克隆策略

使用深度学习技术预测股票价格¶

版本 v1.0

目录¶

### 深度学习策略的交易规则
### 策略构建步骤
### 策略的实现

正文¶

一、深度学习策略的交易规则¶

买入条件：预测的上涨概率>0.5，则买入或保持已有持仓。
卖出条件 :预测的上涨概率<0.5，则卖出已有股票。

二、策略构建步骤¶

1、确定股票池和数据起止时间¶

在证券代码列表m24和m28模块中输入要回测的单只股票，以及数据的起止日期(分别为训练集和验证集)。

2、确定因子¶

在输入特征列表m8模块中输入用于预测的N个因子表达式。

3、获取基础数据¶

通过基础特征数据抽取模块m22和m16获取指定股票池的基础数据，如收盘价等字段。

4、确定并计算模型标注¶

通过自动标注股票模块m21计算需要的标注指标，本例中首先计算未来10天收益，然后根据其正负来给每日数据标注1或0，来标识涨跌。

5、抽取因子数据¶

通过衍生数据抽取模块m23和m26计算因子数据。

6、合并标注与因子数据¶

通过连接数据m17模块合并因子数据和标注数据。

7、生成序列窗口滚动数据集¶

通过序列窗口滚动(深度学习)模块m25和m27将训练集和预测集的数据生成固定窗口长度的数据序列，为后续模型训练和预测做准备。

8、构建LSTM + CNN模型构架¶

在画布左侧模块列表中依次拖入输入层模块、Reshape层模块、Conv2D层模块、Reshape层模块、LSTM层模块、Dropout层模块和全连接层模块(两组)，构成深度学习网络构架，

最后通过“构建(深度学习)”模块组装各层。这里需要注意：

输入层的shape参数是窗口滚动数据集的大小 X 因子数量 , 本例为 50 行 X 5个因子

ReShape层的参数是窗口滚动数据集的大小 X 因子数量 X 1 ，本例为 50 行 X 5个因子 X1

Conv2D层中的 kernel_size参数是滑动窗口的尺寸，本例中使用 3行 X 5列的窗口, 每次滑动的步长为 1行 X 1列，卷积核数目为32，这里的窗口设置决定了后面ReShape层的参数

ReShape层中的target_shape 参数，这是由窗口滚动数据集 X 因子数量和 Conv2D层中设置的窗口尺寸以及步长决定的。本例中 50行 X 5因子的输入数据，使用 3行 X5列的窗口滑动取数据，

每次移动1行，共计可以得到48次数据(即可以通过滑动3行 X 5列的窗口48次来获取完整的数据)，因此target_shape= 48 X 卷积核数32

LSTM层的输出空间维度设置为卷积核数32，并设置激活函数

Dropout层是防止过度拟合采用的主动裁剪数据技术，这里设置rate 为0.8

全连接层共两层，第一层的输出空间维度与LSTM的输出维度保持一致为32，第二层将第一层的32维数据转变为1维数据输出，即获取预测的label值，此例为0到1之间的连续值，可以认为是上涨的概率。

9、训练深度学习模型¶

在画布左侧模块列表中拖入“训练(深度学习)”模块m6,设置属性中的优化器、目标函数、评估指标、每次训练的数据量batch_size、迭代次数epochs和GPU的数量以及日志输出频率。

10、使用深度学习模型预测¶

在画布左侧模块列表中拖入“预测(深度学习)”模块m7，并将“训练(深度学习)”模块m6的模型输出和验证集的序列窗口滚动数据集传给预测模块，通过预测模块即根据股票验证集的数据预测上涨的概率。

11、将预测结果与时间拼接¶

通过自定义模块m2将预测的每个滚动序列窗口的最后一个值最为当日的预测结果，并与预测集数据的时间列拼接，形成最终的每日预测结果。

12、根据模型预测结果构建策略¶

如果当日预测的上涨概率大于0.5，则保持持仓或买入
如果当日预测的上涨概率小于0.5，则卖出股票或保持空仓。

13、模拟回测¶

通过 trade 模块中的初始化函数定义交易手续费和滑点，通过 context.prediction 获取每日的上涨概率预测结果；
通过 trade 模块中的主函数(handle函数)查看每日的买卖交易信号，按照买卖原则执行相应的买入/卖出操作。

