{"description":"实验创建于2023/2/10","graph":{"edges":[{"to_node_id":"-127:sql","from_node_id":"-106:data"},{"to_node_id":"-106:sql1","from_node_id":"-115:data"},{"to_node_id":"-88:sql","from_node_id":"-118:data"},{"to_node_id":"-106:sql2","from_node_id":"-118:data"},{"to_node_id":"-103:data","from_node_id":"-110:predictions"},{"to_node_id":"-75:data","from_node_id":"-127:data"},{"to_node_id":"-75:validation_ds","from_node_id":"-127:data"},{"to_node_id":"-110:model","from_node_id":"-75:model"},{"to_node_id":"-110:data","from_node_id":"-88:data"}],"nodes":[{"node_id":"-106","module_id":"BigQuantSpace.sql_join_2.sql_join_2-v1","parameters":[{"name":"sql_join","value":"WITH\nsql1 AS (\n {sql1}\n),\nsql2 AS (\n {sql2}\n)\n\nSELECT * FROM sql1 JOIN sql2 USING (date, instrument)\nORDER BY date, instrument\n","type":"Literal","bound_global_parameter":null}],"input_ports":[{"name":"sql1","node_id":"-106"},{"name":"sql2","node_id":"-106"}],"output_ports":[{"name":"data","node_id":"-106"}],"cacheable":true,"seq_num":3,"comment":"3. 合并因子和标注数据","comment_collapsed":false,"x":-453,"y":-242},{"node_id":"-115","module_id":"BigQuantSpace.input_features_dai.input_features_dai-v6","parameters":[{"name":"sql","value":"/*\n使用DAI SQL为量化模型预测生成标签数据。标签反映了未来5日的收益率，并且被离散化为20个桶，每个桶代表一个收益率范围。这样，我们就可以训练模型来预测未来的收益率范围，而不仅仅是具体的收益率值。\n\n1. 首先定义了一个名为label_data的临时表，用于计算和存储未来5日收益率，其1%和99%分位数，以及离散化后的收益率（被分为20个桶，每个桶代表一个收益率范围）。\n2. 对未来5日收益率进行了截断处理，只保留在1%和99%分位数之间的值。\n3. 选择了标签值不为空，并且非涨跌停（未来一天的最高价不等于最低价）的数据\n4. 从这个临时表中选择了日期、股票代码和标签字段，以供进模型训练。\n*/\n\nSELECT\n -- 计算的是未来5日的收益率。这是通过将5天后的收盘价除以第二天的开盘价得到的。这里使用的是一个叫做m_lead的函数，它可以获取某个字段在未来某天的值。\n -- _future_return 是一个中间变量名，以 _ 开始的别名列不会在最终结果返回\n m_lead(close, 5) / m_lead(open, 1) AS _future_return,\n\n -- 计算未来5日收益率的1%分位数。all_quantile_cont是一个分位数函数，它能够计算出某个字段值的分位数，这里是计算1%的分位数。\n c_quantile_cont(_future_return, 0.01) AS _future_return_1pct,\n\n -- 计算未来5日收益率的99%分位数。同样，all_quantile_cont函数用来计算99%的分位数。\n c_quantile_cont(_future_return, 0.99) AS _future_return_99pct,\n\n -- 对未来5日收益率进行截断处理，值位于1%和99%分位数之间的数据被保留，超过这个范围的值将被设为边界值。\n clip(_future_return, _future_return_1pct, _future_return_99pct) AS _clipped_return,\n\n -- 将截断后的未来5日收益率分为20个不同的桶，每个桶代表一个范围。all_wbins函数用于将数据离散化为多个桶。\n cbins(_clipped_return, 20) AS _binned_return,\n\n -- 将离散化后的数据作为标签使用，这是我们预测的目标。\n _binned_return AS label,\n\n -- 日期，这是每个股票每天的数据\n date,\n\n -- 股票代码，代表每一支股票\n instrument\n-- 从cn_stock_bar1d这个表中选择数据，这个表存储的是股票的日线数据\nFROM cn_stock_bar1d\nQUALIFY\n COLUMNS(*) IS NOT NULL\n -- 标签值不为空，且非涨跌停(未来一天的最高价不等于最低价)\n AND m_lead(high, 1) != m_lead(low, 1)\nORDER BY date, instrument","type":"Literal","bound_global_parameter":null}],"input_ports":[],"output_ports":[{"name":"data","node_id":"-115"}],"cacheable":true,"seq_num":1,"comment":"1. 