用Python实现一个最简单的量化工具(转载)
原文地址: http://blog.csdn.net/robertsong2004/article/details/50643198
在之前的blog里写了一个最简单的例子:
http://blog.csdn.net/robertsong2004/article/details/50642655
现在试一下简单的分析,即设定一个策略:以20日线为标准,当前股价低于20日线的时候就卖出,高于20日线的时候就买入。
然后计算一下这个策略的效果。
主要用 TuShare 里的 get_hist_data 这个接口,用于获取到目前为止3年的历史数据。
主要的用法参照如下:
获取个股历史交易数据(包括均线数据),可以通过参数设置获取日k线、周k线、月k线,以及5分钟、15分钟、30分钟和60分钟k线数据。本接口只能获取近3年的日线数据,适合搭配均线数据进行选股和分析,如果需要全部历史数据,请调用下一个接口get_h_data()。
参数说明:
- code:股票代码,即6位数字代码,或者指数代码(sh=上证指数 sz=深圳成指 hs300=沪深300指数 sz50=上证50 zxb=中小板 cyb=创业板)
- start:开始日期,格式YYYY-MM-DD
- end:结束日期,格式YYYY-MM-DD
- ktype:数据类型,D=日k线 W=周 M=月 5=5分钟 15=15分钟 30=30分钟 60=60分钟,默认为D
- retry_count:当网络异常后重试次数,默认为3
- pause:重试时停顿秒数,默认为0
返回值说明:
- date:日期
- open:开盘价
- high:最高价
- close:收盘价
- low:最低价
- volume:成交量
- price_change:价格变动
- p_change:涨跌幅
- ma5:5日均价
- ma10:10日均价
- ma20:20日均价
- v_ma5:5日均量
- v_ma10:10日均量
- v_ma20:20日均量
- turnover:换手率[注:指数无此项]
那我们只要比较 v_ma20 和 close 这两个值就可以了。
OK,写了一个很简单的程序:
1 #!/usr/bin/python 2 # coding: UTF-8 3 4 """This script parse stock info""" 5 6 import tushare as ts 7 8 def parse(code_list): 9 '''process stock''' 10 is_buy = 0 11 buy_val = [] 12 buy_date = [] 13 sell_val = [] 14 sell_date = [] 15 df = ts.get_hist_data(STOCK) 16 ma20 = df[u'ma20'] 17 close = df[u'close'] 18 rate = 1.0 19 idx = len(ma20) 20 21 while idx > 0: 22 idx -= 1 23 close_val = close[idx] 24 ma20_val = ma20[idx] 25 if close_val > ma20_val: 26 if is_buy == 0: 27 is_buy = 1 28 buy_val.append(close_val) 29 buy_date.append(close.keys()[idx]) 30 elif close_val < ma20_val: 31 if is_buy == 1: 32 is_buy = 0 33 sell_val.append(close_val) 34 sell_date.append(close.keys()[idx]) 35 36 print "stock number: %s" %STOCK 37 print "buy count : %d" %len(buy_val) 38 print "sell count : %d" %len(sell_val) 39 40 for i in range(len(sell_val)): 41 rate = rate * (sell_val[i] * (1 - 0.002) / buy_val[i]) 42 print "buy date : %s, buy price : %.2f" %(buy_date[i], buy_val[i]) 43 print "sell date: %s, sell price: %.2f" %(sell_date[i], sell_val[i]) 44 45 print "rate: %.2f" % rate 46 47 if __name__ == '__main__': 48 STOCK = '600000' ##浦发银行 49 parse(STOCK)
主要就是调 parse 这个函数,这个函数就会从数据中读 20 日线的值和当日收盘价做对比,如果当日收盘价比20日线值高且之前没有买入,那就买入。
如果当日收盘价比20日线值低且之前已经买入,那就卖出。
然后加了一个 rate 参数,这是用于统计每次买卖后的收益。这里有个0.002的值,表示两块一个是印花税卖出后有1%的印花税要抽,一个是券商的佣金,按万5来算,
一来一回也是千分之一。
跑一下结果如下,
stock number: 600000 buy count : 39 sell count : 38 buy date : 2013-02-28, buy price : 11.06 sell date: 2013-03-04, sell price: 10.39 buy date : 2013-03-05, buy price : 11.07 sell date: 2013-03-07, sell price: 10.76 buy date : 2013-03-19, buy price : 10.65 sell date: 2013-03-26, sell price: 10.70 buy date : 2013-04-22, buy price : 10.22 sell date: 2013-04-23, sell price: 9.84 buy date : 2013-05-03, buy price : 10.01 sell date: 2013-06-03, sell price: 9.84 buy date : 2013-07-11, buy