【干货转载】：一个利用JQ做指标监控的例子

接到一个朋友的需要，做一个简单的指标监控，就是当多个周期的均线和ķ线满足某种情况的时候发出信号。其实这种预警很多图表操盘软件也可以做，但是问题是多个品种监控的时候效率非常低，还容易死机，同时监控信息的发出手段不灵活，例如不方便发到QQ或者微信群里面于是想到可以通过蟒蛇来做这个事情。
主程序如下：

Mon = Monitor()
Mon.addRule(RuleResonanceUp())

with open("flist.txt",'r') as f:
    flist = f.readlines()
    for line in flist:
        print("Processing "+line)
        bd = BarData(line.strip('\n'))
        Mon.Run(bd)

初始化一个监控器监控器，并给这个监控器里面放入所需要监控的规则类RuleResonanceUp 
所有品种的代码放在一个flist.txt文件中，读出来以后由BarData这个类进行初始化，获得多周期的交易数据，作为监控器要处理的数据上下文。
然后运行监控器，监控器会逐一将规则套入上下文数据中进行判断，满足条件的就发出消息通知。

 

其他的代码：

from jqdatasdk import *
import talib    #用来做指标计算，算是最流行的库了吧
import matplotlib.pyplot as plt     #用来做图示，这个是调试用的，调好后基本就不用了
from matplotlib.pylab import date2num     #用来做图示的，调试用，调好后就基本不用了
import mpl_finance as mpf         #用来图示的。
import matplotlib.ticker as ticker      #用来改图示的横坐标
import pandas as pd       
import pandas.io.excel       #用来输出文件用的
import datetime
import numpy as np

class BarData(object):       #用来获取品种的实际数据
    def __init__(self, stockCode, enddate=None):
        self.stockCode = stockCode
        self.endDate = enddate
        self.get_560Bars()
        self.apply_indicators()

    def get_560Bars(self):      #这个系统用到周线日线小时和5分钟
        self.df1w = get_bars(self.stockCode, 100, unit='1w', fields=['date', 'open', 'high', 'low', 'close'], include_now=True,
                      end_dt = self.endDate)
        self.df1d = get_bars(self.stockCode, 100, unit='1d', fields=['date', 'open', 'high', 'low', 'close'], include_now=True,
                      end_dt = self.endDate)
        self.df60m = get_bars(self.stockCode, 100, unit='60m', fields=['date', 'open', 'high', 'low', 'close'], include_now=True,
                        end_dt=self.endDate)
        self.df5m = get_bars(self.stockCode, 100, unit='5m', fields=['date', 'open', 'high', 'low', 'close'], include_now=True,
                        end_dt=self.endDate)

    def apply_indicators(self):   #这个系统用到20  60  和经典参数的macd
        def apply_ma60(df):
            df["MA60"] = talib.MA(df["close"], timeperiod=60)
        def apply_ma20(df):
            df["MA20"] = talib.MA(df["close"], timeperiod=20)
        def apply_macd(df):
            #close = [float(x) for x in df['close']]
            df["DIFF"], df['DEA'], df['MACDhist'] = talib.MACD(df['close'], fastperiod=12, slowperiod=26, signalperiod=9)
        def apply_hist_signals(df):
            """

            :type df: Dataframe
            """
            #在原来的dataframe里面加入所有历史的信号，每个信号加一列
            #需要支持的值是： c-ma60, ma60Slope, c-ma20, macdGold, macdDead, macdGoldsoon, macdDeadsoon, DEASlope
            df['c_minus_ma60'] = df['close'] - df["MA60"]   #用来判断是否在ma60之上

            df['ma60Slope'] = None
            for i in range(1,len(df)):
                df.loc[i,'ma60Slope'] = df["MA60"][i] - df["MA60"][i-1]
             #用来判断ma60的斜率

