MACD

Moving Average Convergence/Divergence，意为异同移动平均线。它刻画的是股价变化的速度

MACD算法

指标	含义	公式
短期EMA	短期收盘价指数移动均线(12天)	前一日EMA(12)11/13 +今日收盘价2/13
长期EMA	长期收盘价指数移动均线(26天)	前一日EMA(26)25/27 +今日收盘价2/27
DIF	短期EMA与长期EMA差值	EMA(12)-EMA(26)
DEA	DIF线的M日指数移动均钱	前一日DEA8/10 +今日DEA2/10
MACD	DIF线与DEA线的差*2	(DIF-DEA)*2

MACD实现

ewm
Pandas中指数加权移动窗口函数采用ewm函数+mean（）快捷计算MACD
bar
Matplotlib柱状图函数，高效绘制MACD中的柱状

class TestMACD(TestCase):def cal_macd(self,df,fastperiod=12,slowperiod=26,signalperiod=9):ewma12 = df['close'].ewm(span=fastperiod,adjust=False).mean()ewma26 = df['close'].ewm(span=slowperiod,adjust=False).mean()df['dif'] = ewma12 - ewma26df['dea'] = df['dif'].ewm(span=signalperiod,adjust=False).mean()df['bar'] = (df['dif'] - df['dea']) * 2return dfdef test_MACD(self):file_name = "D:\lhjytest\demo.csv"df = pd.read_csv(file_name)df.columns = ["stock_id","date","close","open","high","low","volume"]df = df[["date","close","open","high","low","volume"]]df["date"] = pd.to_datetime(df["date"])df_macd = self.cal_macd(df)print(df_macd)plt.figure()df_macd['dea'].plot(color="red",label='dea')df_macd['dif'].plot(color="blue",label='dif')plt.legend(loc='best')pos_bar = []pos_index = []neg_bar = []neg_index = []for index,row in df_macd.iterrows():if(row['bar'] > 0) :pos_bar.append(row['bar'])pos_index.append(index)else:neg_bar.append(row['bar'])neg_index.append(index)# 大于0用红色表示plt.bar(pos_index,pos_bar,width=0.5,color='red')# 小于等于0则用绿色表示plt.bar(neg_index,neg_bar,width=0.5,color='green')major_index = df_macd.index[df_macd.index]major_xtics = df_macd['date'][df_macd.index]plt.xticks(major_index,major_xtics)plt.setp(plt.gca().get_xticklabels(),rotation=30)plt.grid(linestyle='-.')plt.title('000001平安银行MACD图')plt.rcParams['axes.unicode_minus'] = Falseplt.rcParams['font.sans-serif'] = ['SimHei']plt.show()

KDJ

中文名叫随机指数。通过价格波动的真实波幅来反映价格走势的强弱和超买超卖现象，在价格尚未上升或下降之前发出买卖信号的一种技术分析指标,适用于短期行情走势分析

KDJ算法

指标	含义	公式
RSV	未成熟随机指标值	(Cn-Ln)/(Hn-Ln)x100(N日)
K	当天K值	2/3x前一日K值+1/3x当日RSV
D	当天D值	2/3x前一日D值+1/3x当日K值
J	当天J值	3当日K值-2当日D值
备注	若无前一日K 值与D值，则可分别用50来代替

KDJ实现

rolling
Pandas中移动窗口函数
每个窗口都是指定的固定大小，快捷计算Ln与Hn
expanding
Pandas中扩展窗口函数
只设置最小的观测值数量，不固定窗口大小，实现累计计算，即不断扩展，连用expanding().max()->创新高

class TestKDJ(TestCase):def cal_kdj(self,df):low_list = df['low'].rolling(9,min_periods=9).min()low_list.fillna(value=df['low'].expanding().min(),inplace=True)high_list = df['high'].rolling(9,min_periods=9).max()high_list.fillna(value=df['high'].expanding().max(),inplace=True)rsv = (df['close'] - low_list) / (high_list - low_list) * 100df['k'] = pd.DataFrame(rsv).ewm(com=2).mean()df['d'] = df['k'].ewm(com=2).mean()df['j'] = 3 * df['k'] - 2 * df['d']return dfdef test_KDJ(self):file_name = "D:\lhjytest\demo.csv"df = pd.read_csv(file_name)df.columns = ["stock_id","date","close","open","high","low","volume"]df = df[["date","close","open","high","low","volume"]]df["date"] = pd.to_datetime(df["date"])df_kdj = self.cal_kdj(df)print(df_kdj)plt.figure()df_kdj['k'].plot(color="red",label='k')df_kdj['d'].plot(color="yellow",label='d')df_kdj['j'].plot(color="blue",label='j')plt.legend(loc='best')major_index = df_kdj.index[df_kdj.index]major_xtics = df_kdj['date'][df_kdj.index]plt.xticks(major_index,major_xtics)plt.setp(plt.gca().get_xticklabels(),rotation=30)