设置预测目标，例如：根据未来5日收益率预测","comment_collapsed":false,"x":-453,"y":-383},{"node_id":"-118","module_id":"BigQuantSpace.input_features_dai.input_features_dai-v6","parameters":[{"name":"sql","value":"-- 使用DAI SQL获取数据，构建因子等，如下是一个例子作为参考\n-- DAI SQL 语法: https://bigquant.com/wiki/doc/dai-PLSbc1SbZX#h-sql%E5%85%A5%E9%97%A8%E6%95%99%E7%A8%8B\n\nSELECT\n -- 在这里输入因子表达式\n -- DAI SQL 算子/函数: https://bigquant.com/wiki/doc/dai-PLSbc1SbZX#h-%E5%87%BD%E6%95%B0\n -- 数据&字段: 数据文档 https://bigquant.com/data/home\n -- 在时间截面的total_market_cap排名\n -- _return_0 是中间变量，以下划线开始的变量是中间值，不会出现在最后的因子里\n close_0 / m_lag(close_0, 1) AS _return_0,\n close_0 / m_lag(close_0, 6) AS return_5,\n close_0 / m_lag(close_0, 11) AS return_10,\n close_0 / m_lag(close_0, 21) AS return_20,\n m_avg(amount_0, 5) AS avg_amount_5,\n m_avg(amount_0, 20) AS avg_amount_20,\n m_avg(amount_0, 10) AS _avg_amount_10,\n c_pct_rank(amount_0) / c_pct_rank(avg_amount_5),\n c_pct_rank(avg_amount_5) / c_pct_rank(_avg_amount_10),\n c_pct_rank(_return_0) AS rank_return_0,\n c_pct_rank(return_5) AS rank_return_5,\n c_pct_rank(return_10) AS rank_return_10,\n rank_return_0 / rank_return_5,\n rank_return_5 / rank_return_10,\n\n price_limit_status,\n IF(price_limit_status=3, 1, 0) AS f1,\n\n -- 日期和股票代码\n date,\n instrument\n-- 预计算因子和数据 cn_stock_factors https://bigquant.com/data/datasources/cn_stock_factors\nFROM cn_stock_factors\nWHERE\n -- 剔除ST股票\n st_status = 0\n -- 非停牌股\n AND suspended = 0\n -- 不属于北交所\n AND list_sector < 4\n-- 窗口函数内容过滤需要放在 QUALIFY 这里\nQUALIFY\n -- 去掉有空值的行\n COLUMNS(*) IS NOT NULL\n AND m_lag(price_limit_status, 1) == 3\n\n-- 按日期、股票代码排序\nORDER BY date, instrument\n","type":"Literal","bound_global_parameter":null}],"input_ports":[],"output_ports":[{"name":"data","node_id":"-118"}],"cacheable":true,"seq_num":2,"comment":"2. 编写因子","comment_collapsed":false,"x":-110,"y":-380},{"node_id":"-110","module_id":"BigQuantSpace.stock_ranker_dai_predict.stock_ranker_dai_predict-v3","parameters":[],"input_ports":[{"name":"model","node_id":"-110"},{"name":"data","node_id":"-110"}],"output_ports":[{"name":"predictions","node_id":"-110"}],"cacheable":true,"seq_num":14,"comment":"7. 预测","comment_collapsed":false,"x":-227,"y":153},{"node_id":"-103","module_id":"BigQuantSpace.bigtrader.bigtrader-v7","parameters":[{"name":"start_date","value":"","type":"Literal","bound_global_parameter":null},{"name":"end_date","value":"","type":"Literal","bound_global_parameter":null},{"name":"initialize","value":"# 交易引擎：初始化函数，只执行一次\ndef bigquant_run(context):\n # context.ranker_prediction = context.options['data'].read_bdb(as_type=pd.DataFrame)\n # 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\n context.set_commission(PerOrder(buy_cost=0.0003, sell_cost=0.0013, min_cost=5))\n # 预测数据，通过 options 传入进来，使用 read_df 函数，加载到内存 (DataFrame)\n # 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\n stock_count = 5\n # 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, ..]\n context.stock_weights = T.norm([1 / math.log(i + 2) for i in range(0, stock_count)])\n # 设置每只股票占用的最大资金比例\n context.max_cash_per_instrument = 0.2\n context.options['hold_days'] = 5\n","type":"Literal","bound_global_parameter":null},{"name":"before_trading_start","value":"# 交易引擎：每个单位时间开盘前调用一次。