复制链接

克隆策略

In [8]:

# 本代码由可视化策略环境自动生成 2022年12月5日 22:22
# 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。


# 交易引擎：初始化函数，只执行一次
def m19_initialize_bigquant_run(context):
       # 加载股票指标数据,数据继承自m6模块
    context.indicator_data = context.options['data'].read_df()

    # 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))
    
    # 设置数量
    context.stock_num = 10
    
    # 调仓天数,22个交易日大概就是一个月。可以理解为一个月换仓一次
    context.rebalance_days = 22
    
    # 如果策略运行中，需要将数据进行保存，可以借用extension这个对象，类型为dict
    # 比如当前运行的k线的索引，比如个股持仓天数、买入均价
    context.extension.index = 0
    context.subscribe(context.instruments)
 
# 交易引擎：每个单位时间开盘前调用一次。
def m19_before_trading_start_bigquant_run(context, data):
    pass
# 交易引擎：tick数据处理函数，每个tick执行一次
def m19_handle_tick_bigquant_run(context, data):
    pass

# 回测引擎：每日数据处理函数，每天执行一次
def m19_handle_data_bigquant_run(context, data):
    
    context.extension.index += 1
    # 不在换仓日就return,相当于后面的代码只会一个月运行一次，买入的股票会持有1个月
    if  context.extension.index % context.rebalance_days != 0:
        return
    
    # 当前的日期
    date = data.current_dt.strftime('%Y-%m-%d')
    cur_data = context.indicator_data[context.indicator_data['date'] == date]

    stock_to_buy = list(cur_data.instrument[:context.stock_num])
    # 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:
        context.order_target_percent(stock, 0)
    
    # 如果当天没有买入的股票，就返回
    if len(stock_to_buy) == 0:
        print(date,'当天没有买入的股票')
        return

    # 等权重买入 
    weight =  1 / len(stock_to_buy)
    print(stock_to_buy,weight)
    # 买入
    for stock in stock_to_buy:
        context.order_target_percent(stock, weight)
 
# 交易引擎：成交回报处理函数，每个成交发生时执行一次
def m19_handle_trade_bigquant_run(context, data):
    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 m19_handle_order_bigquant_run(context, data):
    pass

# 交易引擎：盘后处理函数，每日盘后执行一次
def m19_after_trading_bigquant_run(context, data):
    pass


g = T.Graph({

    'm11': 'M.input_features.v1',
    'm11.features': """return_5 = close/shift(close, 5)
return_10 = close/shift(close, 10)
return_20 = close/shift(close, 20)
amount/mean(amount,5)
mean(amount,5)/mean(amount,20)
rank(amount)/rank(mean(amount,5))
rank(mean(amount,5))/rank(mean(amount,10))
rank(close/shift(close, 1))
rank(close/shift(close, 5))
rank(close/shift(close, 10))     
rank(close/shift(close, 1))/rank(close/shift(close, 5))    
rank(close/shift(close, 5))/rank(close/shift(close, 10))""",

    'm12': 'M.use_datasource.v1',
    'm12.datasource_id': 'bar1d_CN_CONBOND',
    'm12.start_date': '2017-01-01',
    'm12.end_date': '2020-12-31',
    'm12.m_cached': False,

    'm13': 'M.derived_feature_extractor.v3',
    'm13.input_data': T.Graph.OutputPort('m12.data'),
    'm13.features': T.Graph.OutputPort('m11.data'),
    'm13.date_col': 'date',
    'm13.instrument_col': 'instrument',
    'm13.drop_na': False,
    'm13.remove_extra_columns': False,
    'm13.user_functions': {},

    'm14': 'M.auto_labeler_on_datasource.v1',
    'm14.input_data': T.Graph.OutputPort('m12.data'),
    'm14.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, 20)

# 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)
""",
    'm14.drop_na_label': True,
    'm14.cast_label_int': True,
    'm14.date_col': 'date',
    'm14.instrument_col': 'instrument',
    'm14.user_functions': {},

    'm15': 'M.join.v3',
    'm15.data1': T.Graph.OutputPort('m14.data'),
    'm15.data2': T.Graph.OutputPort('m13.data'),
    'm15.on': 'date,instrument',
    'm15.how': 'inner',
    'm15.sort': False,

    'm18': 'M.dropnan.v1',
    'm18.input_data': T.Graph.OutputPort('m15.data'),
    'm18.m_cached': False,

