StockRanker多因子选股策略¶

In [81]:

# 本代码由可视化策略环境自动生成 2022年5月30日 15:48
# 本代码单元只能在可视化模式下编辑。您也可以拷贝代码，粘贴到新建的代码单元或者策略，然后修改。


# Python 代码入口函数，input_1/2/3 对应三个输入端，data_1/2/3 对应三个输出端
def m2_run_bigquant_run(input_1, input_2, input_3):
    # 示例代码如下。在这里编写您的代码
    df = DataSource("bar1d_CN_CONBOND").read(start_date="2017-06-01", end_date="2020-06-30")
    df2 = df.drop(['close'],axis = 1)
    data_1 = DataSource.write_df(df)
    data_2 = DataSource.write_df(df2)
    return Outputs(data_1=data_1, data_2=data_2, data_3=None)

# 后处理函数，可选。输入是主函数的输出，可以在这里对数据做处理，或者返回更友好的outputs数据格式。此函数输出不会被缓存。
def m2_post_run_bigquant_run(outputs):
    return outputs

def del_input(input):
    df = input.read_df()
    df = df.drop(["trigger_cond_item_desc",'revise_item_desc','trigger_item_desc'],axis=1)
    df = df.dropna(axis=0, how='all', thresh=None, subset=None, inplace=False)
    print(df.columns)
    #df.sort_values(by=['date','double_low'],axis=0,ascending=True,inplace = True)
    df = df.sort_index(axis = 1)
    df = df.reset_index(drop = True)
    df = df.groupby('date').head(10)
    #print(df[['double_low','date']])
    data_1 = DataSource.write_df(df)
    return data_1
# Python 代码入口函数，input_1/2/3 对应三个输入端，data_1/2/3 对应三个输出端
def m23_run_bigquant_run(input_1, input_2, input_3):
    data_1 = del_input(input_1)
    data_2 = del_input(input_2)
    return Outputs(data_1=data_1, data_2=data_2, data_3=None)
# 后处理函数，可选。输入是主函数的输出，可以在这里对数据做处理，或者返回更友好的outputs数据格式。此函数输出不会被缓存。
def m23_post_run_bigquant_run(outputs):
    return outputs

# Python 代码入口函数，input_1/2/3 对应三个输入端，data_1/2/3 对应三个输出端
def m15_run_bigquant_run(input_1, input_2, input_3):
    # 示例代码如下。在这里编写您的代码
    df = DataSource("bar1d_CN_CONBOND").read(start_date="2020-06-30", end_date="2021-06-30")
    df = df.drop(['close'],axis = 1)
    data_1 = DataSource.write_df(df)
    data_2 = DataSource.write_pickle(df)
    return Outputs(data_1=data_1, data_2=data_2, data_3=None)

# 后处理函数，可选。输入是主函数的输出，可以在这里对数据做处理，或者返回更友好的outputs数据格式。此函数输出不会被缓存。
def m15_post_run_bigquant_run(outputs):
    return outputs

# Python 代码入口函数，input_1/2/3 对应三个输入端，data_1/2/3 对应三个输出端
def m13_run_bigquant_run(input_1, input_2, input_3):
    # 示例代码如下。在这里编写您的代码
    df = input_1.read_df()
    #print(df.columns)
    #df = df1[['date','instrument','close','label']]
    #df = df.drop(["trigger_cond_item_desc",'revise_item_desc','trigger_item_desc'],axis=1)
    df = df.dropna(axis=0, how='all', thresh=None, subset=None, inplace=False)
    print(df.columns)
    df.sort_values(by=['date','double_low'],axis=0,ascending=True,inplace = True)
    df = df.sort_index(axis = 1)
    df = df.reset_index(drop = True)
    df = df.groupby('date').head(10)
    #print(df[['double_low','date']])
    data_1 = DataSource.write_df(df)
    data_2 = DataSource.write_pickle(df)
    return Outputs(data_1=data_1, data_2=data_2, data_3=None)
# 后处理函数，可选。输入是主函数的输出，可以在这里对数据做处理，或者返回更友好的outputs数据格式。此函数输出不会被缓存。
def m13_post_run_bigquant_run(outputs):
    return outputs