\ndef bigquant_run(context, data):\n # 盘前处理，订阅行情等\n pass\n","type":"Literal","bound_global_parameter":null},{"name":"handle_tick","value":"# 交易引擎：tick数据处理函数，每个tick执行一次\ndef bigquant_run(context, tick):\n pass\n","type":"Literal","bound_global_parameter":null},{"name":"handle_data","value":"# 回测引擎：每日数据处理函数，每天执行一次\ndef bigquant_run(context, data):\n # 按日期过滤得到今日的预测数据\n ranker_prediction = context.data[context.data.date == data.current_dt.strftime('%Y-%m-%d')]\n\n # 1. 资金分配\n # 平均持仓时间是hold_days，每日都将买入股票，每日预期使用 1/hold_days 的资金\n # 实际操作中，会存在一定的买入误差，所以在前hold_days天，等量使用资金；之后，尽量使用剩余资金（这里设置最多用等量的1.5倍）\n is_staging = context.trading_day_index < context.options['hold_days'] # 是否在建仓期间（前 hold_days 天）\n cash_avg = context.portfolio.portfolio_value / context.options['hold_days']\n cash_for_buy = min(context.portfolio.cash, (1 if is_staging else 1.5) * cash_avg)\n cash_for_sell = cash_avg - (context.portfolio.cash - cash_for_buy)\n positions = {e: p.amount * p.last_sale_price\n for e, p in context.portfolio.positions.items()}\n\n # 2. 生成卖出订单：hold_days天之后才开始卖出；对持仓的股票，按机器学习算法预测的排序末位淘汰\n if not is_staging and cash_for_sell > 0:\n equities = {e: e for e, p in context.portfolio.positions.items()}\n instruments = list(reversed(list(ranker_prediction.instrument[ranker_prediction.instrument.apply(\n lambda x: x in equities)])))\n\n for instrument in instruments:\n context.order_target(context.symbol(instrument), 0)\n cash_for_sell -= positions[instrument]\n if cash_for_sell <= 0:\n break\n\n # 3. 生成买入订单：按机器学习算法预测的排序，买入前面的stock_count只股票\n buy_cash_weights = context.stock_weights\n buy_instruments = list(ranker_prediction.instrument[:len(buy_cash_weights)])\n max_cash_per_instrument = context.portfolio.portfolio_value * context.max_cash_per_instrument\n for i, instrument in enumerate(buy_instruments):\n cash = cash_for_buy * buy_cash_weights[i]\n if cash > max_cash_per_instrument - positions.get(instrument, 0):\n # 确保股票持仓量不会超过每次股票最大的占用资金量\n cash = max_cash_per_instrument - positions.get(instrument, 0)\n if cash > 0:\n context.order_value(context.symbol(instrument), cash)\n","type":"Literal","bound_global_parameter":null},{"name":"handle_trade","value":"# 交易引擎：成交回报处理函数，每个成交发生时执行一次\ndef bigquant_run(context, trade):\n pass\n","type":"Literal","bound_global_parameter":null},{"name":"handle_order","value":"# 交易引擎：委托回报处理函数，每个委托变化时执行一次\ndef bigquant_run(context, order):\n pass\n","type":"Literal","bound_global_parameter":null},{"name":"after_trading","value":"# 交易引擎：盘后处理函数，每日盘后执行一次\ndef bigquant_run(context, data):\n pass\n","type":"Literal","bound_global_parameter":null},{"name":"capital_base","value":1000000,"type":"Literal","bound_global_parameter":null},{"name":"frequency","value":"daily","type":"Literal","bound_global_parameter":null},{"name":"product_type","value":"股票","type":"Literal","bound_global_parameter":null},{"name":"before_start_days","value":0,"type":"Literal","bound_global_parameter":null},{"name":"volume_limit","value":1,"type":"Literal","bound_global_parameter":null},{"name":"order_price_field_buy","value":"open","type":"Literal","bound_global_parameter":