python-双均线策略的预测能力研究

做量化，不仅要懂编程，有交易经验，而且要数学和统计学要好。恶补统计理论中，好后悔上统计课的时候做白日梦了。。。。。。。。。。。

下面试着尝试了分析双均线能否产生预测力，我们知道很多回测产生的盈利，有时候并不是策略本身的问题，有可能是数据的趋势。本文尝试通过消除平均趋势或者采用随机基准来对比策略的收益，似乎，双均线蛮好。

# -*- coding: utf-8 -*-
"""
Created on Tue Jun 6 14:50:04 2017

@author: yunjinqi

E-mail:yunjinqi@qq.com

Differentiate yourself in the world from anyone else.
"""
# -*- coding: utf-8 -*-
"""
Created on Thu May 25 08:55:12 2017

@author: yunjinqi

E-mail:yunjinqi@qq.com

Differentiate yourself in the world from anyone else.
"""
import pandas as pd
import numpy as np
import datetime
import time
#获取数据
df=pd.read_csv('C:/Users/HXWD/Desktop/rb000.csv',encoding='gbk')
df.head()
df.columns=['date','open','high','low','close','volume','amt']
df.head()
value=[]
for i in range(5,6):
for j in range(20,21):
df['ma5']=df['close'].rolling(i).mean()
df['ma20']=df['close'].rolling(j).mean()
df.ix[df['ma5']>df['ma20'],'cross']=1
df.ix[df['ma5']<=df['ma20'],'cross']=-1
#df[['close','ma5','ma20']][-200:].plot()
df['ret']=(df['close']-df['close'].shift(1))
df['profit']=df['ret']*df['cross']
#df['profit'].plot()
target=df['profit'].sum()
s=[i,j,target]
ts=time.strftime('%Y-%m-%d %X', time.localtime() )
value.append(s)
print('当前时间:{}短期参数:{},长期参数:{}优化完毕,净利润{}'.format(ts,i,j,s))
data=pd.DataFrame(value)
#data.to_csv('参数优化.csv')
#消除趋势后的
df_up=df.cross[df.cross>0]
df_down=df.cross[df.cross<0]
p=len(df_up)/(len(df_down)+len(df_up))
ll=len(df)
mu=p*len(df)
mu

import random
suiji_value=[]
for i in range(10000):
total=[]
for i in range(len(df)):
v=random.randint(0,ll)
if v<mu:
total.append(1)
if v>=mu:
total.append(-1)
total=pd.Series(total)
suiji_profit=(df['ret']*total).sum()
print(suiji_profit)
suiji_value.append(suiji_profit)
suiji_value=pd.Series(suiji_value).sum()/10000
suiji_value
#随机信号只能够产生平均83.9的收益，而双均线能够产生15593的利润，
#有理由相信这是一个好策略
#我们尝试消除平均趋势之后再计算相应的盈利
df['new_ret']=(df['close']-df['close'].shift(1))/df['close'].shift(1)
aa=df['new_ret'].mean()
df['new_ret1']=df['new_ret']-pd.Series([aa]*len(df))
df['ret2']=df['new_ret1']*df['close'].shift(1)
(df['ret2']*df['cross']).sum()
#计算出来消除平均趋势之后的收益15571，只差了20多。
#似乎均线策略，马马虎虎证明了，是可以产生预测能力的。