add new strategy from th666@joinquant & optimize select engine

DevilYuan.pyproj

@@ -300,6 +300,7 @@
     <Compile Include="Stock\Select\Strategy\Cta\DySS_SwingTrough.py">
       <SubType>Code</SubType>
     </Compile>
+    <Compile Include="Stock\Select\Strategy\Cta\DySS_Th666.py" />
     <Compile Include="Stock\Select\Strategy\Other\DySS_ETF.py">
       <SubType>Code</SubType>
     </Compile>

DyCommon/DyTalib.py

@@ -1,4 +1,9 @@
 import numpy as np
+import pandas as pd
+
+"""
+    Defaultly, in/out paramters are  numpy array.
+"""
 
 
 DyTalibExpAdjust = True # pandas默认是True，同花顺默认是False。True时，最近的值权重会大些。数据足够多的时候，趋向一致。
@@ -148,4 +153,45 @@ def WATR(highs, lows, closes, W, timeperiod=14):
 
     atr.insert(0, np.nan)
 
-    return atr
+    return atr
+
+def HHV(v, N):
+    """
+        @v: numpy array or list
+        @return: numpy array
+    """
+    return pd.Series(v).rolling(N).max().values
+
+
+def LLV(v, N):
+    """
+        @v: numpy array or list
+        @return: numpy array
+    """
+    return pd.Series(v).rolling(N).min().values
+
+def KDJ(highs, lows, closes, N=9, M1=3, M2=3, adjust=DyTalibExpAdjust):
+    """
+        @closes, highs, lows: numpy array
+        @return: numpy array
+    """
+    C = closes
+    H = highs
+    L = lows
+
+    RSV = (C - LLV(L, N)) / (HHV(H, N) - LLV(L, N)) * 100
+    RSV = RSV[~np.isnan(RSV)]
+    K = SMA(RSV, N, M1, adjust=adjust)
+    D = SMA(K, N, M2, adjust=adjust)
+    J = 3 * np.array(K) - 2 * np.array(D)
+    return K, D, J
+
+def SINGLE_GOLDEN_CROSS(A, B):
+    if A[-2] < B[-2] and A[-1] > B[-1]:
+        return True
+
+    return False
+
+def GOLDEN_CROSS(A, B):
+    var = np.where(A < B, 1, 0)
+    return (pd.Series(var).diff() < 0).values

Stock/Select/Engine/DyStockSelectSelectEngine.py

@@ -1,3 +1,5 @@
+import traceback
+
 from ...Data.Engine.DyStockDataEngine import *
 from EventEngine.DyEvent import *
 from ..DyStockSelectCommon import *
@@ -209,7 +211,7 @@ def _runDaysLoop(self):
         self._info.print("开始运行日线数据...")
 
         # init progress
-        self._progress.init(len(self._daysEngine.stockAllCodes), 100, 5)
+        self._progress.init(len(self._daysEngine.stockAllCodesFunds), 100, 5)
 
         # index loop
         for index in self._daysEngine.stockIndexes:
@@ -219,6 +221,14 @@ def _runDaysLoop(self):
 
             self._progress.update()
 
+        # ETF loop
+        for etfCode in self._daysEngine.stockFunds:
+            df = self._daysEngine.getDataFrame(etfCode, self._startDay, self._endDay)
+            if df is not None:
+                self._strategy.onEtfDays(etfCode, df)
+
+            self._progress.update()
+
         # stock loop
         for code in self._daysEngine.stockCodes:
             df = self._daysEngine.getDataFrame(code, self._startDay, self._endDay)
@@ -282,7 +292,8 @@ def _run(self):
         try:
             self._strategy.onInit(self._dataEngine, self._errorDataEngine)
         except Exception as ex:
-            self._info.print('策略onInit异常', DyLogData.error)
+            traceback.print_exc()
+            self._info.print('策略onInit异常: {}'.format(ex), DyLogData.error)
             return False
 