三、策略的实现¶

可视化策略实现如下：

In [2]:

# 本代码由可视化策略环境自动生成 2019年10月12日 09:40
# 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):

    test_data = input_2.read_pickle()
    pred_label = input_1.read_pickle()
    pred_result = pred_label.reshape(pred_label.shape[0]) 
    dt = input_3.read_df()['date'][-1*len(pred_result):]
    pred_df = pd.Series(pred_result, index=dt)
    ds = DataSource.write_df(pred_df)
    
    return Outputs(data_1=ds)

# 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

# 回测引擎：初始化函数，只执行一次
def m1_initialize_bigquant_run(context):
    # 加载预测数据
    context.prediction = 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))
# 回测引擎：每日数据处理函数，每天执行一次
def m1_handle_data_bigquant_run(context, data):
    # 按日期过滤得到今日的预测数据
    try:
        prediction = context.prediction[data.current_dt.strftime('%Y-%m-%d')]
    except KeyError as e:
        return
    
    instrument = context.instruments[0]
    sid = context.symbol(instrument)
    cur_position = context.portfolio.positions[sid].amount
    
    # 交易逻辑
    if prediction > 0.5 and cur_position == 0:
        context.order_target_percent(context.symbol(instrument), 1)
        print(data.current_dt, '买入！')
        
    elif prediction < 0.5 and cur_position > 0:
        context.order_target_percent(context.symbol(instrument), 0)
        print(data.current_dt, '卖出！')
    
# 回测引擎：准备数据，只执行一次
def m1_prepare_bigquant_run(context):
    pass

# 回测引擎：每个单位时间开始前调用一次，即每日开盘前调用一次。
def m1_before_trading_start_bigquant_run(context, data):
    pass


m3 = M.dl_layer_input.v1(
    shape='50,5',
    batch_shape='',
    dtype='float32',
    sparse=False,
    name=''
)

m13 = M.dl_layer_reshape.v1(
    inputs=m3.data,
    target_shape='50,5,1',
    name=''
)

m14 = M.dl_layer_conv2d.v1(
    inputs=m13.data,
    filters=32,
    kernel_size='3,5',
    strides='1,1',
    padding='valid',
    data_format='channels_last',
    dilation_rate='1,1',
    activation='relu',
    use_bias=True,
    kernel_initializer='glorot_uniform',
    bias_initializer='Zeros',
    kernel_regularizer='None',
    kernel_regularizer_l1=0,
    kernel_regularizer_l2=0,
    bias_regularizer='None',
    bias_regularizer_l1=0,
    bias_regularizer_l2=0,
    activity_regularizer='None',
    activity_regularizer_l1=0,
    activity_regularizer_l2=0,
    kernel_constraint='None',
    bias_constraint='None',
    name=''
)

m15 = M.dl_layer_reshape.v1(
    inputs=m14.data,
    target_shape='48,32',
    name=''
)

m4 = M.dl_layer_lstm.v1(
    inputs=m15.data,
    units=32,
    activation='tanh',
    recurrent_activation='hard_sigmoid',
    use_bias=True,
    kernel_initializer='glorot_uniform',
    recurrent_initializer='Orthogonal',
    bias_initializer='Ones',
    unit_forget_bias=True,
    kernel_regularizer='None',
    kernel_regularizer_l1=0,
    kernel_regularizer_l2=0,
    recurrent_regularizer='None',
    recurrent_regularizer_l1=0,
    recurrent_regularizer_l2=0,
    bias_regularizer='None',
    bias_regularizer_l1=0,
    bias_regularizer_l2=0,
    activity_regularizer='None',
    activity_regularizer_l1=0,
    activity_regularizer_l2=0,
    kernel_constraint='None',
    recurrent_constraint='None',
    bias_constraint='None',
    dropout=0,
    recurrent_dropout=0,
    return_sequences=False,
    implementation='0',
    name=''
)