price : 8.86 sell date: 2013-07-19, sell price: 8.03 buy date : 2013-08-12, buy price : 8.31 sell date: 2013-09-26, sell price: 10.17 buy date : 2013-10-29, buy price : 10.48 sell date: 2013-11-06, sell price: 10.03 buy date : 2013-11-12, buy price : 10.12 sell date: 2013-11-13, sell price: 9.64 buy date : 2013-11-18, buy price : 10.10 sell date: 2013-11-19, sell price: 9.97 buy date : 2013-11-29, buy price : 9.99 sell date: 2013-12-11, sell price: 9.93 buy date : 2014-01-10, buy price : 9.41 sell date: 2014-01-15, sell price: 9.23 buy date : 2014-01-22, buy price : 9.34 sell date: 2014-01-23, sell price: 9.22 buy date : 2014-01-29, buy price : 9.29 sell date: 2014-01-30, sell price: 9.17 buy date : 2014-02-10, buy price : 9.29 sell date: 2014-02-18, sell price: 9.26 buy date : 2014-02-19, buy price : 9.49 sell date: 2014-02-21, sell price: 9.29 buy date : 2014-03-14, buy price : 8.99 sell date: 2014-03-18, sell price: 8.89 buy date : 2014-03-19, buy price : 9.05 sell date: 2014-04-17, sell price: 9.76 buy date : 2014-04-22, buy price : 9.90 sell date: 2014-04-24, sell price: 9.87 buy date : 2014-05-12, buy price : 9.93 sell date: 2014-05-15, sell price: 9.73 buy date : 2014-05-26, buy price : 9.74 sell date: 2014-05-27, sell price: 9.66 buy date : 2014-06-13, buy price : 9.73 sell date: 2014-06-24, sell price: 9.04 buy date : 2014-07-22, buy price : 9.06 sell date: 2014-08-20, sell price: 9.64 buy date : 2014-09-02, buy price : 9.60 sell date: 2014-10-23, sell price: 9.75 buy date : 2014-10-30, buy price : 9.81 sell date: 2015-01-19, sell price: 14.82 buy date : 2015-01-21, buy price : 15.74 sell date: 2015-01-26, sell price: 15.43 buy date : 2015-02-26, buy price : 14.62 sell date: 2015-03-03, sell price: 14.00 buy date : 2015-03-09, buy price : 14.51 sell date: 2015-03-10, sell price: 14.10 buy date : 2015-03-11, buy price : 14.25 sell date: 2015-05-04, sell price: 17.81 buy date : 2015-05-22, buy price : 17.76 sell date: 2015-05-28, sell price: 17.04 buy date : 2015-06-01, buy price : 17.93 sell date: 2015-06-19, sell price: 17.06 buy date : 2015-07-06, buy price : 17.22 sell date: 2015-07-08, sell price: 15.98 buy date : 2015-07-10, buy price : 17.31 sell date: 2015-07-14, sell price: 16.74 buy date : 2015-07-15, buy price : 17.07 sell date: 2015-07-20, sell price: 16.68 buy date : 2015-08-31, buy price : 14.97 sell date: 2015-09-07, sell price: 14.41 buy date : 2015-09-08, buy price : 15.10 sell date: 2015-12-11, sell price: 18.61 buy date : 2015-12-21, buy price : 19.16 sell date: 2015-12-22, sell price: 18.89 buy date : 2015-12-24, buy price : 19.03 sell date: 2015-12-28, sell price: 18.72 rate: 1.23
最终收益率为23%,这样就是一个最简单的量化工具了。
这里还有一些问题,上述的接口是没有复权的,这个如果除权或分红比较多的话就有比较大的偏差,要加上前复权的处理。
另外就是有时候非必是买卖都能完成的,比如无量涨停和无量跌停以及之前的熔断机制,这种情况可以往后推直到涨跌停破了为止。