close()
print(f"dataSize = {len(self.dataSet)}")
class MyStrategy(CoreEngineStrategy):
def __init__(self):
self.sw = SWImpl()
self.mDateHoldCountMap = {} ##统计产生收集个数的次数
def generatePredictOrder(self,
engine: CoreEngine,
predict: PredictData,
debugPrams: {} = None) -> PredictOrder:
if debugPrams is None:
debugPrams = {}
code = predict.collectData.occurBars[-1].symbol
name = self.sw.getSw2Name(code)
order = PredictOrder(dimen=predict.dimen, code=code, name=name)
predict_sell_pct = predict.getPredictSellPct(engine.getEngineModel())
predict_buy_pct = predict.getPredictBuyPct(engine.getEngineModel())
start_price = engine.getEngineModel().getYBasePrice(
predict.collectData)
order.suggestSellPrice = start_price * (1 + predict_sell_pct / 100)
order.suggestBuyPrice = start_price * (1 + predict_buy_pct / 100)
order.power_rate = engine.queryQuantData(predict.dimen).getPowerRate()
timeKey = utils.to_start_date(
predict.collectData.occurBars[-1].datetime)
if self.mDateHoldCountMap.get(timeKey) is None:
self.mDateHoldCountMap[timeKey] = 0
##for backTest
occurBar: BarData = predict.collectData.occurBars[-2]
skipBar: BarData = predict.collectData.occurBars[-1]
buy_price = skipBar.close_price
predict_sell_pct = 100 * (order.suggestSellPrice -
start_price) / start_price
predict_buy_pct = 100 * (order.suggestBuyPrice -
start_price) / start_price
buy_point_pct = 100 * (
buy_price - occurBar.close_price) / occurBar.close_price ##买入的价格
abilityData = engine.queryPredictAbilityData(predict.dimen)
quantData = engine.queryQuantData(predict.dimen)
delta = abs(quantData.sellCenterPct) - abs(quantData.buyCenterPct)
if abs(delta) < 0.05:
# 多空力量差不多
power = 0
if delta > 0:
# 适合做多
power = (quantData.sellCenterPct +
quantData.buyCenterPct) / quantData.sellCenterPct
else:
power = -(quantData.sellCenterPct +
quantData.buyCenterPct) / quantData.buyCenterPct
extraCondition = True
quant_power = debugPrams.get("quant_power")
if not quant_power is None:
extraCondition = extraCondition and predict.quantData.getPowerRate(
) >= quant_power
predict_buy_pct_param = debugPrams.get("predict_buy_pct")
if not predict_buy_pct_param is None:
extraCondition = extraCondition and predict_buy_pct >= predict_buy_pct_param
if extraCondition and predict_sell_pct - buy_point_pct > 1 \
and abilityData.trainData.biasSellLoss < 10:
order.status = PredictOrderStatus.HOLD
order.buyPrice = buy_price
timeKey = utils.to_start_date(
predict.collectData.occurBars[-1].datetime)
self.mDateHoldCountMap[timeKey] += 1
else:
order.status = PredictOrderStatus.STOP
return order
def updatePredictOrder(self,
order: PredictOrder,
bar: BarData,
isTodayLastBar: bool,
debugParams: {} = None):
if (order.status == PredictOrderStatus.HOLD):
if bar.high_price >= order.suggestSellPrice:
order.sellPrice = order.suggestSellPrice
order.status = PredictOrderStatus.CROSS
return
order.holdDay += 1
if order.holdDay >= 2:
order.sellPrice = bar.close_price
order.status = PredictOrderStatus.CROSS
return
def computeHoldBarIndictor(indictor:IndicatorItem)->HoldBarData:
sw = SWImpl()
lists = sw.getSW2List()
chart = Chart()
total_cost_pcts = []#收益
avg_eran_cost_pcts = []#每个盈利holdbard的平均盈利
total_days = []
total_eran_days = []
total_holdbars = []