m11 = M.dl_layer_dropout.v1(
    inputs=m4.data,
    rate=0.8,
    noise_shape='',
    name=''
)

m10 = M.dl_layer_dense.v1(
    inputs=m11.data,
    units=32,
    activation='tanh',
    use_bias=True,
    kernel_initializer='glorot_uniform',
    bias_initializer='Zeros',
    kernel_regularizer='None',
    kernel_regularizer_l1=0,
    kernel_regularizer_l2=0,
    bias_regularizer='None',
    bias_regularizer_l1=0,
    bias_regularizer_l2=0,
    activity_regularizer='None',
    activity_regularizer_l1=0,
    activity_regularizer_l2=0,
    kernel_constraint='None',
    bias_constraint='None',
    name=''
)

m12 = M.dl_layer_dropout.v1(
    inputs=m10.data,
    rate=0.8,
    noise_shape='',
    name=''
)

m9 = M.dl_layer_dense.v1(
    inputs=m12.data,
    units=1,
    activation='sigmoid',
    use_bias=True,
    kernel_initializer='glorot_uniform',
    bias_initializer='Zeros',
    kernel_regularizer='None',
    kernel_regularizer_l1=0,
    kernel_regularizer_l2=0,
    bias_regularizer='None',
    bias_regularizer_l1=0,
    bias_regularizer_l2=0,
    activity_regularizer='None',
    activity_regularizer_l1=0,
    activity_regularizer_l2=0,
    kernel_constraint='None',
    bias_constraint='None',
    name=''
)

m5 = M.dl_model_init.v1(
    inputs=m3.data,
    outputs=m9.data
)

m8 = M.input_features.v1(
    features="""(close_0/close_1-1)*10
(high_0/high_1-1)*10
(low_0/low_1-1)*10
(open_0/open_1-1)*10
(volume_0/volume_1-1)*10"""
)

m24 = M.instruments.v2(
    start_date='2015-01-01',
    end_date=T.live_run_param('trading_date', '2017-03-01'),
    market='CN_STOCK_A',
    instrument_list='600009.SHA',
    max_count=0
)

m21 = M.advanced_auto_labeler.v2(
    instruments=m24.data,
    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
#   添加benchmark_前缀，可使用对应的benchmark数据
# 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)
where(shift(close, -10) / close -1>0,1,0)

# 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)
""",
    start_date='',
    end_date='',
    benchmark='000300.SHA',
    drop_na_label=True,
    cast_label_int=True,
    user_functions={}
)

m22 = M.general_feature_extractor.v7(
    instruments=m24.data,
    features=m8.data,
    start_date='',
    end_date='',
    before_start_days=90
)

m23 = M.derived_feature_extractor.v3(
    input_data=m22.data,
    features=m8.data,
    date_col='date',
    instrument_col='instrument',
    drop_na=False,
    remove_extra_columns=False,
    user_functions={}
)

m17 = M.join.v3(
    data1=m21.data,
    data2=m23.data,
    on='date',
    how='inner',
    sort=True
)

m18 = M.dropnan.v1(
    input_data=m17.data
)

m25 = M.dl_convert_to_bin.v2(
    input_data=m18.data,
    features=m8.data,
    window_size=50,
    feature_clip=5,
    flatten=False,
    window_along_col=''
)

m6 = M.dl_model_train.v1(
    input_model=m5.data,
    training_data=m25.data,
    optimizer='Adam',
    loss='binary_crossentropy',
    metrics='accuracy',
    batch_size=2048,
    epochs=10,
    n_gpus=1,
    verbose='1:输出进度条记录'
)

m28 = M.instruments.v2(
    start_date='2017-03-01',
    end_date='2019-01-01',
    market='CN_STOCK_A',
    instrument_list='600009.SHA',
    max_count=0
)

m16 = M.general_feature_extractor.v7(
    instruments=m28.data,
    features=m8.data,
    start_date='',
    end_date='',
    before_start_days=90
)

m26 = M.derived_feature_extractor.v3(
    input_data=m16.data,
    features=m8.data,
    date_col='date',
    instrument_col='instrument',
    drop_na=False,
    remove_extra_columns=False,
    user_functions={}
)

