1. # coding=utf-8
      2. from __future__ import print_function, absolute_import, unicode_literals
      4. import numpy as np
      5. from gm.api import *
      6. from pandas import DataFrame
      8. '''
      9. 本策略每隔1个月定时触发,根据Fama-French三因子模型对每只股票进行回归，得到其alpha值。
      10. 假设Fama-French三因子模型可以完全解释市场，则alpha为负表明市场低估该股，因此应该买入。
      11. 策略思路：
      12. 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,
      13. 根据分类得到的组合分别计算其市值加权收益率、SMB和HML. 
      14. 对各个股票进行回归(假设无风险收益率等于0)得到alpha值.
      15. 选取alpha值小于0并为最小的10只股票进入标的池
      16. 平掉不在标的池的股票并等权买入在标的池的股票
      17. 回测数据:SHSE.000300的成份股
      18. 回测时间:2017-07-01 08:00:00到2017-10-01 16:00:00
      19. '''
      22. def init(context):
      23.     # 每月第一个交易日的09:40 定时执行algo任务
      24.     schedule(schedule_func=algo, date_rule='1m', time_rule='09:40:00')
      25.     print(order_target_percent(symbol='SHSE.600000', percent=0.5, order_type=OrderType_Market,
      26.                          position_side=PositionSide_Long))
      27.     # 数据滑窗
      28.     context.date = 20
      29.     # 设置开仓的最大资金量
      30.     context.ratio = 0.8
      31.     # 账面市值比的大/中/小分类
      32.     context.BM_BIG = 3.0
      33.     context.BM_MID = 2.0
      34.     context.BM_SMA = 1.0
      35.     # 市值大/小分类
      36.     context.MV_BIG = 2.0
      37.     context.MV_SMA = 1.0
      40. # 计算市值加权的收益率,MV为市值的分类,BM为账目市值比的分类
      41. def market_value_weighted(stocks, MV, BM):
      42.     select = stocks[(stocks.NEGOTIABLEMV == MV) & (stocks.BM == BM)]
      43.     market_value = select['mv'].values
      44.     mv_total = np.sum(market_value)
      45.     mv_weighted = [mv / mv_total for mv in market_value]
      46.     stock_return = select['return'].values
      47.     # 返回市值加权的收益率的和
      48.     return_total = []
      49.     for i in range(len(mv_weighted)):
      50.         return_total.append(mv_weighted[i] * stock_return[i])
      51.     return_total = np.sum(return_total)
      52.     return return_total
      55. def algo(context):
      56.     # 获取上一个交易日的日期
      57.     last_day = get_previous_trading_date(exchange='SHSE', date=context.now)
      58.     # 获取沪深300成份股
      59.     context.stock300 = get_history_constituents(index='SHSE.000300', start_date=last_day,
      60.                                                 end_date=last_day)[0]['constituents'].keys()
      61.     # 获取当天有交易的股票
      62.     not_suspended = get_history_instruments(symbols=context.stock300, start_date=last_day, end_date=last_day)
      63.     not_suspended = [item['symbol'] for item in not_suspended if not item['is_suspended']]
      64.     fin = get_fundamentals(table='tq_sk_finindic', symbols=not_suspended, start_date=last_day, end_date=last_day,
      65.                            fields='PB,NEGOTIABLEMV', df=True)
      67.     # 计算账面市值比,为P/B的倒数
      68.     fin['PB'] = (fin['PB'] ** -1)
      69.     # 计算市值的50%的分位点,用于后面的分类
      70.     size_gate = fin['NEGOTIABLEMV'].quantile(0.50)
      71.     # 计算账面市值比的30%和70%分位点,用于后面的分类
      72.     bm_gate = [fin['PB'].quantile(0.30), fin['PB'].quantile(0.70)]
      73.     fin.index = fin.symbol
      74.     x_return = []
      75.     # 对未停牌的股票进行处理
      76.     for symbol in not_suspended:
      77.         # 计算收益率
      78.         close = history_n(symbol=symbol, frequency='1d', count=context.date + 1, end_time=last_day, fields='close',
      79.                           skip_suspended=True, fill_missing='Last', adjust=ADJUST_PREV, df=True)['close'].values
      80.         stock_return = close[-1] / close[0] - 1
      81.         pb = fin['PB'][symbol]
      82.         market_value = fin['NEGOTIABLEMV'][symbol]
      83.         # 获取[股票代码. 股票收益率, 账面市值比的分类, 市值的分类, 流通市值]
      84.         if pb < bm_gate[0]:
      85.             if market_value < size_gate:
      86.                 label = [symbol, stock_return, context.BM_SMA, context.MV_SMA, market_value]
      87.             else:
      88.                 label = [symbol, stock_return, context.BM_SMA, context.MV_BIG, market_value]
      89.         elif pb < bm_gate[1]:
      90.             if market_value < size_gate:
      91.                 label = [symbol, stock_return, context.BM_MID, context.MV_SMA, market_value]
      92.             else:
      93.                 label = [symbol, stock_return, context.BM_MID, context.MV_BIG, market_value]
      94.         elif market_value < size_gate:
      95.             label = [symbol, stock_return, context.BM_BIG, context.MV_SMA, market_value]
      96.         else:
      97.             label = [symbol, stock_return, context.BM_BIG, context.MV_BIG, market_value]
      98.         if len(x_return) == 0:
      99.             x_return = label
      100.         else:
      101.             x_return = np.vstack([x_return, label])
      103.     stocks = DataFrame(data=x_return, columns=['symbol', 'return', 'BM', 'NEGOTIABLEMV', 'mv'])
      104.     stocks.index = stocks.symbol
      105.     columns = ['return', 'BM', 'NEGOTIABLEMV', 'mv']
      106.     for column in columns:
      107.         stocks[column] = stocks[column].astype(np.float64)
      108.     # 计算SMB.HML和市场收益率
      109.     # 获取小市值组合的市值加权组合收益率
      110.     smb_s = (market_value_weighted(stocks, context.MV_SMA, context.BM_SMA) +
      111.              market_value_weighted(stocks, context.MV_SMA, context.BM_MID) +
      112.              market_value_weighted(stocks, context.MV_SMA, context.BM_BIG)) / 3
      114.     # 获取大市值组合的市值加权组合收益率
      115.     smb_b = (market_value_weighted(stocks, context.MV_BIG, context.BM_SMA) +
      116.              market_value_weighted(stocks, context.MV_BIG, context.BM_MID) +
      117.              market_value_weighted(stocks, context.MV_BIG, context.BM_BIG)) / 3
      119.     smb = smb_s - smb_b
      120.     # 获取大账面市值比组合的市值加权组合收益率
      121.     hml_b = (market_value_weighted(stocks, context.MV_SMA, 3) +
      122.              market_value_weighted(stocks, context.MV_BIG, context.BM_BIG)) / 2
      123.     # 获取小账面市值比组合的市值加权组合收益率
      124.     hml_s = (market_value_weighted(stocks, context.MV_SMA, context.BM_SMA) +
      125.              market_value_weighted(stocks, context.MV_BIG, context.BM_SMA)) / 2
      127.     hml = hml_b - hml_s
      128.     close = history_n(symbol='SHSE.000300', frequency='1d', count=context.date + 1,
      129.                       end_time=last_day, fields='close', skip_suspended=True,
      130.                       fill_missing='Last', adjust=ADJUST_PREV, df=True)['close'].values
      131.     market_return = close[-1] / close[0] - 1
      132.     coff_pool = []
      133.     # 对每只股票进行回归获取其alpha值
      134.     for stock in stocks.index:
      135.         x_value = np.array([[market_return], [smb], [hml], [1.0]])
      136.         y_value = np.array([stocks['return'][stock]])
      137.         # OLS估计系数
      138.         coff = np.linalg.lstsq(x_value.T, y_value)[0][3]
      139.         coff_pool.append(coff)
      141.     # 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)
      142.     stocks['alpha'] = coff_pool
      143.     stocks = stocks[stocks.alpha < 0].sort_values(by='alpha').head(10)
      145.     symbols_pool = stocks.index.tolist()
      146.     positions = context.account().positions()
      148.     # 平不在标的池的股票
      149.     for position in positions:
      150.         symbol = position['symbol']
      151.         if symbol not in symbols_pool:
      152.             order_target_percent(symbol=symbol, percent=0, order_type=OrderType_Market,
      153.                                  position_side=PositionSide_Long)
      154.             print('市价单平不在标的池的', symbol)
      156.     # 获取股票的权重
      157.     percent = context.ratio / len(symbols_pool)
      158.     # 买在标的池中的股票
      159.     for symbol in symbols_pool:
      160.         order_target_percent(symbol=symbol, percent=percent, order_type=OrderType_Market,
      161.                              position_side=PositionSide_Long)
      162.         print(symbol, '以市价单调多仓到仓位', percent)
      165. if __name__ == '__main__':
      166.     '''
      167.     strategy_id策略ID,由系统生成
      168.     filename文件名,请与本文件名保持一致
      169.     mode实时模式:MODE_LIVE回测模式:MODE_BACKTEST
      170.     token绑定计算机的ID,可在系统设置-密钥管理中生成
      171.     backtest_start_time回测开始时间
      172.     backtest_end_time回测结束时间
      173.     backtest_adjust股票复权方式不复权:ADJUST_NONE前复权:ADJUST_PREV后复权:ADJUST_POST
      174.     backtest_initial_cash回测初始资金
      175.     backtest_commission_ratio回测佣金比例
      176.     backtest_slippage_ratio回测滑点比例
      177.     '''
      178.     run(strategy_id='strategy_id',
      179.         filename='main.py',
      180.         mode=MODE_BACKTEST,
      181.         token='token_id',
      182.         backtest_start_time='2017-07-01 08:00:00',
      183.         backtest_end_time='2017-10-01 16:00:00',
      184.         backtest_adjust=ADJUST_PREV,
      185.         backtest_initial_cash=10000000,
      186.         backtest_commission_ratio=0.0001,
      187.         backtest_slippage_ratio=0.0001)

    [多因子选股(股票)]

    原文: https://www.myquant.cn/docs/python_strategyies/103