    'm16': 'M.stock_ranker_train.v6',
    'm16.training_ds': T.Graph.OutputPort('m18.data'),
    'm16.features': T.Graph.OutputPort('m11.data'),
    'm16.learning_algorithm': '排序',
    'm16.number_of_leaves': 30,
    'm16.minimum_docs_per_leaf': 1000,
    'm16.number_of_trees': 20,
    'm16.learning_rate': 0.1,
    'm16.max_bins': 1023,
    'm16.feature_fraction': 1,
    'm16.data_row_fraction': 1,
    'm16.plot_charts': True,
    'm16.ndcg_discount_base': 1,
    'm16.m_lazy_run': False,

    'm1': 'M.use_datasource.v1',
    'm1.datasource_id': 'market_performance_CN_CONBOND',
    'm1.start_date': '2021-01-01',
    'm1.end_date': '2022-09-16',
    'm1.m_cached': False,

    'm2': 'M.trade_data_generation.v1',
    'm2.input': T.Graph.OutputPort('m1.data'),
    'm2.category': 'CN_STOCK',
    'm2.m_cached': False,

    'm28': 'M.use_datasource.v1',
    'm28.instruments': T.Graph.OutputPort('m2.instrument_list'),
    'm28.datasource_id': 'bar1d_CN_CONBOND',
    'm28.start_date': '2018-01-01',
    'm28.end_date': '2021-09-15',

    'm30': 'M.derived_feature_extractor.v3',
    'm30.input_data': T.Graph.OutputPort('m28.data'),
    'm30.features': T.Graph.OutputPort('m11.data'),
    'm30.date_col': 'date',
    'm30.instrument_col': 'instrument',
    'm30.drop_na': False,
    'm30.remove_extra_columns': False,
    'm30.user_functions': {},

    'm17': 'M.stock_ranker_predict.v5',
    'm17.model': T.Graph.OutputPort('m16.model'),
    'm17.data': T.Graph.OutputPort('m30.data'),
    'm17.m_lazy_run': False,

    'm19': 'M.hftrade.v2',
    'm19.instruments': T.Graph.OutputPort('m2.instrument_list'),
    'm19.options_data': T.Graph.OutputPort('m17.predictions'),
    'm19.start_date': '',
    'm19.end_date': '',
    'm19.initialize': m19_initialize_bigquant_run,
    'm19.before_trading_start': m19_before_trading_start_bigquant_run,
    'm19.handle_tick': m19_handle_tick_bigquant_run,
    'm19.handle_data': m19_handle_data_bigquant_run,
    'm19.handle_trade': m19_handle_trade_bigquant_run,
    'm19.handle_order': m19_handle_order_bigquant_run,
    'm19.after_trading': m19_after_trading_bigquant_run,
    'm19.capital_base': 1000001,
    'm19.frequency': 'daily',
    'm19.price_type': '真实价格',
    'm19.product_type': '可转债',
    'm19.before_start_days': '0',
    'm19.volume_limit': 1,
    'm19.order_price_field_buy': 'close',
    'm19.order_price_field_sell': 'close',
    'm19.benchmark': '000300.HIX',
    'm19.plot_charts': True,
    'm19.disable_cache': False,
    'm19.replay_bdb': False,
    'm19.show_debug_info': False,
    'm19.backtest_only': False,
})

# g.run({})


def m3_run_bigquant_run(
    bq_graph,
    inputs,
    trading_days_market='CN', # 使用那个市场的交易日历, TODO
    train_instruments_mid='m12', # 训练数据 证券代码列表 模块id
    test_instruments_mid='m1', # 测试数据 证券代码列表 模块id
    predict_mid='m17', # 预测 模块id
    trade_mid='m19', # 回测 模块id
    start_date='2017-01-01', # 数据开始日期
    end_date=T.live_run_param('trading_date', '2022-09-16'), # 数据结束日期
    train_update_days=22, # 更新周期，按交易日计算，每多少天更新一次
    train_update_days_for_live=None, #模拟实盘模式下的更新周期，按交易日计算，每多少天更新一次。如果需要在模拟实盘阶段使用不同的模型更新周期，可以设置这个参数
    train_data_min_days=250, # 最小数据天数，按交易日计算，所以第一个滚动的结束日期是 从开始日期到开始日期+最小数据天数
    train_data_max_days=250, # 最大数据天数，按交易日计算，0，表示没有限制，否则每一个滚动的开始日期=max(此滚动的结束日期-最大数据天数, 开始日期
    rolling_count_for_live=1, #实盘模式下滚动次数，模拟实盘模式下，取最后多少次滚动。一般在模拟实盘模式下，只用到最后一次滚动训练的模型，这里可以设置为1；如果你的滚动训练数据时间段很短，以至于期间可能没有训练数据，这里可以设置大一点。0表示没有限制
):
    def merge_datasources(input_1):
        df_list = [ds[0].read_df().set_index('date').loc[ds[1]:].reset_index() for ds in input_1]
        df = pd.concat(df_list)
        instrument_data = {
            'start_date': df['date'].min().strftime('%Y-%m-%d'),
            'end_date': df['date'].max().strftime('%Y-%m-%d'),
            'instruments': list(set(df['instrument'])),
        }
        return Outputs(data=DataSource.write_df(df), instrument_data=DataSource.write_pickle(instrument_data))