total_holdbars_earn = []
max_cost_pcts =[] #最大收益
min_cost_pcts = []
for code in lists:
if len(sw.getSW2Stocks(code)) < 10:
continue
bars = sw.getSW2Daily(code, start, end)
#print(f"bar.size = {bars.__len__()}")
chart.run(bars, indictor)
holdbarList = indictor.getHoldBars()
#holdbarList = HoldBarUtils.filterHoldBar(holdbarList)
data = HoldBarUtils.computeHoldBarIndictor(holdbarList);
if data is None:
continue
total_cost_pcts.append(data.total_cost_pct)
max_cost_pcts.append(data.max_cost_pct)
min_cost_pcts.append(data.min_cost_pct)
total_days.append(data.total_day)
total_holdbars.append(data.total_holdbar)
total_holdbars_earn.append(data.total_holdbar_earn)
avg_eran_cost_pcts.append(data.avg_eran_cost_pct)
total_eran_days.append(data.total_earn_day)
ret = HoldBarData()
total_cost_pcts = np.array(total_cost_pcts)
total_days = np.array(total_days)
max_cost_pcts = np.array(max_cost_pcts)
min_cost_pcts = np.array(min_cost_pcts)
total_holdbars = np.array(total_holdbars)
total_holdbars_earn = np.array(total_holdbars_earn)
avg_eran_cost_pcts = np.array(avg_eran_cost_pcts)
total_eran_days = np.array(total_eran_days)
ret.total_min_cost_pct = total_cost_pcts.min()
ret.total_max_cost_pct = total_cost_pcts.max()
ret.total_cost_pct = total_cost_pcts.mean()
ret.total_cost_pct_std = np.std(total_cost_pcts)
ret.total_day = total_days.mean()
ret.max_cost_pct = max_cost_pcts.mean()
ret.min_cost_pct = min_cost_pcts.mean()
ret.total_holdbar = total_holdbars.mean()
ret.total_holdbar_earn = total_holdbars_earn.mean()
ret.avg_eran_cost_pct = avg_eran_cost_pcts.mean()
ret.total_earn_day = total_eran_days.mean()
return ret
def compute_SW_KEncode_parseAlgro1_split(
pct_split=[-7, -5, -3, -1.5, -0.5, 0.5, 1.5, 3, 5, 7],
extra_split=[1, 2, 3]):
from earnmi.data.SWImpl import SWImpl
sw = SWImpl()
lists = sw.getSW2List()
start = datetime(2014, 5, 1)
end = datetime(2020, 8, 17)
dataSet = {}
pct_split = [-7, -5, -3, -1.5, -0.5, 0.5, 1.5, 3, 5, 7]
extra_split = [0.5, 1.0, 1.5, 2.0, 2.5, 2.0]
total_count = 0
pct_code_count = np.zeros(len(pct_split) + 1)
high_extra_pct_code_count = np.zeros(len(extra_split) + 1)
low_extra_pct_code_count = np.zeros(len(extra_split) + 1)
for code in lists:
#for code in lists:
barList = sw.getSW2Daily(code, start, end)
indicator = Indicator(40)
preBar = None
for i in range(0, len(barList)):
bar = barList[i]
indicator.update_bar(bar)
if (i > 10):
k_code, pct_code, high_extra_pct_code, low_extra_pct_code = KEncode.parseAlgro1(
indicator.close[-2], indicator.open[-1],
indicator.high[-1], indicator.low[-1], indicator.close[-1],
pct_split, extra_split)
high_extra_pct_code_count[
high_extra_pct_code] = high_extra_pct_code_count[
high_extra_pct_code] + 1
low_extra_pct_code_count[
low_extra_pct_code] = low_extra_pct_code_count[
low_extra_pct_code] + 1
pct_code_count[pct_code] = pct_code_count[pct_code] + 1
total_count += 1
preBar = bar
##打印当前形态
print(f"pct_split: {pct_split}")
print(f"extra_split: {extra_split}")
count_list = pct_code_count
print(f"应该服从正太分布")
print(f"\n：pct_code_count分布,avg = %.4f%%" % (100 / len(count_list)))
for codeId in range(0, len(count_list)):
item_count = count_list[codeId]
item_occur_rate = 100 * item_count / total_count