            df['c_minus_ma20'] =  df['close'] - df["MA20"]   #用来判断是否在ma20之上

            df["macdGold"] = False                          # macd 金叉死叉
            df["macdDead"] = False
            for i in range(len(df)):
                if (df["DIFF"][i] - df["DEA"][i]) > 0 :
                    df.loc[i,"macdGold"] = True
                if (df["DIFF"][i] - df["DEA"][i]) < 0 :
                    df.loc[i,"macdDead"] = True

            df["macdGoldsoon"] = None                       # macd 将要金叉死叉
            df["macdDeadsoon"] = None
            for i in range(len(df)-2) :
                if df["DEA"][i+1] > df["DIFF"][i+1]:
                    if (df["DEA"][i+1] - df["DIFF"][i+1])-((df["DEA"][i]-df["DIFF"][i]) - (df["DEA"][i+1]-df["DIFF"][i+1])) < 0:
                        df.loc[i+2,"macdGoldsoon"] = True

                if df["DIFF"][i + 1] > df["DEA"][i + 1]:
                    if (df["DIFF"][i + 1] - df["DEA"][i + 1]) - (
                            (df["DIFF"][i] - df["DEA"][i]) - (df["DIFF"][i + 1] - df["DEA"][i + 1])) < 0:
                        df.loc[i+2,"macdDeadsoon"] = True


            df['DEASlope'] = None
            for i in range(1, len(df)):
                df.loc[i, 'DEASlope'] = df["DEA"][i] - df["DEA"][i - 1]

            df["KDir"] = 0
            for i in range(1,len(df)):                  # K线方向判断，如果高点抬高 低点抬高就多，高点低点降低就空，内包外包就保持
                if (df["high"][i] > df["high"][i-1]) and (df["low"][i] > df["low"][i-1]):
                    df.loc[i,"KDir"] = 1
                elif df["high"][i] < df["high"][i-1] and df["low"][i] < df["low"][i-1]:
                    df.loc[i,"KDir"] = -1
                else :
                    df.loc[i,"KDir"] = df["KDir"][i-1]

        for dft in [self.df5m, self.df60m, self.df1d, self.df1w]:
            apply_ma60(dft)
            apply_ma20(dft)
            apply_macd(dft)
            apply_hist_signals(dft)

class Rule(object):    #预警规则的基类
    def __init__(self):
        self.name = "RuleBase"

    def ApplyRule(self, bd):
        pass

    def Msg(self):    #当规则发生的时候，要发出来的消息
        return "Non Rule Implemented."

class RuleResonanceUp(Rule):            #多头共振的规则，所有的周期都是多的时候提醒
    def __init__(self):
        self.name = "RuleResonaceUp"
    def ApplyRule(self, bd):
        if bd.df1w["KDir"].tolist()[-1] > 0 and bd.df1d.KDir.tolist()[-1] > 0 and bd.df1d.DEASlope.tolist()[-1] > 0 and bd.df60m.KDir.tolist()[-1] > 0 \
            and bd.df60m.DEASlope.tolist()[-1] > 0 and bd.df5m.ma60Slope.tolist()[-1] > 0 and bd.df5m.DEASlope.tolist()[-1] > 0 \
            and bd.df5m.KDir.tolist()[-1] >0:
            return True
        else:
            return False

    def Msg(self):
        return "周K，日K，日DEA，日K，小时DEA，小时K，5mDEA、5mMA60、5mK，多方共振。"


class Monitor(object):
    def __init__(self):
        self.Rules = []
    def addRule(self,r):
        self.Rules += [r]

    def Run(self,bd):      #执行规则，满足规则的时候打印提醒信息，在这里，可以换成ichat或者pyqq啥的，就可以实现微信和qq的通知了。
        for r in self.Rules:
            if r.ApplyRule(bd):
                print(datetime.datetime.now()," "+get_security_info(bd.stockCode).display_name+': '+r.Msg())
    pass

以上就是一个简单的利用jqdata进行多品种均线情况监控的规则框架。

 

原文链接：https://www.joinquant.com/view/community/detail/16682

原文作者：聚宽/fwoods