    def gen_rolling_dates(trading_days_market, start_date, end_date, train_update_days, train_update_days_for_live, train_data_min_days, train_data_max_days, rolling_count_for_live):
        # 是否实盘模式
        tdays = list(D.trading_days(market=trading_days_market, start_date=start_date, end_date=end_date)['date'])
        is_live_run = T.live_run_param('trading_date', None) is not None

        if is_live_run and train_update_days_for_live:
            train_update_days = train_update_days_for_live

        rollings = []
        train_end_date = train_data_min_days
        while train_end_date < len(tdays):
            if train_data_max_days is not None and train_data_max_days > 0:
                train_start_date = max(train_end_date - train_data_max_days, 0)
            else:
                train_start_date = 0
            rollings.append({
                'train_start_date': tdays[train_start_date].strftime('%Y-%m-%d'),
                'train_end_date': tdays[train_end_date - 1].strftime('%Y-%m-%d'),
                'test_start_date': tdays[train_end_date].strftime('%Y-%m-%d'),
                'test_end_date': tdays[min(train_end_date + train_update_days, len(tdays)) - 1].strftime('%Y-%m-%d'),
            })
            train_end_date += train_update_days

        if not rollings:
            raise Exception('没有滚动需要执行，请检查配置')

        if is_live_run and rolling_count_for_live:
            rollings = rollings[-rolling_count_for_live:]

        return rollings

    g = bq_graph

    rolling_dates = gen_rolling_dates(
        trading_days_market, start_date, end_date, train_update_days, train_update_days_for_live, train_data_min_days, train_data_max_days, rolling_count_for_live)

    # 训练和预测
    results = []
    for rolling in rolling_dates:
        parameters = {}
        # 先禁用回测
        parameters[trade_mid + '.__enabled__'] = False
        parameters[train_instruments_mid + '.start_date'] = rolling['train_start_date']
        parameters[train_instruments_mid + '.end_date'] = rolling['train_end_date']
        parameters[test_instruments_mid + '.start_date'] = rolling['test_start_date']
        parameters[test_instruments_mid + '.end_date'] = rolling['test_end_date']
        # print('------ rolling_train:', parameters)
        results.append(g.run(parameters))

    # 合并预测结果并回测
    mx = M.cached.v3(run=merge_datasources, input_1=[[result[predict_mid].predictions, result[test_instruments_mid].data.read_pickle()['start_date']] for result in results])
    parameters = {}
    parameters['*.__enabled__'] = False
    parameters[trade_mid + '.__enabled__'] = True
    parameters[trade_mid + '.instruments'] = mx.instrument_data
    parameters[trade_mid + '.options_data'] = mx.data

    trade = g.run(parameters)

    return {'rollings': results, 'trade': trade}


m3 = M.hyper_rolling_train.v1(
    run=m3_run_bigquant_run,
    run_now=True,
    bq_graph=g
)

[2022-12-05 22:18:54.920876] INFO: moduleinvoker: input_features.v1 开始运行..

[2022-12-05 22:18:54.928655] INFO: moduleinvoker: 命中缓存

[2022-12-05 22:18:54.930763] INFO: moduleinvoker: input_features.v1 运行完成[0.009895s].

[2022-12-05 22:18:54.939427] INFO: moduleinvoker: use_datasource.v1 开始运行..

[2022-12-05 22:18:55.343127] INFO: moduleinvoker: use_datasource.v1 运行完成[0.403704s].

[2022-12-05 22:18:55.349986] INFO: moduleinvoker: use_datasource.v1 开始运行..