# Python 代码入口函数，input_1/2/3 对应三个输入端，data_1/2/3 对应三个输出端
def m14_run_bigquant_run(input_1, input_2, input_3, stock_count, hold_days):
    # 示例代码如下。在这里编写您的代码
    param = {
        "stock_count": stock_count,
        "hold_days": hold_days
        }
    data_1 = DataSource.write_pickle(param)
    return Outputs(data_1=data_1, data_2=None, data_3=None)

# 后处理函数，可选。输入是主函数的输出，可以在这里对数据做处理，或者返回更友好的outputs数据格式。此函数输出不会被缓存。
def m14_post_run_bigquant_run(outputs):
    return outputs

# Python 代码入口函数，input_1/2/3 对应三个输入端，data_1/2/3 对应三个输出端
def m11_run_bigquant_run(input_1, input_2, input_3):
    # 示例代码如下。在这里编写您的代码
    df = input_1.read()
    param = input_2.read()
    
    data = {
        "param": param,
        "data": df
    }
    data_1 = DataSource.write_pickle(data)
    return Outputs(data_1=data_1, data_2=None, data_3=None)

# 后处理函数，可选。输入是主函数的输出，可以在这里对数据做处理，或者返回更友好的outputs数据格式。此函数输出不会被缓存。
def m11_post_run_bigquant_run(outputs):
    return outputs

# 回测引擎：初始化函数，只执行一次
def m21_initialize_bigquant_run(context):
    print(context.options)
    # 加载预测数据
    context.ranker_prediction = context.options.get('data').read()['data']
    print(context.ranker_prediction)
    context.param = context.options['data'].read()["param"]
    # 系统已经设置了默认的交易手续费和滑点，要修改手续费可使用如下函数
    context.set_commission(PerOrder(buy_cost=0.0003, sell_cost=0.0013, min_cost=5))
    # 预测数据，通过options传入进来，使用 read_df 函数，加载到内存 (DataFrame)
    # 设置买入的股票数量，这里买入预测股票列表排名靠前的5只
    stock_count = context.param["stock_count"]
    # 每只的股票的权重，如下的权重分配会使得靠前的股票分配多一点的资金，[0.339160, 0.213986, 0.169580, ..]
    context.stock_weights = T.norm([1 / math.log(i + 2) for i in range(0, stock_count)])
    # 设置每只股票占用的最大资金比例
    context.max_cash_per_instrument = 0.4
    context.hold_days = context.param["hold_days"]

# 交易引擎：每个单位时间开盘前调用一次。
def m21_before_trading_start_bigquant_run(context, data):
    # 盘前处理，订阅行情等
    pass

# 交易引擎：tick数据处理函数，每个tick执行一次
def m21_handle_tick_bigquant_run(context, tick):
    pass

# 回测引擎：每日数据处理函数，每天执行一次
def m21_handle_data_bigquant_run(context, data):
    print('处理data',data)
    context.ranker_prediction = context.options.get('data').read()['data']
    # 相隔几天（hold_days）进行一下换仓
    if context.trading_day_index % context.hold_days != 0:
        return 
    
    # 按日期过滤得到今日的预测数据
    ranker_prediction = context.ranker_prediction[
        context.ranker_prediction.date == data.current_dt.strftime('%Y-%m-%d')]
    # 目前持仓
    positions = {e: p.amount * p.last_sale_price for e, p in context.portfolio.positions.items()}
    # 权重
    buy_cash_weights = context.stock_weights
    # 今日买入股票列表
    stock_to_buy = list(ranker_prediction.instrument[:len(buy_cash_weights)])
    # 持仓上限
    max_cash_per_instrument = context.portfolio.portfolio_value * context.max_cash_per_instrument

    # 通过positions对象，使用列表生成式的方法获取目前持仓的股票列表
    stock_hold_now = [equity for equity in context.portfolio.positions ]
    # 继续持有的股票：调仓时，如果买入的股票已经存在于目前的持仓里，那么应继续持有
    no_need_to_sell = [i for i in stock_hold_now if i in stock_to_buy]
    # 需要卖出的股票
    stock_to_sell = [i for i in stock_hold_now if i not in no_need_to_sell]
  
