recommendation based on subject
最后更新于：2022-04-01 21:57:07
# recommendation based on subject
> 来源：https://uqer.io/community/share/549d0203f9f06c4bb8863242
## 策略思路：
+ step1：计算昨日所有主题的涨跌幅，根据涨跌幅排名，挑出涨幅最高的前`n_sub`个主题
+ step2：根据昨日成交量挑选出每个主题的龙头股`n_bigstk`只
+ 买入策略：昨日涨幅最高的前`n_sub`个主题，每个主题龙头股`n_bigstk`只，当日一共买入`m*n`只个股
+ 卖出策略：持有固定天数`hold_days`，卖出。
此实验中，`n_sub=5`, `n_bigstk=5`, `hold_days=10`
文件`sub_stk_info.txt`是根据`dataapi`获得的文件，里面储存了多个主题及对应的股票列表，点击[这里](https://app.yinxiang.com/shard/s52/sh/cfee1210-3b74-4fdf-b7d1-9806815dc1cb/9f232abc1f0ac12c583528ba73d0a7b6)下载
```py
a2=read('sub_stk_info.txt')
b2=a2.split('\r\n')
b2=b2[:-1]
sub_stk_dic={}
universe1=set([])
IDmap=lambda x:x +'.XSHG' if x[0]=='6' else x+'.XSHE'
for i2 in b2:
i2=i2.split(':')
i3=i2[1].split(',')
sub_stk_dic[i2[0]]=map(IDmap,i3)
universe1 |= set(i3)
start = datetime(2013, 6, 23) # 回测起始时间
end = datetime(2014, 12, 23) # 回测结束时间
benchmark = 'HS300' # 使用沪深 300 作为参考标准
universe = map(IDmap, list(universe1))
capital_base = 100000 # 起始资金
#print len(universe)
hold_days=10
sell_stk_list=[]
for i in range(hold_days):
sell_stk_list.append({})
j=hold_days
def initialize(account): # 初始化虚拟账户状态
add_history('hist',1)
def handle_data(account): # 每个交易日的买入卖出指令
global sell_stk_list
global j
#计算昨日主题涨跌幅
sub_increase_rate_dic={}
sub_bigstk_dic={}
#print 'today:',account.current_date
for (subid,stkid_list) in sub_stk_dic.items():
increase_rate_list=[]
turnvol_list=[]
#记录每只股票的成交量
stk_turnvol_dic={}
for stk in stkid_list:
#停盘的情况
if (stk not in account.universe) :
continue
close_price=account.hist[stk].iloc[0,3]
pre_close_price=account.hist[stk].iloc[0,4]
turnoverVol=account.hist[stk].iloc[0,5]
stk_turnvol_dic[stk]=turnoverVol
increase_rate_yes=(close_price-pre_close_price)*turnoverVol/pre_close_price
increase_rate_list.append(increase_rate_yes)
turnvol_list.append(turnoverVol)
big_stk_list=sorted(stk_turnvol_dic.keys(),key=lambda x:stk_turnvol_dic[x], reverse=True)
#买龙头股，每个主题买n只龙头股
n_bigstk=5
big_stk_list=big_stk_list[0:n_bigstk]
sub_bigstk_dic[subid]=big_stk_list
increase_rate_w=sum(increase_rate_list)/sum(turnvol_list)
sub_increase_rate_dic[subid]=increase_rate_w
sub_increase_rate_dic_sorted=sorted(sub_increase_rate_dic.keys(), key = lambda x:sub_increase_rate_dic[x], reverse = True)
n_sub=5
buy_subject_list=sub_increase_rate_dic_sorted[0:n_sub]
buy_stk_list=[]
for sub_id in buy_subject_list:
buy_stk_list +=sub_bigstk_dic[sub_id]
sell_next_dic={}
for stk in buy_stk_list:
if j>0:
amount=int(account.position.cash/hold_days/len(buy_stk_list)/account.hist[stk].iloc[0,3])
j -=1
else:
amount=int(account.position.cash/len(buy_stk_list)/account.hist[stk].iloc[0,3])
order(stk,amount)
sell_next_dic[stk]=amount
sell_stk_list.insert(0,sell_next_dic)
#print 'sell_stk_list:',sell_stk_list
sell_today_dic=sell_stk_list.pop()
#print 'sell_today_dic',sell_today_dic
if sell_today_dic!={}:
for (stk,amt) in sell_today_dic.items():
#如果股票今天不能交易,就过hold_days再卖
if stk not in account.universe:
sell_stk_list[0][stk]=amt
else:
order(stk,-amt)
```
';