print(f"\t{codeId}: %.4f %% count:%d" %
(item_occur_rate, item_count))
count_list = high_extra_pct_code_count
print(f"\n：high_extra_pct_code_count分布,avg = %.4f%%" %
(100 / len(count_list)))
for codeId in range(0, len(count_list)):
item_count = count_list[codeId]
item_occur_rate = 100 * item_count / total_count
print(f"\t{codeId}: %.4f %% count:%d" %
(item_occur_rate, item_count))
count_list = low_extra_pct_code_count
print(f"\n：low_extra_pct_code_count分布,avg = %.4f%%" %
(100 / len(count_list)))
for codeId in range(0, len(count_list)):
item_count = count_list[codeId]
item_occur_rate = 100 * item_count / total_count
print(f"\t{codeId}: %.4f %% count:%d" %
(item_occur_rate, item_count))
class Generate_Feature_KPattern_skip1_predit2(
More_detail_KPattern_skip1_predit2):
def __init__(self, limit_close_pct=1, kPatters: [] = None):
super().__init__(limit_close_pct, kPatters)
self.traceDatas = []
self.sw = SWImpl()
self.predictTraceDatas: ['Skip1_Predict2_TraceData'] = []
#pct_split = [-7, -5, -3, -1.5, -0.5, 0.5, 1.5, 3, 5, 7]
#self.pctEncoder = FloatEncoder(pct_split)
def onCreate(self):
super().onCreate()
self.traceDatas.clear()
def doWantedTraceData(self, traceData: Skip1_Predict2_TraceData,
countData: CountData):
super().doWantedTraceData(traceData, countData)
self.traceDatas.append(traceData)
pass
def onTraceStop(self, traceData: Skip1_Predict2_TraceData):
super().onTraceStop(traceData)
if (not traceData.skipBar is None):
self.predictTraceDatas.append(traceData)
def generateData(
self,
traceDatas: Sequence["Skip1_Predict2_TraceData"]) -> pd.DataFrame:
trainDataSet = []
for traceData in traceDatas:
occurBar = traceData.occurBar
skipBar = traceData.skipBar
sell_pct = 100 * ((skipBar.high_price + skipBar.close_price) / 2 -
occurBar.close_price) / occurBar.close_price
buy_pct = 100 * ((skipBar.low_price + skipBar.close_price) / 2 -
occurBar.close_price) / occurBar.close_price
data = []
data.append(self.code)
data.append(self.sw.getSw2Name(self.code))
data.append(traceData.kPatternValue)
data.append(buy_pct)
data.append(sell_pct)
data.append(traceData.indicator_k)
data.append(traceData.indicator_j)
data.append(traceData.sell_pct)
data.append(traceData.buy_pct)
trainDataSet.append(data)
cloumns = [
"code", "name", "kPattern", "buy_price", "sell_price", "k", "j",
"label_sell_price", "label_buy_price"
]
return pd.DataFrame(trainDataSet, columns=cloumns)
def getPandasData(self) -> pd.DataFrame:
return self.generateData(self.traceDatas)
def writeToXml(self) -> bool:
size = len(self.traceDatas)
if size < 1:
print("no need to write")
return False
wxl = self.getPandasData()
writer = pd.ExcelWriter('files/sw_train_data_sample.xlsx')
wxl.to_excel(writer, sheet_name="sample", index=False)
writer.save()
writer.close()
print(f"write dataSize = {size}")
return True
def onDestroy(self):
super().onDestroy()
return True
start = datetime(2014, 5, 1)
end = datetime(2020, 8, 17)
# code = "600196"
#
#
# market = MarketImpl()
# market.addNotice(code)
# market.setToday(end)
#
#
# bars = market.getHistory().getKbarFrom(code,start)
sw = SWImpl()
lists = sw.getSW2List()
chart = Chart()
for code in lists:
bars = sw.getSW2Daily(code, start, end)
print(f"bar.size = {bars.__len__()}")
