The most commonly used technical analysis tool in the futures and stock markets, the KDJ indicator, full name Stochastics, was created by Dr. George Lane. It combines the notion of momentum, strengths and weaknesses of some indicators. The KDJ indicator is calculated as the ratio between the highest, lowest and closing prices that have appeared in a given period.
RSVt = ((Ct-L9)/ ((H9-L9)) * 100 (Ct = closing price on the day; L9 = lowest price in 9 days; H9 = highest price in 9 days)
K is the RSV value of the 3-day smooth moving average, and the formula is: Kt = RSVt/3+2*t-1/3
Dt = Kt/3+2*Dt-1/3
The J value is 3 times the K value minus 2 times the D value, and the formula is:Dt-2Kt
There are several aspects to consider when applying the KDJ indicator:
1. The valuation of K and D, ranging from 0 to 100, 80+ markets exhibit overbought, 20+ exhibit overbought.
2.买进信号：K值在上涨趋势中﹤D值，K线向上突破D线时；卖出信号：K值在下跌趋势中﹥D值，K线向下跌破D线。
3.交易不活跃、发行量小的股票并不适用KD指标，而对大盘和热门大盘的准确性却很高。
4.在KD处在高位或低位，如果出现与股价走向的背离，则是采取行动的信号。
5.J's take>100 is overbought and <0 is oversold, both in the price abnormal range.
6.短期转势预警信号：K值和D值上升或者下跌的速度减弱，倾斜度趋于平缓
Generally, K, D and J are three values between 20 and 80 for the threshold range, and it is desirable to observe, in terms of sensitivity, the strongest is the J value, followed by K, the slowest is D, and in terms of safety, just the opposite.
import numpy as np
import pandas as pd
from pandas import DataFrame
import talib as ta
start = '2006-01-01' # 回测起始时间
end = '2015-08-17' # 回测结束时间
benchmark = 'HS300' # 策略参考标准
universe = set_universe('HS300')
capital_base = 100000 # 起始资金
refresh_rate = 1 # 调仓频率，即每 refresh_rate 个交易日执行一次 handle_data() 函数
longest_history=20
MA=[5,10,20,30,60,120] #移动均线参数
def initialize(account):
account.kdj=[]
def handle_data(account):
# 每个交易日的买入卖出指令
sell_pool=[]
hist = account.get_history(longest_history)
#data=DataFrame(hist['600006.XSHG'])
stock_pool,all_data=Get_all_indicators(hist)
pool_num=len(stock_pool)
if account.secpos==None:
print 'null'
for i in stock_pool:
buy_num=int(float(account.cash/pool_num)/account.referencePrice[i]/100.0)*100
order(i, buy_num)
else:
for x in account.valid_secpos:
if all_data[x].iloc[-1]['closePrice']<all_data[x].iloc[-1]['ma1'] and (all_data[x].iloc[-1]['ma1']-all_data[x].iloc[-1]['closePrice'])/all_data[x].iloc[-1]['ma1']>0.05 :
sell_pool.append(x)
order_to(x, 0)
if account.cash>500 and pool_num>0:
try:
sim_buy_money=float(account.cash)/pool_num
for l in stock_pool:
#print sim_buy_money,account.referencePrice[l]
buy_num=int(sim_buy_money/account.referencePrice[l]/100.0)*100
#buy_num=10000
order(l, buy_num)
except Exception as e:
#print e
pass
def Get_kd_ma(data):
indicators={}
#计算kd指标
indicators['k'],indicators['d']=ta.STOCH(np.array(data['highPrice']),np.array(data['lowPrice']),np.array(data['closePrice']),\
fastk_period=9,slowk_period=3,slowk_matype=0,slowd_period=3,slowd_matype=0)
indicators['ma1']=pd.rolling_mean(data['closePrice'], MA[0])
indicators['ma2']=pd.rolling_mean(data['closePrice'], MA[1])
indicators['ma3']=pd.rolling_mean(data['closePrice'], MA[2])
indicators['ma4']=pd.rolling_mean(data['closePrice'], MA[3])
indicators['ma5']=pd.rolling_mean(data['closePrice'], MA[4])
indicators['closePrice']=data['closePrice']
indicators=pd.DataFrame(indicators)
return indicators
def Get_all_indicators(hist):
stock_pool=[]
all_data={}
for i in hist:
try:
indicators=Get_kd_ma(hist[i])
all_data[i]=indicators
except Exception as e:
#print 'error:%s'%e
pass
if indicators.iloc[-2]['k']<indicators.iloc[-2]['d'] and indicators.iloc[-1]['k']>indicators.iloc[-2]['d']:
stock_pool.append(i)
elif indicators.iloc[-1]['k']>=10 and indicators.iloc[-1]['d']<=20 and indicators.iloc[-1]['k']>indicators.iloc[-2]['k'] and indicators.iloc[-2]['k']<indicators.iloc[-3]['k']:
stock_pool.append(i)
return stock_pool,all_data
Translated from Programmatic Trader