m20 = M.dropnan.v1(
    input_data=m26.data
)

m27 = M.dl_convert_to_bin.v2(
    input_data=m20.data,
    features=m8.data,
    window_size=50,
    feature_clip=5,
    flatten=False,
    window_along_col=''
)

m7 = M.dl_model_predict.v1(
    trained_model=m6.data,
    input_data=m27.data,
    batch_size=10240,
    n_gpus=0,
    verbose='2:每个epoch输出一行记录'
)

m2 = M.cached.v3(
    input_1=m7.data,
    input_2=m27.data,
    input_3=m20.data,
    run=m2_run_bigquant_run,
    post_run=m2_post_run_bigquant_run,
    input_ports='',
    params='{}',
    output_ports=''
)

m1 = M.trade.v4(
    instruments=m28.data,
    options_data=m2.data_1,
    start_date='',
    end_date='',
    initialize=m1_initialize_bigquant_run,
    handle_data=m1_handle_data_bigquant_run,
    prepare=m1_prepare_bigquant_run,
    before_trading_start=m1_before_trading_start_bigquant_run,
    volume_limit=0.025,
    order_price_field_buy='open',
    order_price_field_sell='close',
    capital_base=1000000,
    auto_cancel_non_tradable_orders=True,
    data_frequency='daily',
    price_type='真实价格',
    product_type='股票',
    plot_charts=True,
    backtest_only=False,
    benchmark=''
)

[2019-10-12 09:24:07.108028] INFO: bigquant: cached.v3 开始运行..

[2019-10-12 09:24:07.155895] INFO: bigquant: 命中缓存

[2019-10-12 09:24:07.158261] INFO: bigquant: cached.v3 运行完成[0.050269s].

[2019-10-12 09:24:07.203340] INFO: bigquant: input_features.v1 开始运行..

[2019-10-12 09:24:07.258154] INFO: bigquant: 命中缓存

[2019-10-12 09:24:07.260520] INFO: bigquant: input_features.v1 运行完成[0.057193s].

[2019-10-12 09:24:07.274096] INFO: bigquant: instruments.v2 开始运行..

[2019-10-12 09:24:07.339699] INFO: bigquant: instruments.v2 运行完成[0.06559s].

[2019-10-12 09:24:07.347389] INFO: bigquant: advanced_auto_labeler.v2 开始运行..

[2019-10-12 09:24:07.576235] INFO: 自动标注(股票): 加载历史数据: 525 行

[2019-10-12 09:24:07.578543] INFO: 自动标注(股票): 开始标注 ..

[2019-10-12 09:24:07.679810] INFO: bigquant: advanced_auto_labeler.v2 运行完成[0.332399s].

[2019-10-12 09:24:07.807610] INFO: bigquant: general_feature_extractor.v7 开始运行..

[2019-10-12 09:24:07.964461] INFO: 基础特征抽取: 年份 2014, 特征行数=61

[2019-10-12 09:24:07.995513] INFO: 基础特征抽取: 年份 2015, 特征行数=244

[2019-10-12 09:24:08.025246] INFO: 基础特征抽取: 年份 2016, 特征行数=244

[2019-10-12 09:24:08.054138] INFO: 基础特征抽取: 年份 2017, 特征行数=37

[2019-10-12 09:24:08.128576] INFO: 基础特征抽取: 总行数: 586

[2019-10-12 09:24:08.131698] INFO: bigquant: general_feature_extractor.v7 运行完成[0.324101s].

[2019-10-12 09:24:08.138851] INFO: bigquant: derived_feature_extractor.v3 开始运行..