null},{"name":"order_price_field_sell","value":"close","type":"Literal","bound_global_parameter":null},{"name":"benchmark","value":"000300.SH","type":"Literal","bound_global_parameter":null},{"name":"plot_charts","value":"True","type":"Literal","bound_global_parameter":null},{"name":"disable_cache","value":"True","type":"Literal","bound_global_parameter":null},{"name":"debug","value":"False","type":"Literal","bound_global_parameter":null},{"name":"backtest_only","value":"False","type":"Literal","bound_global_parameter":null}],"input_ports":[{"name":"data","node_id":"-103"},{"name":"options_data","node_id":"-103"},{"name":"history_ds","node_id":"-103"},{"name":"benchmark_ds","node_id":"-103"}],"output_ports":[{"name":"raw_perf","node_id":"-103"}],"cacheable":true,"seq_num":8,"comment":"8. 交易，设置处理函数等","comment_collapsed":false,"x":-129,"y":293},{"node_id":"-127","module_id":"BigQuantSpace.extract_data_dai.extract_data_dai-v7","parameters":[{"name":"start_date","value":"2018-01-01","type":"Literal","bound_global_parameter":null},{"name":"start_date_bound_to_trading_date","value":"False","type":"Literal","bound_global_parameter":null},{"name":"end_date","value":"2020-12-31","type":"Literal","bound_global_parameter":null},{"name":"end_date_bound_to_trading_date","value":"False","type":"Literal","bound_global_parameter":null},{"name":"before_start_days","value":90,"type":"Literal","bound_global_parameter":null},{"name":"debug","value":"False","type":"Literal","bound_global_parameter":null}],"input_ports":[{"name":"sql","node_id":"-127"}],"output_ports":[{"name":"data","node_id":"-127"}],"cacheable":true,"seq_num":4,"comment":"4. 训练数据，设置训练开始时间和结束时间","comment_collapsed":false,"x":-451,"y":-120},{"node_id":"-75","module_id":"BigQuantSpace.stock_ranker_dai_train.stock_ranker_dai_train-v2","parameters":[{"name":"learning_algorithm","value":"排序","type":"Literal","bound_global_parameter":null},{"name":"number_of_leaves","value":"30","type":"Literal","bound_global_parameter":null},{"name":"min_docs_per_leaf","value":"200","type":"Literal","bound_global_parameter":null},{"name":"number_of_trees","value":"10","type":"Literal","bound_global_parameter":null},{"name":"learning_rate","value":"0.1","type":"Literal","bound_global_parameter":null},{"name":"max_bins","value":"1023","type":"Literal","bound_global_parameter":null},{"name":"feature_fraction","value":"1","type":"Literal","bound_global_parameter":null},{"name":"data_row_fraction","value":"1","type":"Literal","bound_global_parameter":null},{"name":"plot_charts","value":"True","type":"Literal","bound_global_parameter":null},{"name":"ndcg_discount_base","value":"1","type":"Literal","bound_global_parameter":null}],"input_ports":[{"name":"data","node_id":"-75"},{"name":"validation_ds","node_id":"-75"},{"name":"base_model","node_id":"-75"}],"output_ports":[{"name":"model","node_id":"-75"}],"cacheable":true,"seq_num":5,"comment":"5. 