国信证券学习系列（1）

        软件不错，满足了我对股票，期货，期权的全部要求。而且数据可以提供下载，简直没话说了。

        数据清洗问题，我其实很早以前就在思考这个问题，回测，到底在测什么？什么样的数据可以用来回测？什么样的回测才是好的回测，才是可以拿来实盘的？说什么tick回测更精准，我到哪去找？我交易也没指望那么高频率，误差出现，我能hold住就行。

不管了，就这样，默认他的数据有用，直接上来干吧！

        先做多因子选股策略：

        择时函数：

def signal(ContextInfo):
    buy = {i:0 for i in ContextInfo.s}
    sell = {i:0 for i in ContextInfo.s}
    data_high = ContextInfo.get_history_data(22,'1d','high',3)
    data_high_pre = ContextInfo.get_history_data(2,'1d','high',3)
    data_close60 = ContextInfo.get_history_data(62,'1d','close',3)
    #print data_high
    #print data_close
    #print data_close60
#选择买入点
    for k in ContextInfo.s:
        if k in data_close60:
            if len(data_high_pre[k]) == 2 and len(data_high[k]) == 22 and len(data_close60[k]) == 62:
                if data_high_pre[k][-2] > max(data_high[k][:-2]):
                    buy[k] = 1           #超过20日最高价，加入买入备选
                elif data_high_pre[k][-2] < np.mean(data_close60[k][:-2]):
                    sell[k] = 1           #低于60日均线，加入卖出备选
    #print buy
    #print sell
    return buy,sell           #买入卖出备选

获取数据排序 

for k in list(buys.keys()):
    if buys[k] == 1:
#找因子
        rank1[k] = ext_data_rank('atr',k[-2:]+k[0:6],0,ContextInfo)
        rank2[k] = ext_data_rank('adtm',k[-2:]+k[0:6],0,ContextInfo)
        #print rank1[k], rank2[k]
        rank_total[k] = 0.5 * rank1[k]- 0.5 * rank2[k]    #因子的权重需要人为设置，此处取了0.5
        print (1111111, rank1[k])

策略代码

#coding:gbk
"""
回测模型示例（非实盘交易策略）
#HS300日线下运行，20个交易日进行 一次调仓，每次买入在买入备选中因子评分前10的股票，每支股票各分配当前可用资金的10%（权重可调整）
#扩展数据需要在补完HS300成分股数据之后生成，本模型中扩展数据暂时使用VBA指标ATR和ADTM生成，命名为atr和adtm
"""
import pandas as pd
import numpy as np
import time
import datetime
#对所有股票进行择时交易
def signal(ContextInfo):
    buy = {i:0 for i in ContextInfo.s}
    sell = {i:0 for i in ContextInfo.s}
    data_high = ContextInfo.get_history_data(22,'1d','high',3)
    data_high_pre = ContextInfo.get_history_data(2,'1d','high',3)
    data_close60 = ContextInfo.get_history_data(62,'1d','close',3)
    #print data_high
    #print data_close
    #print data_close60
    for k in ContextInfo.s:
        if k in data_close60:
            if len(data_high_pre[k]) == 2 and len(data_high[k]) == 22 and len(data_close60[k]) == 62:
                if data_high_pre[k][-2] > max(data_high[k][:-2]):
                    buy[k] = 1           #超过20日最高价，加入买入备选
                elif data_high_pre[k][-2] < np.mean(data_close60[k][:-2]):
                    sell[k] = 1           #低于60日均线，加入卖出备选
    #print buy
    #print sell
    return buy,sell           #买入卖出备选

def init(ContextInfo):
    ContextInfo.s = ContextInfo.get_sector('000300.SH')
    ContextInfo.set_universe(ContextInfo.s)
    ContextInfo.day = 0
    ContextInfo.holdings = {i:0 for i in ContextInfo.s}
    ContextInfo.weight = [0.1]*10         #设置资金分配权重
    ContextInfo.buypoint = {}
    ContextInfo.money = ContextInfo.capital
    ContextInfo.profit = 0
    ContextInfo.accountID='testS'
#获取发出买入信号的股票，进行因子评级
def handlebar(ContextInfo):
    rank1 = {}
    rank2 = {}
    rank_total = {}
    tmp_stock = {}
    d = ContextInfo.barpos
    price = ContextInfo.get_history_data(1,'1d','open',3)
    if d > 60 and d % 20 == 0:               #每月一调仓
        nowDate = timetag_to_datetime(ContextInfo.get_bar_timetag(d),'%Y%m%d')
        print(nowDate)
        buys, sells = signal(ContextInfo)
        order = {}
        for k in list(buys.keys()):
            if buys[k] == 1:
                rank1[k] = ext_data_rank('atr',k[-2:]+k[0:6],0,ContextInfo)
                rank2[k] = ext_data_rank('adtm',k[-2:]+k[0:6],0,ContextInfo)
                #print rank1[k], rank2[k]
                rank_total[k] = 0.5 * rank1[k]- 0.5 * rank2[k]           #因子的权重需要人为设置，此处取了0.5和-0.5
                print (1111111, rank1[k])
        tmp = sorted(list(rank_total.items()), key = lambda item:item[1])
        #print tmp
        if len(tmp) >= 10:
            tmp_stock = {i[0] for i in tmp[:10]}
        else:
            tmp_stock = {i[0] for i in tmp}                              #买入备选中若超过10只股票则选10支，不足10支则全选
        for k in list(buys.keys()):
            if k not in tmp_stock:
                buys[k] = 0
        if tmp_stock:
            print('stock pool:',tmp_stock)
            for k in ContextInfo.s:
                if ContextInfo.holdings[k] > 0 and sells[k] == 1:
                    print('ready to sell')
                    order_shares(k,-ContextInfo.holdings[k]*100,'fix',price[k][-1],ContextInfo,ContextInfo.accountID)
                    ContextInfo.money += price[k][-1] * ContextInfo.holdings[k] * 100 - 0.0003*ContextInfo.holdings[k]*100*price[k][-1]                  #手续费按万三设定
                    ContextInfo.profit += (price[k][-1]-ContextInfo.buypoint[k]) * ContextInfo.holdings[k] * 100 - 0.0003*ContextInfo.holdings[k]*100*price[k][-1]
                    #print price[k][-1]
                    print(k)
                    #print ContextInfo.money
                    ContextInfo.holdings[k] = 0
            ContextInfo.money_distribution = {k:i*ContextInfo.money for (k,i) in zip(tmp_stock,ContextInfo.weight)}
            for k in tmp_stock:
                if ContextInfo.holdings[k] == 0 and buys[k] == 1:
                    print('ready to buy')
                    order[k] = int(ContextInfo.money_distribution[k]/(price[k][-1]))/100
                    order_shares(k,order[k]*100,'fix',price[k][-1],ContextInfo,ContextInfo.accountID)
                    ContextInfo.buypoint[k] = price[k][-1]
                    ContextInfo.money -= price[k][-1] * order[k] * 100 - 0.0003*order[k]*100*price[k][-1]
                    ContextInfo.profit -= 0.0003*order[k]*100*price[k][-1]
                    print(k)
                    ContextInfo.holdings[k] = order[k]
            print(ContextInfo.money,ContextInfo.profit,ContextInfo.capital)
    profit = ContextInfo.profit/ContextInfo.capital
    if not ContextInfo.do_back_test:
        ContextInfo.paint('profit_ratio', profit, -1, 0)