         # run loop

Stock/Select/Strategy/Cta/DySS_Th666.py

@@ -0,0 +1,172 @@
+import operator
+
+from ..DyStockSelectStrategyTemplate import *
+from ....Data.Utility.DyStockDataUtility import *
+from DyCommon.DyTalib import *
+
+
+class DySS_Th666(DyStockSelectStrategyTemplate):
+    """
+        https://www.joinquant.com/post/15297?tag=algorithm
+        1.首先趋势判断，中证500趋势判断，日线级别kdj金叉多区间，上一交易日为大阴线3/5实体（并且50日atr大于2/3平均值），obos（超买超卖指标）小于-100，满足条件则选股
+        2.选股9：25-9：30之间高开五个点以上（包含），不含st，开盘前五秒上涨，选出并打印
+
+        魔元注:
+            由于obos是绝对值，是不是可以改成相对值? 即 OBOS=(N日内上涨家数移动总和－N日内下跌家数移动总和)/N日内上涨下跌家数总和 * 100
+
+            OBOS=N日内上涨家数移动总和－N日内下跌家数移动总和
+            N日的采样统计一般设定为10日。
+    """
+    name = 'DySS_Th666'
+    chName = 'th666聚宽求助'
+
+    autoFillDays = True
+    optimizeAutoFillDays = True
+    continuousTicks = True
+
+    colNames = ['代码', '名称']
+
+    param = OrderedDict\
+                ([
+                    ('基准日期', '2018-11-15'),
+                    ('中证500: 上一交易日阴线实体比例>=', 0.1), # 0.6
+                    ('中证500: 50日ATR/平均值>', 0.67),
+                    ('中证500: OBOS<', 1000), # -100
+                ])
+
+
+    def __init__(self, param, info):
+        super().__init__(param, info)
+
+        # unpack parameters
+        self._baseDate              = param['基准日期']
+        self._bodyRatio             = param['中证500: 上一交易日阴线实体比例>=']
+        self._atrOverMean           = param['中证500: 50日ATR/平均值>']
+        self._obos                  = param['中证500: OBOS<']
+
+        self._canContinue = True
+        self._preCloses = {}
+
+    def onDaysLoad(self):
+        return self._baseDate, -50-14
+
+    def onTicksLoad(self):
+        return self._baseDate, 0
+
+    def _get500(self, daysEngine):
+        # 选取中小创流通市值前500只
+        codes = {}
+        for code in daysEngine.stockCodes:
+            # 由于DY里没有中证500成分股信息，所以用中小创代替
+            indexCode = DyStockCommon.getIndex(code)
+            if indexCode not in [DyStockCommon.cybIndex, DyStockCommon.zxbIndex]:
+                continue
+
+            df = self._daysEngine.getDataFrame(code)
+            if df is None:
+                continue
+
+            codes[code] = df['amt'][-1]/df['turn'][-1]
+
+        sortedCodes = sorted(codes, key=lambda k: codes[k], reverse=True)
+        return sortedCodes[:500]
+
+    def onInit(self, dataEngine, errorDataEngine):
+        print("onInit...")
+
+        self._daysEngine = dataEngine.daysEngine
+        self._stockAllCodes = self._daysEngine.stockAllCodes
+
+        selectedCodes = self._get500(self._daysEngine)
+
+        # obos（超买超卖指标）小于-100
+        N = 10
+
+        # date range
+        df = self._daysEngine.getDataFrame(DyStockCommon.szIndex)
+        dateRange = df[:self._baseDate].tail(N+1).index
+        startDate, endDate = dateRange[0].strftime("%Y-%m-%d"), dateRange[-1].strftime("%Y-%m-%d")
+
+        zeros = pd.Series([0]*N, index=df[:self._baseDate].tail(N).index)
+        ups = zeros
+        downs = zeros
+        for code in selectedCodes:
+            df = self._daysEngine.getDataFrame(code)
+            if df is None:
+                continue
+
+            closes = df['close'][startDate:endDate]
+            pctChanges = closes.pct_change().dropna()
+
+            # 上涨
+            temp = zeros + pctChanges.apply(lambda x: 1 if x > 0 else 0)
+            ups = ups + temp.fillna(0)
+
+            # 下跌
+            temp = zeros + pctChanges.apply(lambda x: 1 if x < 0 else 0)
+            downs = downs + temp.fillna(0)
+
+        obos = ups.sum() - downs.sum()
+        if obos >= self._obos:
+            print("中证500: OBOS: {}".format(obos))
+            self._canContinue = False
+
+    def onEtfDays(self, code, df):
+        """ Etf日线数据 """
+        if not self._canContinue:
+            return
+
+        # 日线级别kdj金叉多区间
+        if code == DyStockCommon.etf500:
+            K, D, J = KDJ(df['high'].values, df['low'].values, df['close'].values, adjust=False)
+            if K[-1] < D[-1]:
+                print("中证500: KDJ不是金叉区间")
+                self._canContinue = False
+                return
+
+            # 上一交易日为大阴线3/5实体
+            if df['close'][-2] >= df['open'][-2]:
+                print("中证500: 上一交易日不是阴线")
+                self._canContinue = False
+                return
+
+            bodyRatio = abs(df['close'][-2] - df['open'][-2])/abs(df['low'][-2] - df['high'][-2])
+            if bodyRatio < self._bodyRatio:
+                print("中证500: 上一交易日阴线实体比例: {}".format(bodyRatio))
+                self._canContinue = False
+                return
+
+            # 50日atr大于2/3平均值
+            atr = np.array(ATR(df['high'].values, df['low'].values, df['close'].values))
+            atrOverMean = float(atr[-1]/atr[-50:].mean())
+            if atrOverMean <= self._atrOverMean:
+                print("中证500: 50日atr/平均值: {}".format(atrOverMean))
+                self._canContinue = False
+                return
+
+    def onStockDays(self, code, df):
+        if not self._canContinue:
+            return
+
+        # 选股9：25-9：30之间高开五个点以上（包含），不含st
+        openIncrease = (df['open'][-1] - df['close'][-2])/df['close'][-2] * 100
+        if openIncrease < 5:
+            return
+
+        if 'st' in self._stockAllCodes[code] or 'ST' in self._stockAllCodes[code]:
+            return
+
+        self._preCloses[code] = df['close'][-2]
+
+        # 设置结果
+        row = [code, self._stockAllCodes[code]]
+        print(row)
+        self._result.append(row)
+
+    def onStockTicks(self, code, dfs):
+        # 开盘前五秒上涨
+        prices = dfs['price']
+        date = dfs.index[0].strftime("%Y-%m-%d")
+        if self._preCloses[code] >= prices[:"{} 09:30:05".format(date)][-1]:
+            print("Remove {}".format([code, self._stockAllCodes[code]]))
+            self.removeFromResult(code)

Stock/Select/Strategy/DyStockSelectStrategyTemplate.py

@@ -123,6 +123,10 @@ def onIndexDays(self, code, df):
         """ 指数日线数据 """
         pass
 
+    def onEtfDays(self, code, df):
+        """ ETF日线数据 """
+        pass
+
     def onStockDays(self, code, df):
         """ 个股日线数据 """
         pass