# coding=utf-8
from __future__ import print_function, absolute_import, unicode_literals
from gm.api import *
import datetime
import numpy as np
import pandas as pd
'''
示例策略仅供参考,不建议直接实盘使用。
本策略基于Fama-French三因子模型。
假设三因子模型可以完全解释市场,以三因子模型对每股股票进行回归计算其Alpha值,当alpha为负表明市场低估该股,因此应该买入。
策略思路:
计算市场收益率、个股的账面市值比和市值,并对后两个进行了分类,
根据分类得到的组合分别计算其市值加权收益率、SMB和HML.
对各个股票进行回归(假设无风险收益率等于0)得到Alpha值.
选取Alpha值小于0并为最小的10只股票进入标的池,每月初移仓换股
'''
def init(context):
# 成分股指数
context.index_symbol = 'SHSE.000300'
# 数据滑窗
context.date = 20
# 设置开仓的最大资金量
context.ratio = 0.8
# 账面市值比的大/中/小分类
context.BM_HIGH = 3.0
context.BM_MIDDLE = 2.0
context.BM_LOW = 1.0
# 市值大/小分类
context.MV_BIG = 2.0
context.MV_SMALL = 1.0
# 每个交易日的09:40 定时执行algo任务
schedule(schedule_func=algo, date_rule='1d', time_rule='09:30:00')
def algo(context):
# 当前时间
now = context.now
now_str = now.strftime('%Y-%m-%d')
# 获取上一个交易日的日期
last_day = get_previous_n_trading_dates(exchange='SHSE', date=now_str, n=1)[0]
# 判断是否为每个月第一个交易日
if now.month!=pd.Timestamp(last_day).month:
# 获取沪深300成份股
stock300 = stk_get_index_constituents(index=context.index_symbol, trade_date=last_day)['symbol'].tolist()
# 过滤停牌、ST、退市及未上市的股票
stocks_info = get_symbols(sec_type1=1010, symbols=stock300, trade_date=now.strftime('%Y-%m-%d'), skip_suspended=True, skip_st=True)
stock300 = [item['symbol'] for item in stocks_info if item['listed_date']<now and item['delisted_date']>now]
# 获取所有股票市值
fin = stk_get_daily_mktvalue_pt(symbols=stock300, fields='tot_mv', trade_date=last_day, df=True).sort_values(by='tot_mv')
# 净资产
ttl_eqy = stk_get_fundamentals_balance_pt(symbols=stock300, date=last_day, fields='ttl_eqy', df=True)
ttl_eqy['max_rpt_date'] = ttl_eqy.groupby(['symbol'])['rpt_date'].max == ttl_eqy['rpt_date']
ttl_eqy = ttl_eqy[ttl_eqy['max_rpt_date'] == True]
# 计算PB
fin = fin.merge(ttl_eqy,on=['symbol'],how='left')
fin['PB'] = fin['tot_mv']/fin['ttl_eqy']
# 计算账面市值比,为P/B的倒数
fin.loc[:,'PB'] = (fin['PB'] ** -1)
# 计算市值的50%的分位点,用于后面的分类
size_gate = fin['tot_mv'].quantile(0.50)
# 计算账面市值比的30%和70%分位点,用于后面的分类
bm_gate = [fin['PB'].quantile(0.30), fin['PB'].quantile(0.70)]
fin.index = fin.symbol
# 设置存放股票收益率的变量
data_df = pd.DataFrame()
# 对未停牌的股票进行处理
for symbol in fin.symbol:
# 计算收益率
close = history_n(symbol=symbol, frequency='1d', count=context.date + 1, end_time=last_day, fields='close',
skip_suspended=True, fill_missing='Last', adjust=ADJUST_PREV, df=True)['close'].values
stock_return = close[-1] / close[0] - 1
pb = fin['PB'][symbol]
market_value = fin['tot_mv'][symbol]
# 获取[股票代码, 股票收益率, 账面市值比的分类, 市值的分类, 市值]
# 其中账面市值比的分类为:高(3)、中(2)、低(1)
# 市值的分类:大(2)、小(1)
if pb < bm_gate[0]:
if market_value < size_gate:
label = [symbol, stock_return, context.BM_LOW, context.MV_SMALL, market_value]# 小市值/低BM
else:
label = [symbol, stock_return, context.BM_LOW, context.MV_BIG, market_value]# 大市值/低BM
elif pb < bm_gate[1]:
if market_value < size_gate:
label = [symbol, stock_return, context.BM_MIDDLE, context.MV_SMALL, market_value]# 小市值/中BM
else:
label = [symbol, stock_return, context.BM_MIDDLE, context.MV_BIG, market_value]# 大市值/中BM
elif market_value < size_gate:
label = [symbol, stock_return, context.BM_HIGH, context.MV_SMALL, market_value]# 小市值/高BM
else:
label = [symbol, stock_return, context.BM_HIGH, context.MV_BIG, market_value]# 大市值/高BM
data_df = pd.concat([data_df,pd.DataFrame(label,index=['symbol', 'return', 'BM', 'tot_mv', 'mv']).T])
data_df.set_index('symbol',inplace=True)
# 调整数据类型
for column in data_df.columns:
data_df[column] = data_df[column].astype(np.float64)
# 计算小市值组合的收益率(组内以市值加权计算收益率,组间以等权计算收益率)
smb_s = (market_value_weighted(data_df, context.MV_SMALL, context.BM_LOW) +
market_value_weighted(data_df, context.MV_SMALL, context.BM_MIDDLE) +
market_value_weighted(data_df, context.MV_SMALL, context.BM_HIGH)) / 3
# 计算大市值组合的收益率(组内以市值加权计算收益率,组间以等权计算收益率)
smb_b = (market_value_weighted(data_df, context.MV_BIG, context.BM_LOW) +
market_value_weighted(data_df, context.MV_BIG, context.BM_MIDDLE) +
market_value_weighted(data_df, context.MV_BIG, context.BM_HIGH)) / 3
# 计算规模因子的收益率(小市值组收益率-大市值组收益率)
smb = smb_s - smb_b
# 计算高BM组合的收益率(组内以市值加权计算收益率,组间以等权计算收益率)
hml_b = (market_value_weighted(data_df, context.MV_SMALL, context.BM_HIGH) +
market_value_weighted(data_df, context.MV_BIG, context.BM_HIGH)) / 2
# 计算低BM组合的收益率(组内以市值加权计算收益率,组间以等权计算收益率)
hml_s = (market_value_weighted(data_df, context.MV_SMALL, context.BM_LOW) +
market_value_weighted(data_df, context.MV_BIG, context.BM_LOW)) / 2
# 计算价值因子的收益率(高BM组收益率-低BM市值组收益率)
hml = hml_b - hml_s
# 获取市场收益率
close = history_n(symbol=context.index_symbol, frequency='1d', count=context.date + 1,
end_time=last_day, fields='close', skip_suspended=True,
fill_missing='Last', adjust=ADJUST_PREV, df=True)['close'].values
market_return = close[-1] / close[0] - 1
coff_pool = []
# 对每只股票进行回归获取其alpha值
for stock in data_df.index:
x_value = np.array([[market_return], [smb], [hml], [1.0]])
y_value = np.array([data_df['return'][stock]])
# OLS估计系数
coff = np.linalg.lstsq(x_value.T, y_value, rcond=None)[0][3]
coff_pool.append(coff)
# 获取alpha最小并且小于0的10只的股票进行操作(若少于10只则全部买入)
data_df.loc[:,'alpha'] = coff_pool
symbols_pool = data_df[data_df.alpha < 0].sort_values(by='alpha').head(10).index.tolist()
positions = get_position()
# 平不在标的池的股票(注:本策略交易以开盘价为交易价格,当调整定时任务时间时,需调整对应价格)
for position in positions:
symbol = position['symbol']
if symbol not in symbols_pool:
# 开盘价(日频数据)
new_price = history_n(symbol=symbol, frequency='1d', count=1, end_time=context.now, fields='open', adjust=ADJUST_PREV, adjust_end_time=context.backtest_end_time, df=False)[0]['open']