item = AroonItem()
chart.run(bars, item)
print(f"holdbars = {len(item.getHoldBars())}")
barList = []
close_price = None
for holdBar in item.getHoldBars():
if close_price is None:
def findKPatternThatIn3Day(first_day_pct: float = 3, targe_pct=3):
sw = SWImpl()
lists = sw.getSW2List()
start = datetime(2014, 5, 1)
end = datetime(2020, 8, 17)
dataSet = {}
total_count = 0
for code in lists:
# for code in lists:
barList = sw.getSW2Daily(code, start, end)
indicator = Indicator(40)
traceItems: ['TraceIn3DayItem'] = []
for bar in barList:
###跟踪数据
toDeleteList = []
for traceItem in traceItems:
traceItem.onTraceBar(bar)
if traceItem.isFinished():
toDeleteList.append(traceItem)
if traceItem.isWanted():
###归纳到统计里面
dataItem: CountItem = dataSet.get(traceItem.kPattern)
if dataItem is None:
dataItem = CountItem()
dataSet[traceItem.kPattern] = dataItem
pct = traceItem.current_sell_pct
total_count += 1
dataItem.count_total += 1
dataItem.pct_total += pct
if traceItem.isSuccess():
dataItem.count_earn += 1
dataItem.pct_earn += pct
pass
for traceItem in toDeleteList:
traceItems.remove(traceItem)
indicator.update_bar(bar)
kEncodeValue = KPattern.encode2KAgo1(indicator)
if kEncodeValue is None:
continue
traceItem = TraceIn3DayItem(kEncodeValue, bar)
traceItems.append(traceItem)
##打印当前形态
occur_count = 0
print(f"总共分析{total_count}个形态，识别出{len(dataSet)}个形态，有意义的形态有：")
max_succ_rate = 0
min_succ_rate = 100
ret_list = []
for key, dataItem in dataSet.items():
success_rate = 100 * dataItem.count_earn / dataItem.count_total
if dataItem.count_total < 300:
continue
if success_rate < 40:
continue
ret_list.append(key)
if dataItem.count_earn > 0:
earn_pct = dataItem.pct_earn / dataItem.count_earn
else:
earn_pct = 0
avg_pct = dataItem.pct_total / dataItem.count_total
occur_count += dataItem.count_total
occur_rate = 100 * dataItem.count_total / total_count
max_succ_rate = max(success_rate, max_succ_rate)
min_succ_rate = min(success_rate, min_succ_rate)
print(
f"{key}： total={dataItem.count_total},suc=%.2f%%,occur_rate=%.2f%%,earn_pct:%.2f%%,avg_pct:%.2f%%)"
% (success_rate, occur_rate, earn_pct, avg_pct))
total_occur_rate = 100 * occur_count / total_count
print(f"总共：occur_rate=%.2f%%, min_succ_rate=%.2f%%, max_succ_rate=%.2f%%" %
(total_occur_rate, min_succ_rate, max_succ_rate))
print(f"{ret_list}")
def printKPatterMoreDetail(kPatters=[
535, 359, 1239, 1415, 1072, 712, 1412, 1240, 1413, 888, 2823, 706, 1414,
1064
]):
from vnpy.trader.constant import Exchange
from vnpy.trader.constant import Interval
sw = SWImpl()
lists = sw.getSW2List()
start = datetime(2014, 5, 1)
end = datetime(2020, 8, 17)
pct_split = [-7, -5, -3, -1.5, -0.5, 0.5, 1.5, 3, 5, 7]
#pct_split = [-7, -5, -3, -1.0, 0, 1, 3, 5, 7]
pct_split = [2]
pctEncoder = FloatEncoder(pct_split)
kPattersMap = {}
for value in kPatters:
kPattersMap[value] = True
class InnerData(object):
kValue: int ##
sell_disbute = np.zeros(pctEncoder.mask()) ##卖方力量分布情况
buy_disbute = np.zeros(pctEncoder.mask()) #买方力量分布情况
pass
dataSet = {}
occurDayMap = {}
allTrayDay = 1
for code in lists:
# for code in lists:
barList = sw.getSW2Daily(code, start, end)
indicator = Indicator(40)
traceItems: ['TraceIn3DayItem'] = []
allTrayDay = max(allTrayDay, len(barList))