使用StockRanker算法训练","comment_collapsed":false,"x":-447,"y":28},{"node_id":"-88","module_id":"BigQuantSpace.extract_data_dai.extract_data_dai-v7","parameters":[{"name":"start_date","value":"2021-01-01","type":"Literal","bound_global_parameter":null},{"name":"start_date_bound_to_trading_date","value":"True","type":"Literal","bound_global_parameter":null},{"name":"end_date","value":"2021-12-31","type":"Literal","bound_global_parameter":null},{"name":"end_date_bound_to_trading_date","value":"True","type":"Literal","bound_global_parameter":null},{"name":"before_start_days","value":90,"type":"Literal","bound_global_parameter":null},{"name":"debug","value":"False","type":"Literal","bound_global_parameter":null}],"input_ports":[{"name":"sql","node_id":"-88"}],"output_ports":[{"name":"data","node_id":"-88"}],"cacheable":true,"seq_num":6,"comment":"6. 预测数据，设置预测时间，开启模拟交易时绑定交易日期","comment_collapsed":false,"x":-107,"y":-119}],"node_layout":"<node_postions><node_position Node='-106' Position='-453,-242,200,200'/><node_position Node='-115' Position='-453,-383,200,200'/><node_position Node='-118' Position='-110,-380,200,200'/><node_position Node='-110' Position='-227,153,200,200'/><node_position Node='-103' Position='-129,293,200,200'/><node_position Node='-127' Position='-451,-120,200,200'/><node_position Node='-75' Position='-447,28,200,200'/><node_position Node='-88' Position='-107,-119,200,200'/></node_postions>"},"nodes_readonly":false,"studio_version":"v2"}
[2023-11-21 20:54:40.641503] INFO: moduleinvoker:603499493.py:111: input_features_dai.v6 开始运行..
[2023-11-21 20:54:40.682851] INFO: moduleinvoker:603499493.py:111: 命中缓存
[2023-11-21 20:54:40.687964] INFO: moduleinvoker:603499493.py:111: input_features_dai.v6 运行完成[0.046524s].
[2023-11-21 20:54:40.735560] INFO: moduleinvoker:603499493.py:156: input_features_dai.v6 开始运行..
[2023-11-21 20:54:40.864362] INFO: moduleinvoker:603499493.py:156: input_features_dai.v6 运行完成[0.128782s].
[2023-11-21 20:54:40.923963] INFO: moduleinvoker:603499493.py:208: extract_data_dai.v7 开始运行..
2023-11-21 20:54:40 [info ] start_date='2021-01-01', end_date='2021-12-31', query_start_date='2020-10-03' ..
2023-11-21 20:54:57 [info ] data extracted: (7914, 16)
[2023-11-21 20:54:57.875300] INFO: moduleinvoker:603499493.py:208: extract_data_dai.v7 运行完成[16.951284s].
[2023-11-21 20:54:57.948310] INFO: moduleinvoker:603499493.py:219: sql_join_2.v1 开始运行..
[2023-11-21 20:54:58.170881] INFO: moduleinvoker:603499493.py:219: sql_join_2.v1 运行完成[0.222564s].
[2023-11-21 20:54:58.225320] INFO: moduleinvoker:603499493.py:236: extract_data_dai.v7 开始运行..
2023-11-21 20:54:58 [info ] start_date='2018-01-01', end_date='2020-12-31', query_start_date='2017-10-03' ..
2023-11-21 20:55:31 [info ] data extracted: (15877, 17)
[2023-11-21 20:55:32.054277] INFO: moduleinvoker:603499493.py:236: extract_data_dai.v7 运行完成[33.829051s].
[2023-11-21 20:55:32.231305] INFO: moduleinvoker:603499493.py:247: stock_ranker_dai_train.v2 开始运行..
2023-11-21 20:55:32 [info ] data loaded: train=(15877, 17), validation=(15877, 17)
2023-11-21 20:55:32 [info ] 数据预处理 ..
2023-11-21 20:55:32 [info ] 特征预处理 ..