# # 当前价(tick数据,免费版本有时间权限限制;实时模式,返回当前最新 tick 数据,回测模式,返回回测当前时间点的最近一分钟的收盘价)
# new_price = current(symbols=symbol)[0]['price']
order_info = order_target_percent(symbol=symbol, percent=0, order_type=OrderType_Limit,position_side=PositionSide_Long,price=new_price)
# 获取股票的权重
percent = context.ratio / len(symbols_pool)
# 买在标的池中的股票(注:本策略交易以开盘价为交易价格,当调整定时任务时间时,需调整对应价格)
for symbol in symbols_pool:
# 开盘价(日频数据)
new_price = history_n(symbol=symbol, frequency='1d', count=1, end_time=context.now, fields='open', adjust=ADJUST_PREV, adjust_end_time=context.backtest_end_time, df=False)[0]['open']
# # 当前价(tick数据,免费版本有时间权限限制;实时模式,返回当前最新 tick 数据,回测模式,返回回测当前时间点的最近一分钟的收盘价)
# new_price = current(symbols=symbol)[0]['price']
order_info = order_target_percent(symbol=symbol, percent=percent, order_type=OrderType_Limit,position_side=PositionSide_Long,price=new_price)
def market_value_weighted(df, MV, BM):
"""
计算市值加权下的收益率
:param MV:MV为市值的分类对应的组别
:param BM:BM账目市值比的分类对应的组别
"""
select = df[(df['tot_mv'] == MV) & (df['BM'] == BM)] # 选出市值为MV,账目市值比为BM的所有股票数据
mv_weighted = select['mv']/np.sum(select['mv'])# 市值加权的权重
return_weighted = select['return']*mv_weighted# 市值加权下的收益率
return np.sum(return_weighted)
def on_order_status(context, order):
# 标的代码
symbol = order['symbol']
# 委托价格
price = order['price']
# 委托数量
volume = order['volume']
# 目标仓位
target_percent = order['target_percent']
# 查看下单后的委托状态,等于3代表委托全部成交
status = order['status']
# 买卖方向,1为买入,2为卖出
side = order['side']
# 开平仓类型,1为开仓,2为平仓
effect = order['position_effect']
# 委托类型,1为限价委托,2为市价委托
order_type = order['order_type']
if status == 3:
if effect == 1:
if side == 1:
side_effect = '开多仓'
else:
side_effect = '开空仓'
else:
if side == 1:
side_effect = '平空仓'
else:
side_effect = '平多仓'
order_type_word = '限价' if order_type==1 else '市价'
print('{}:标的:{},操作:以{}{},委托价格:{},目标仓位:{:.2%}'.format(context.now,symbol,order_type_word,side_effect,price,target_percent))
def on_backtest_finished(context, indicator):
print('*'*50)
print('回测已完成,请通过右上角“回测历史”功能查询详情。')
if __name__ == '__main__':
'''
strategy_id策略ID,由系统生成
filename文件名,请与本文件名保持一致
mode实时模式:MODE_LIVE回测模式:MODE_BACKTEST
token绑定计算机的ID,可在系统设置-密钥管理中生成
backtest_start_time回测开始时间
backtest_end_time回测结束时间
backtest_adjust股票复权方式不复权:ADJUST_NONE前复权:ADJUST_PREV后复权:ADJUST_POST
backtest_initial_cash回测初始资金
backtest_commission_ratio回测佣金比例
backtest_slippage_ratio回测滑点比例
backtest_match_mode市价撮合模式,以下一tick/bar开盘价撮合:0,以当前tick/bar收盘价撮合:1
'''
run(strategy_id='strategy_id',
filename='main.py',
mode=MODE_BACKTEST,
token='{{token}}',
backtest_start_time='2021-08-01 08:00:00',
backtest_end_time='2022-02-10 16:00:00',
backtest_adjust=ADJUST_PREV,
backtest_initial_cash=1000000,
backtest_commission_ratio=0.0001,
backtest_slippage_ratio=0.0001,
backtest_match_mode=1)