for bar in barList:
###跟踪数据
toDeleteList = []
for traceItem in traceItems:
traceItem.onTraceBar(bar)
if traceItem.isFinished():
toDeleteList.append(traceItem)
if traceItem.isWanted():
occurBar = traceItem.firstBar
dayKey = occurBar.datetime.year * 13 * 35 + occurBar.datetime.month * 13 + occurBar.datetime.day
occurDayMap[dayKey] = True
###归纳到统计里面
innerData: InnerData = dataSet.get(traceItem.kPattern)
if innerData is None:
innerData = InnerData()
innerData.kValue = traceItem.kPattern
dataSet[traceItem.kPattern] = innerData
sell_pct = traceItem.current_sell_pct
buy_pct = traceItem.current_buy_pct
innerData.buy_disbute[pctEncoder.encode(buy_pct)] += 1
innerData.sell_disbute[pctEncoder.encode(
sell_pct)] += 1
pass
for traceItem in toDeleteList:
traceItems.remove(traceItem)
indicator.update_bar(bar)
kEncodeValue = KPattern.encode2KAgo1(indicator)
if kEncodeValue is None or kPattersMap.get(kEncodeValue) is None:
continue
traceItem = TraceIn3DayItem(kEncodeValue, bar)
traceItems.append(traceItem)
print(f"所有交易日中，有意义的k线形态出现占比：%.2f%%" %
(100 * len(occurDayMap) / allTrayDay))
for kValue, dataItem in dataSet.items():
total_count1 = 0
total_count2 = 0
for cnt in dataItem.sell_disbute:
total_count1 += cnt
for cnt in dataItem.buy_disbute:
total_count2 += cnt
assert total_count1 == total_count2
assert total_count1 > 0
print(f"\n\nk:%6d, " % (kValue))
print(f" 卖方价格分布：")
for encode in range(0, len(dataItem.sell_disbute)):
occurtRate = 100 * dataItem.sell_disbute[encode] / total_count1
print(f" {pctEncoder.descriptEncdoe(encode)}：%.2f%%" %
(occurtRate))
print(f" 买方价格分布：")
for encode in range(0, len(dataItem.buy_disbute)):
occurtRate = 100 * dataItem.buy_disbute[encode] / total_count1
print(f" {pctEncoder.descriptEncdoe(encode)}：%.2f%%" %
(occurtRate))
pass
def __init__(self):
self.kPattersMap = {}
self.sw = SWImpl()
self.dataSet = []
for value in self.KPattern:
self.kPattersMap[value] = True
if __name__ == "__main__":
import pickle
from earnmi.model.CoreEngineImpl import CoreEngineImpl
from earnmi.data.SWImpl import SWImpl
def saveCollectData(bars: []):
fileName = "files/testSaveCollectData.bin"
with open(fileName, 'wb') as fp:
pickle.dump(bars, fp, -1)
def loadCollectData():
bars = None
fileName = "files/testSaveCollectData.bin"
with open(fileName, 'rb') as fp:
bars = pickle.load(fp)
return bars
start = datetime(2014, 5, 1)
end = datetime(2020, 8, 17)
sw = SWImpl()
code = sw.getSW2List()[3]
bars = sw.getSW2Daily(code, start, end)
#saveCollectData(bars)
bars2 = loadCollectData()
assert bars == bars2
assert len(bars) == len(bars2) and len(bars2) != 0
def compute_SW_KPattern_data():
from earnmi.data.SWImpl import SWImpl
from earnmi.chart.Chart import Chart, IndicatorItem, Signal
sw = SWImpl()
lists = sw.getSW2List()
start = datetime(2014, 5, 1)
end = datetime(2020, 8, 17)
dataSet = {}
total_count = 0
for code in lists:
#for code in lists:
barList = sw.getSW2Daily(code, start, end)
indicator = Indicator()
preBar = None
for bar in barList:
##先识别形态
rets = KPattern.matchIndicator(indicator)
size = len(rets)
if size > 0 and not preBar is None:
"""有形态识别出来
"""
for item in rets:
name = item.name
value = item.value
total_count += 1
dataItem = None