[2022-12-05 22:18:55.506024] INFO: moduleinvoker: use_datasource.v1 运行完成[0.156035s].

[2022-12-05 22:18:55.516519] INFO: moduleinvoker: auto_labeler_on_datasource.v1 开始运行..

[2022-12-05 22:18:56.027929] INFO: 自动标注(任意数据源): 开始标注 ..

[2022-12-05 22:18:56.244072] INFO: moduleinvoker: auto_labeler_on_datasource.v1 运行完成[0.727539s].

[2022-12-05 22:18:56.253565] INFO: moduleinvoker: derived_feature_extractor.v3 开始运行..

[2022-12-05 22:18:56.772202] INFO: derived_feature_extractor: 提取完成 return_5 = close/shift(close, 5), 0.020s

[2022-12-05 22:18:56.799116] INFO: derived_feature_extractor: 提取完成 return_10 = close/shift(close, 10), 0.025s

[2022-12-05 22:18:56.854485] INFO: derived_feature_extractor: 提取完成 return_20 = close/shift(close, 20), 0.053s

[2022-12-05 22:18:57.055333] INFO: derived_feature_extractor: 提取完成 amount/mean(amount,5), 0.199s

[2022-12-05 22:18:57.411254] INFO: derived_feature_extractor: 提取完成 mean(amount,5)/mean(amount,20), 0.353s

[2022-12-05 22:18:57.625669] INFO: derived_feature_extractor: 提取完成 rank(amount)/rank(mean(amount,5)), 0.212s

[2022-12-05 22:18:58.032175] INFO: derived_feature_extractor: 提取完成 rank(mean(amount,5))/rank(mean(amount,10)), 0.404s

[2022-12-05 22:18:58.086559] INFO: derived_feature_extractor: 提取完成 rank(close/shift(close, 1)), 0.046s

[2022-12-05 22:18:58.161157] INFO: derived_feature_extractor: 提取完成 rank(close/shift(close, 5)), 0.073s

[2022-12-05 22:18:58.212117] INFO: derived_feature_extractor: 提取完成 rank(close/shift(close, 10)), 0.049s

[2022-12-05 22:18:58.316585] INFO: derived_feature_extractor: 提取完成 rank(close/shift(close, 1))/rank(close/shift(close, 5)), 0.102s

[2022-12-05 22:18:58.444554] INFO: derived_feature_extractor: 提取完成 rank(close/shift(close, 5))/rank(close/shift(close, 10)), 0.104s

[2022-12-05 22:18:59.002665] INFO: derived_feature_extractor: /data, 135983

[2022-12-05 22:18:59.216456] INFO: moduleinvoker: derived_feature_extractor.v3 运行完成[2.962878s].

[2022-12-05 22:18:59.228943] INFO: moduleinvoker: trade_data_generation.v1 开始运行..

[2022-12-05 22:18:59.409187] INFO: moduleinvoker: trade_data_generation.v1 运行完成[0.18022s].

[2022-12-05 22:18:59.421337] INFO: moduleinvoker: join.v3 开始运行..

[2022-12-05 22:19:00.218207] INFO: join: /data, 行数=5587/135983, 耗时=0.678388s

[2022-12-05 22:19:00.291273] INFO: join: 最终行数: 5587

[2022-12-05 22:19:00.298211] INFO: moduleinvoker: join.v3 运行完成[0.876881s].

[2022-12-05 22:19:00.304767] INFO: moduleinvoker: use_datasource.v1 开始运行..

[2022-12-05 22:19:00.505267] INFO: moduleinvoker: use_datasource.v1 运行完成[0.200497s].

[2022-12-05 22:19:00.517503] INFO: moduleinvoker: dropnan.v1 开始运行..

[2022-12-05 22:19:00.636391] INFO: dropnan: /data, 776/5587

[2022-12-05 22:19:00.671187] INFO: dropnan: 行数: 776/5587

[2022-12-05 22:19:00.674059] INFO: moduleinvoker: dropnan.v1 运行完成[0.156553s].

[2022-12-05 22:19:00.682209] INFO: moduleinvoker: derived_feature_extractor.v3 开始运行..