    # 卖出
    for stock in stock_to_sell:
        # 如果该股票停牌，则没法成交。因此需要用can_trade方法检查下该股票的状态
        # 如果返回真值，则可以正常下单，否则会出错
        # 因为stock是字符串格式，我们用symbol方法将其转化成平台可以接受的形式：Equity格式
        if data.can_trade(context.symbol(stock)):
            # order_target_percent是平台的一个下单接口，表明下单使得该股票的权重为0，
            #   即卖出全部股票，可参考回测文档
            context.order_target_percent(context.symbol(stock), 0)
    
    # 如果当天没有买入的股票，就返回
    if len(stock_to_buy) == 0:
        return
    
    # 买入
    for i, instrument in enumerate(stock_to_buy):
        cash = context.portfolio.portfolio_value * buy_cash_weights[i]
        if cash > max_cash_per_instrument - positions.get(instrument, 0):
            # 确保股票持仓量不会超过每次股票最大的占用资金量
            cash = max_cash_per_instrument - positions.get(instrument, 0)
        if cash > 0:
            context.order_value(context.symbol(instrument), cash)
# 交易引擎：成交回报处理函数，每个成交发生时执行一次
def m21_handle_trade_bigquant_run(context, trade):
    pass

# 交易引擎：委托回报处理函数，每个委托变化时执行一次
def m21_handle_order_bigquant_run(context, order):
    pass

# 交易引擎：盘后处理函数，每日盘后执行一次
def m21_after_trading_bigquant_run(context, data):
    pass


g = T.Graph({

    'm3': 'M.input_features.v1',
    'm3.features': """# #号开始的表示注释
# 多个特征，每行一个，可以包含基础特征和衍生特征
double_low = close + bond_prem_ratio
remain_size
rank_swing_volatility_5 = nanstd((high-low)/pre_close, 5)*sqrt(200)*100""",

    'm2': 'M.cached.v3',
    'm2.run': m2_run_bigquant_run,
    'm2.post_run': m2_post_run_bigquant_run,
    'm2.input_ports': '',
    'm2.params': '{}',
    'm2.output_ports': '',

    'm10': 'M.auto_labeler_on_datasource.v1',
    'm10.input_data': T.Graph.OutputPort('m2.data_1'),
    'm10.label_expr': """# #号开始的表示注释
# 0. 每行一个，顺序执行，从第二个开始，可以使用label字段
# 1. 可用数据字段见 https://bigquant.com/docs/develop/datasource/deprecated/history_data.html
# 2. 可用操作符和函数见 `表达式引擎 <https://bigquant.com/docs/develop/bigexpr/usage.html>`_

# 计算收益：5日收盘价(作为卖出价格)除以明日开盘价(作为买入价格)
shift(close, -5) / shift(open, -1)

# 极值处理：用1%和99%分位的值做clip
clip(label, all_quantile(label, 0.01), all_quantile(label, 0.99))

# 将分数映射到分类，这里使用20个分类
all_wbins(label, 10)

# 过滤掉一字涨停的情况 (设置label为NaN，在后续处理和训练中会忽略NaN的label)
where(shift(high, -1) == shift(low, -1), NaN, label)
""",
    'm10.drop_na_label': True,
    'm10.cast_label_int': True,
    'm10.date_col': 'date',
    'm10.instrument_col': 'instrument',
    'm10.user_functions': {},

    'm1': 'M.use_datasource.v1',
    'm1.datasource_id': 'market_performance_CN_CONBOND',
    'm1.start_date': '2017-06-01',
    'm1.end_date': '2020-06-30',

    'm5': 'M.use_datasource.v1',
    'm5.datasource_id': 'market_performance_CN_CONBOND',
    'm5.start_date': '2020-07-01',
    'm5.end_date': '2021-06-30',

    'm12': 'M.trade_data_generation.v1',
    'm12.input': T.Graph.OutputPort('m5.data'),
    'm12.category': 'CN_STOCK',
    'm12.m_cached': False,

    'm23': 'M.cached.v3',
    'm23.input_1': T.Graph.OutputPort('m1.data'),
    'm23.input_2': T.Graph.OutputPort('m5.data'),
    'm23.run': m23_run_bigquant_run,
    'm23.post_run': m23_post_run_bigquant_run,
    'm23.input_ports': '',
    'm23.params': '{}',
    'm23.output_ports': '',
    'm23.m_cached': False,