if dataSet.__contains__(name):
dataItem = dataSet[name]
else:
dataItem = CountItem()
dataItem.values = [] ##形态被识别的值。
dataItem.pcts = [] ##识别之后第二天的盈利情况
dataSet[name] = dataItem
##第二天的收益
short_pct = ((bar.high_price + bar.close_price) / 2 -
preBar.close_price) / preBar.close_price
#long_pct = ((bar.high_price + bar.close_price) / 2 - preBar.close_price) / preBar.close_price
###pct = (bar.close_price - preBar.close_price) / preBar.close_price
##收录当前形态
dataItem.count_total += 1
dataItem.pct_total += short_pct
if short_pct > 0.000001:
dataItem.count_earn += 1
dataItem.pct_earn += short_pct
dataItem.values.append(value)
dataItem.pcts.append(short_pct)
pass
indicator.update_bar(bar)
preBar = bar
##打印当前形态
print(f"总共分析{total_count}个形态，识别出{len(dataSet)}个形态，有意义的形态有：")
for key, dataItem in dataSet.items():
if dataItem.count_total < 1000:
continue
success_rate = 100 * dataItem.count_earn / dataItem.count_total
if abs(int(success_rate - 50)) < 5:
continue
values = np.array(dataItem.values)
pcts = np.array(dataItem.pcts) * 100
count = len(values)
long_values = []
short_value = []
long_pcts = []
long_ok_cnt = 0
short_pcts = []
short_ok_cnt = 0
for i in range(0, count):
v = values[i]
if v > 0:
long_values.append(v)
long_pcts.append(pcts[i])
if pcts[i] >= 0.000001:
long_ok_cnt = long_ok_cnt + 1
else:
short_value.append(v)
short_pcts.append(pcts[i])
if pcts[i] <= -10.000001:
short_ok_cnt = short_ok_cnt + 1
long_values = np.array(long_values)
short_value = np.array(short_value)
long_pcts = np.array(long_pcts)
short_pcts = np.array(short_pcts)
long_pct = 0
long_std = math.nan
long_success = math.nan
short_pct = 0
short_std = math.nan
short_success = math.nan
if len(long_values) > 0:
long_pct = long_pcts.mean()
long_std = long_pcts.std()
long_success = long_ok_cnt / len(long_values)
if len(short_value) > 0:
short_pct = short_pcts.mean()
short_std = short_pcts.std()
short_success = short_ok_cnt / len(short_value)
print(
f"{key}： count={count},suc_reate=%.2f%%,long(size:{len(long_values)},suc=%.2f%%,pcts:%.2f%%,std=%.2f),short(size:{len(short_value)},suc=%.2f%%,pcts:%.2f%%,std=%.2f)"
% (success_rate, long_success * 100, long_pct, long_std,
short_success * 100, short_pct, short_std))
print("-----------具体情况-----------")
outputKeys = ["CDLADVANCEBLOCK"]
for key in outputKeys:
dataItem = dataSet[key]
values = np.array(dataItem.values)
pcts = np.array(dataItem.pcts) * 100
count = len(dataItem.values)
print(
f"{key}： count={count},values:{values.mean()},pcts:%.2f%%,pcts_std=%.2f"
% (pcts.mean(), pcts.std()))
itemSize = 10
size = int(count / itemSize)
if count % itemSize > 0:
size = size + 1
for i in range(0, size):
lineStr = ""
start = itemSize * i
end = min(start + itemSize, count)
for j in range(start, end):
lineStr = lineStr + (f" %4d->%.2f%%," % (values[j], pcts[j]))
print(lineStr)
print("no need to write")
return False
wxl = self.getPandasData()
writer = pd.ExcelWriter('files/sw_train_data_sample.xlsx')
wxl.to_excel(writer, sheet_name="sample", index=False)
writer.save()
writer.close()
print(f"write dataSize = {size}")
return True
def onDestroy(self):
super().onDestroy()
if __name__ == "__main__":
sw = SWImpl()
start = datetime(2014, 5, 1)
end = datetime(2020, 8, 17)