2023-11-21 20:55:32 [info ] prepare data: training ..
2023-11-21 20:55:33 [info ] prepare data: validation ..
2023-11-21 20:55:33 [info ] 训练集预处理完成 rows=15877, cols=17
2023-11-21 20:55:33 [info ] 验证集预处理完成 rows=15877, cols=17
2023-11-21 20:55:33 [info ] 预处理特征
2023-11-21 20:55:33 [info ] 开始训练 e1119272e85f4544baaf2e1167c6ab5b
2023-11-21 20:55:35 [info ] numThreads = 8
2023-11-21 20:55:39 [info ] 00:00:04.5178823: Starting to train ...
2023-11-21 20:55:40 [info ] 00:00:04.8633224: Finished iteration 1
2023-11-21 20:55:40 [info ] 00:00:04.9303207: Finished iteration 2
2023-11-21 20:55:40 [info ] 00:00:04.9717366: Finished iteration 3
2023-11-21 20:55:40 [info ] 00:00:05.0265500: Finished iteration 4
2023-11-21 20:55:40 [info ] 00:00:05.0761613: Finished iteration 5
2023-11-21 20:55:40 [info ] 00:00:05.1840398: Finished iteration 6
2023-11-21 20:55:40 [info ] 00:00:05.3035926: Finished iteration 7
2023-11-21 20:55:40 [info ] 00:00:05.4212088: Finished iteration 8
2023-11-21 20:55:40 [info ] 00:00:05.4978351: Finished iteration 9
2023-11-21 20:55:40 [info ] 00:00:05.5622598: Finished iteration 10
2023-11-21 20:55:41 [info ] 训练完成
2023-11-21 20:55:41 [info ] 保存模型
[2023-11-21 20:55:42.060307] INFO: bigcharts.impl.render:render.py:408:render_chart Data is None, skip loading it to chart.
[2023-11-21 20:55:42.390006] INFO: moduleinvoker:603499493.py:247: stock_ranker_dai_train.v2 运行完成[10.158799s].
[2023-11-21 20:55:42.678728] INFO: moduleinvoker:603499493.py:263: stock_ranker_dai_predict.v3 开始运行..
2023-11-21 20:55:42 [info ] load data (7914, 16)
[2023-11-21 20:55:43.969115] INFO: moduleinvoker:603499493.py:263: stock_ranker_dai_predict.v3 运行完成[1.290386s].
[2023-11-21 20:55:44.095924] INFO: moduleinvoker:603499493.py:269: bigtrader.v7 开始运行..
2023-11-21 20:55:44 [info ] read data ..
2023-11-21 20:55:44 [info ] start_date='2021-01-04', end_date='2021-12-31', instruments=1900
2023-11-21 20:55:44 [info ] biglearning V1.5.4
2023-11-21 20:55:44 [info ] bigtrader v1.10.5 2023-11-12
2023-11-21 20:56:57 [info ] backtest done, raw_perf_ds:dai.DataSource("_700ca028f3f3484aaf7ad377de822524")
[2023-11-21 20:57:01.994461] INFO: bigcharts.impl.render:render.py:408:render_chart Data is None, skip loading it to chart.
[2023-11-21 20:57:06.770088] INFO: moduleinvoker:603499493.py:269: bigtrader.v7 运行完成[82.674138s].
- 收益率51.61%
- 年化收益率51.35%
- 基准收益率-6.21%
- 阿尔法0.58
- 贝塔0.45
- 夏普比率2.23
- 胜率0.82
- 盈亏比10.15
- 收益波动率18.8%
- 信息比率0.16
- 最大回撤12.37%
| 日期 |
时间 |
股票代码 |
股票名称 |
买/卖 |
数量 |
成交价 |
总成本 |
交易佣金 |
| Loading... (need help?) |
| 日期 |
标的代码 |
标的名称 |
持仓均价 |
收盘价 |
数量 |
持仓价值 |
收益 |
| Loading... (need help?) |
| 时间 |
级别 |
内容 |
| Loading... (need help?) |