    'm4': 'M.join.v3',
    'm4.data1': T.Graph.OutputPort('m2.data_2'),
    'm4.data2': T.Graph.OutputPort('m23.data_1'),
    'm4.on': 'date,instrument',
    'm4.how': 'inner',
    'm4.sort': False,

    'm16': 'M.derived_feature_extractor.v3',
    'm16.input_data': T.Graph.OutputPort('m4.data'),
    'm16.features': T.Graph.OutputPort('m3.data'),
    'm16.date_col': 'date',
    'm16.instrument_col': 'instrument',
    'm16.drop_na': False,
    'm16.remove_extra_columns': False,

    'm7': 'M.join.v3',
    'm7.data1': T.Graph.OutputPort('m10.data'),
    'm7.data2': T.Graph.OutputPort('m16.data'),
    'm7.on': 'date,instrument',
    'm7.how': 'inner',
    'm7.sort': False,

    'm6': 'M.stock_ranker_train.v6',
    'm6.training_ds': T.Graph.OutputPort('m7.data'),
    'm6.features': T.Graph.OutputPort('m3.data'),
    'm6.learning_algorithm': '排序',
    'm6.number_of_leaves': 30,
    'm6.minimum_docs_per_leaf': 1000,
    'm6.number_of_trees': 20,
    'm6.learning_rate': 0.1,
    'm6.max_bins': 1023,
    'm6.feature_fraction': 1,
    'm6.data_row_fraction': 1,
    'm6.plot_charts': True,
    'm6.ndcg_discount_base': 1,
    'm6.m_lazy_run': False,

    'm15': 'M.cached.v3',
    'm15.run': m15_run_bigquant_run,
    'm15.post_run': m15_post_run_bigquant_run,
    'm15.input_ports': '',
    'm15.params': '{}',
    'm15.output_ports': '',

    'm25': 'M.join.v3',
    'm25.data1': T.Graph.OutputPort('m23.data_2'),
    'm25.data2': T.Graph.OutputPort('m15.data_1'),
    'm25.on': 'date,instrument',
    'm25.how': 'inner',
    'm25.sort': False,

    'm18': 'M.derived_feature_extractor.v3',
    'm18.input_data': T.Graph.OutputPort('m25.data'),
    'm18.features': T.Graph.OutputPort('m3.data'),
    'm18.date_col': 'date',
    'm18.instrument_col': 'instrument',
    'm18.drop_na': False,
    'm18.remove_extra_columns': False,

    'm13': 'M.cached.v3',
    'm13.input_1': T.Graph.OutputPort('m18.data'),
    'm13.run': m13_run_bigquant_run,
    'm13.post_run': m13_post_run_bigquant_run,
    'm13.input_ports': '',
    'm13.params': '{}',
    'm13.output_ports': '',
    'm13.m_cached': False,

    'm8': 'M.stock_ranker_predict.v5',
    'm8.model': T.Graph.OutputPort('m6.model'),
    'm8.data': T.Graph.OutputPort('m13.data_1'),
    'm8.m_lazy_run': False,

    'm14': 'M.cached.v3',
    'm14.run': m14_run_bigquant_run,
    'm14.post_run': m14_post_run_bigquant_run,
    'm14.input_ports': '',
    'm14.params': """{
    "stock_count": 3,
    "hold_days": 4 
}""",
    'm14.output_ports': '',

    'm11': 'M.cached.v3',
    'm11.input_1': T.Graph.OutputPort('m8.predictions'),
    'm11.input_2': T.Graph.OutputPort('m14.data_1'),
    'm11.run': m11_run_bigquant_run,
    'm11.post_run': m11_post_run_bigquant_run,
    'm11.input_ports': '',
    'm11.params': '{}',
    'm11.output_ports': '',