##查找有意义的k线形态
findKPatternCollector = Find_KPattern_skip1_predit2()
findKPatternCollector.print_on_destroy = True
#sw.collect(start, end,findKPatternCollector)
##打印更详细的信息
#printMoreDetail = More_detail_KPattern_skip1_predit2(kPatters=[712])
#printMoreDetail.print_on_destroy = True
#sw.collect(start, end,printMoreDetail)
##生成训练数据。
def generateSWTrainData(kPatterns: [], start: datetime,
end: datetime) -> pd.DataFrame:
sw = SWImpl()
lists = sw.getSW2List()
cloumns = [
"code", "name", "kPattern", "k", "d", "dif", "dea", "macd", "open",
"short", "long"
]
datas = []
kPatternMap = {}
for kPatternValues in kPatterns:
kPatternMap[kPatternValues] = True
macd_list = []
for code in lists:
# for code in lists:
name = sw.getSw2Name(code)
barList = sw.getSW2Daily(code, start, end)
indicator = Indicator(34)
preBar = None
for bar in barList:
##先识别形态
kEncodeValue = None
if indicator.inited:
tmpKEncodeValue = KPattern.encode3KAgo1(indicator)
if kPatternMap.__contains__(tmpKEncodeValue):
kEncodeValue = tmpKEncodeValue
if kEncodeValue is None:
indicator.update_bar(bar)
preBar = bar
continue
##昨天的kdj
k, d, j = indicator.kdj(array=False)
dif, dea, macd = indicator.macd(fast_period=12,
slow_period=26,
signal_period=9,
array=False)
##第二天的收益
short_pct = 100 * ((bar.high_price + bar.close_price) / 2 -
preBar.close_price) / preBar.close_price
long_pct = 100 * ((bar.low_price + bar.close_price) / 2 -
preBar.close_price) / preBar.close_price
open_pct = 100 * (bar.open_price -
preBar.close_price) / preBar.close_price
item = []
item.append(code)
item.append(name)
item.append(kEncodeValue)
item.append(k)
item.append(d)
item.append(dif)
item.append(dea)
item.append(macd)
#下个k线数据
item.append(open_pct)
item.append(short_pct)
item.append(long_pct)
datas.append(item)
macd_list.append(macd)
indicator.update_bar(bar)
preBar = bar
macd_list = np.array(macd_list)
print(
f"total size : {len(datas)},mean ={macd_list.mean()},max={macd_list.max()},min={macd_list.min()}"
)
wxl = pd.DataFrame(datas, columns=cloumns)
return wxl
def __init__(self, limit_close_pct=1, kPatters: [] = None):
super().__init__(limit_close_pct, kPatters)
self.traceDatas = []
self.sw = SWImpl()
self.predictTraceDatas: ['Skip1_Predict2_TraceData'] = []
def __init__(self):
self.sw = SWImpl()
def __init__(self):
self.lasted3Bar = np.array([None, None, None])
self.lasted3BarKdj = np.array([None, None, None])
self.sw = SWImpl()
if topData.code == topData2.code:
yestory_data = topData2
break
printTopData(topData)
printTopData(yestory_data, True)
print(f" {topData.predict.getLogInfo()}")
print(f"前天top数据")
for topData in yestoday2_top:
printTopData(topData)
print(f" {topData.predict.getLogInfo()}")
if __name__ == "__main__":
sw = SWImpl()
start = datetime(2014, 5, 1)
end = datetime(2019, 5, 1)
patternList = [
884, 709, 886, 1061, 710, 1062, 1063, 1238, 885, 535, 708, 929, 1060,
887, 1237, 1239, 711, 1236, 8630, 8629, 883, 534, 531, 1059, 533, 1413,
532, 707, 753
]
#patternList = [884, 709, 886, 1061, 710, 1062]
patternList = [886, 1061, 710, 885, 1060, 711, 1236, 1059]
from earnmi_demo.strategy_demo.kbars.analysis_KPattern_skip1_predit2 import \
Generate_Feature_KPattern_skip1_predit2, Skip1_Predict2_TraceData
##建立特征模型