[2022-12-05 22:19:00.740257] INFO: derived_feature_extractor: 提取完成 return_5 = close/shift(close, 5), 0.003s

[2022-12-05 22:19:00.747483] INFO: derived_feature_extractor: 提取完成 return_10 = close/shift(close, 10), 0.005s

[2022-12-05 22:19:00.755226] INFO: derived_feature_extractor: 提取完成 return_20 = close/shift(close, 20), 0.005s

[2022-12-05 22:19:00.775915] INFO: derived_feature_extractor: 提取完成 amount/mean(amount,5), 0.018s

[2022-12-05 22:19:00.802990] INFO: derived_feature_extractor: 提取完成 mean(amount,5)/mean(amount,20), 0.024s

[2022-12-05 22:19:00.849861] INFO: derived_feature_extractor: 提取完成 rank(amount)/rank(mean(amount,5)), 0.045s

[2022-12-05 22:19:00.882598] INFO: derived_feature_extractor: 提取完成 rank(mean(amount,5))/rank(mean(amount,10)), 0.031s

[2022-12-05 22:19:00.892265] INFO: derived_feature_extractor: 提取完成 rank(close/shift(close, 1)), 0.007s

[2022-12-05 22:19:00.944674] INFO: derived_feature_extractor: 提取完成 rank(close/shift(close, 5)), 0.050s

[2022-12-05 22:19:00.955109] INFO: derived_feature_extractor: 提取完成 rank(close/shift(close, 10)), 0.008s

[2022-12-05 22:19:00.969711] INFO: derived_feature_extractor: 提取完成 rank(close/shift(close, 1))/rank(close/shift(close, 5)), 0.012s

[2022-12-05 22:19:00.983338] INFO: derived_feature_extractor: 提取完成 rank(close/shift(close, 5))/rank(close/shift(close, 10)), 0.012s

[2022-12-05 22:19:01.043537] INFO: derived_feature_extractor: /data, 1049

[2022-12-05 22:19:01.105077] INFO: moduleinvoker: derived_feature_extractor.v3 运行完成[0.422869s].

[2022-12-05 22:19:01.114790] INFO: moduleinvoker: stock_ranker_train.v6 开始运行..

[2022-12-05 22:19:01.195768] INFO: StockRanker: 特征预处理 ..

[2022-12-05 22:19:01.227426] INFO: StockRanker: prepare data: training ..

[2022-12-05 22:19:01.334397] INFO: StockRanker训练: c3c674c8 准备训练: 776 行数

[2022-12-05 22:19:01.336903] INFO: StockRanker训练: AI模型训练，将在776*12=0.93万数据上对模型训练进行20轮迭代训练。预计将需要1～2分钟。请耐心等待。

[2022-12-05 22:19:01.596535] INFO: StockRanker训练: 正在训练 ..

[2022-12-05 22:19:01.646753] INFO: StockRanker训练: 任务状态: Pending

[2022-12-05 22:19:11.691096] INFO: StockRanker训练: 任务状态: Running

[2022-12-05 22:20:11.974162] INFO: StockRanker训练: 任务状态: Succeeded

[2022-12-05 22:20:11.986820] ERROR: moduleinvoker: module name: stock_ranker_train, module version: v6, trackeback: Exception: 模型训练失败：可能导致错误的原因是训练数据问题，请检查训练数据, err_code=1 (c3c674c874a711edb9db8efb860ab9fc)

[2022-12-05 22:20:11.990935] ERROR: moduleinvoker: module name: hyper_rolling_train, module version: v1, trackeback: Exception: 模型训练失败：可能导致错误的原因是训练数据问题，请检查训练数据, err_code=1 (c3c674c874a711edb9db8efb860ab9fc)

---------------------------------------------------------------------------
Exception                                 Traceback (most recent call last)
<ipython-input-8-4cee75e7179e> in <module>
    311 
    312 
--> 313 m3 = M.hyper_rolling_train.v1(
    314     run=m3_run_bigquant_run,
    315     run_now=True,

<ipython-input-8-4cee75e7179e> in m3_run_bigquant_run(bq_graph, inputs, trading_days_market, train_instruments_mid, test_instruments_mid, predict_mid, trade_mid, start_date, end_date, train_update_days, train_update_days_for_live, train_data_min_days, train_data_max_days, rolling_count_for_live)
    296         parameters[test_instruments_mid + '.end_date'] = rolling['test_end_date']
    297         # print('------ rolling_train:', parameters)
--> 298         results.append(g.run(parameters))
    299 
    300     # 合并预测结果并回测

Exception: 模型训练失败：可能导致错误的原因是训练数据问题，请检查训练数据, err_code=1 (c3c674c874a711edb9db8efb860ab9fc)