    'm21': 'M.hftrade.v2',
    'm21.instruments': T.Graph.OutputPort('m12.instrument_list'),
    'm21.options_data': T.Graph.OutputPort('m11.data_1'),
    'm21.start_date': '',
    'm21.end_date': '',
    'm21.initialize': m21_initialize_bigquant_run,
    'm21.before_trading_start': m21_before_trading_start_bigquant_run,
    'm21.handle_tick': m21_handle_tick_bigquant_run,
    'm21.handle_data': m21_handle_data_bigquant_run,
    'm21.handle_trade': m21_handle_trade_bigquant_run,
    'm21.handle_order': m21_handle_order_bigquant_run,
    'm21.after_trading': m21_after_trading_bigquant_run,
    'm21.capital_base': 1000000,
    'm21.frequency': 'daily',
    'm21.price_type': '真实价格',
    'm21.product_type': '可转债',
    'm21.before_start_days': '8',
    'm21.order_price_field_buy': 'open',
    'm21.order_price_field_sell': 'close',
    'm21.benchmark': '000300.HIX',
    'm21.plot_charts': True,
    'm21.disable_cache': False,
    'm21.replay_bdb': False,
    'm21.show_debug_info': False,
    'm21.backtest_only': False,
})

# g.run({})


def m9_param_grid_builder_bigquant_run():
    import itertools
    param_grid = {}
    
    period_list = [5,20,60,100,120] 
    
    # 在这里设置需要调优的参数备选
    feature_list = [
    '''
    double_low = close + bond_prem_ratio
    remain_size
    rank_swing_volatility_5 = nanstd((high-low)/pre_close, {0})*sqrt(200)*100
    '''.format(period) for period in period_list
    ]
    param_grid["m14.params"] = [
        """{"stock_count": 3, "hold_days": 3}""",
        """{"stock_count": 4, "hold_days": 4}""",
        """{"stock_count": 3, "hold_days": 1}""",
        """{"stock_count": 4, "hold_days": 2}"""
    ]
    param_grid['m3.features'] = feature_list
    return param_grid
def m9_scoring_bigquant_run(result):
    # 评分：收益/最大回撤
    score = result.get('m7').read_raw_perf()['sharpe'].tail(1)[0]
    return {'score': score}


m9 = M.hyper_parameter_search.v1(
    param_grid_builder=m9_param_grid_builder_bigquant_run,
    scoring=m9_scoring_bigquant_run,
    search_algorithm='网格搜索',
    search_iterations=10,
    workers=1,
    worker_distributed_run=False,
    worker_silent=False,
    run_now=True,
    bq_graph=g
)

[2022-05-30 15:42:47.429987] ERROR: moduleinvoker: module name: hyper_parameter_search, module version: v1, trackeback: KeyError: 'm1'

Fitting 1 folds for each of 3 candidates, totalling 3 fits
[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.
[CV 1/1; 1/3] START m1.end_date=2021-07-01, m1.start_date=2021-06-01............
[CV 1/1; 1/3] END m1.end_date=2021-07-01, m1.start_date=2021-06-01; total time=   0.0s
[Parallel(n_jobs=1)]: Done   1 out of   1 | elapsed:    0.0s remaining:    0.0s
[CV 1/1; 2/3] START m1.end_date=2021-07-01, m1.start_date=2021-06-10............
[CV 1/1; 2/3] END m1.end_date=2021-07-01, m1.start_date=2021-06-10; total time=   0.0s
[Parallel(n_jobs=1)]: Done   2 out of   2 | elapsed:    0.0s remaining:    0.0s
[CV 1/1; 3/3] START m1.end_date=2021-07-01, m1.start_date=2021-06-15............
[CV 1/1; 3/3] END m1.end_date=2021-07-01, m1.start_date=2021-06-15; total time=   0.0s
[Parallel(n_jobs=1)]: Done   3 out of   3 | elapsed:    0.0s remaining:    0.0s
[Parallel(n_jobs=1)]: Done   3 out of   3 | elapsed:    0.0s finished

---------------------------------------------------------------------------
KeyError                                  Traceback (most recent call last)
<ipython-input-81-5ab9ef3d6240> in <module>
     24 
     25 
---> 26 m9 = M.hyper_parameter_search.v1(
     27     param_grid_builder=m9_param_grid_builder_bigquant_run,
     28     scoring=m9_scoring_bigquant_run,

KeyError: 'm1'