[2019-10-12 09:24:08.278492] INFO: derived_feature_extractor: 提取完成 (close_0/close_1-1)*10, 0.002s

[2019-10-12 09:24:08.282731] INFO: derived_feature_extractor: 提取完成 (high_0/high_1-1)*10, 0.002s

[2019-10-12 09:24:08.285716] INFO: derived_feature_extractor: 提取完成 (low_0/low_1-1)*10, 0.001s

[2019-10-12 09:24:08.288645] INFO: derived_feature_extractor: 提取完成 (open_0/open_1-1)*10, 0.001s

[2019-10-12 09:24:08.291632] INFO: derived_feature_extractor: 提取完成 (volume_0/volume_1-1)*10, 0.002s

[2019-10-12 09:24:08.349739] INFO: derived_feature_extractor: /y_2014, 61

[2019-10-12 09:24:08.385737] INFO: derived_feature_extractor: /y_2015, 244

[2019-10-12 09:24:08.419597] INFO: derived_feature_extractor: /y_2016, 244

[2019-10-12 09:24:08.450961] INFO: derived_feature_extractor: /y_2017, 37

[2019-10-12 09:24:08.578944] INFO: bigquant: derived_feature_extractor.v3 运行完成[0.440068s].

[2019-10-12 09:24:08.586030] INFO: bigquant: join.v3 开始运行..

[2019-10-12 09:24:08.733123] INFO: join: /y_2014, 行数=0/61, 耗时=0.028867s

[2019-10-12 09:24:08.774101] INFO: join: /y_2015, 行数=244/244, 耗时=0.036187s

[2019-10-12 09:24:08.807981] INFO: join: /y_2016, 行数=244/244, 耗时=0.029041s

[2019-10-12 09:24:08.835860] INFO: join: /y_2017, 行数=37/37, 耗时=0.026162s

[2019-10-12 09:24:08.914471] INFO: join: 最终行数: 525

[2019-10-12 09:24:08.917469] INFO: bigquant: join.v3 运行完成[0.331441s].

[2019-10-12 09:24:08.923535] INFO: bigquant: dropnan.v1 开始运行..

[2019-10-12 09:24:09.003998] INFO: dropnan: /y_2014, 0/0

[2019-10-12 09:24:09.032231] INFO: dropnan: /y_2015, 244/244

[2019-10-12 09:24:09.060844] INFO: dropnan: /y_2016, 244/244

[2019-10-12 09:24:09.091875] INFO: dropnan: /y_2017, 37/37

[2019-10-12 09:24:09.175157] INFO: dropnan: 行数: 525/525

[2019-10-12 09:24:09.178648] INFO: bigquant: dropnan.v1 运行完成[0.255089s].

[2019-10-12 09:24:09.212157] INFO: bigquant: dl_convert_to_bin.v2 开始运行..

[2019-10-12 09:24:09.608608] INFO: bigquant: dl_convert_to_bin.v2 运行完成[0.396427s].

[2019-10-12 09:24:09.619469] INFO: bigquant: dl_model_train.v1 开始运行..

[2019-10-12 09:24:10.476978] INFO: dl_model_train: 准备训练，训练样本个数：525，迭代次数：10

[2019-10-12 09:24:15.522367] INFO: dl_model_train: 训练结束，耗时：5.04s

[2019-10-12 09:24:15.632669] INFO: bigquant: dl_model_train.v1 运行完成[6.013195s].

[2019-10-12 09:24:15.636824] INFO: bigquant: instruments.v2 开始运行..

[2019-10-12 09:24:15.723619] INFO: bigquant: instruments.v2 运行完成[0.086768s].

[2019-10-12 09:24:15.767058] INFO: bigquant: general_feature_extractor.v7 开始运行..

[2019-10-12 09:24:15.909290] INFO: 基础特征抽取: 年份 2016, 特征行数=22

[2019-10-12 09:24:15.936480] INFO: 基础特征抽取: 年份 2017, 特征行数=244

[2019-10-12 09:24:16.009381] INFO: 基础特征抽取: 年份 2018, 特征行数=243

[2019-10-12 09:24:16.034893] INFO: 基础特征抽取: 年份 2019, 特征行数=0

[2019-10-12 09:24:16.115291] INFO: 基础特征抽取: 总行数: 509

[2019-10-12 09:24:16.119082] INFO: bigquant: general_feature_extractor.v7 运行完成[0.352045s].

[2019-10-12 09:24:16.122262] INFO: bigquant: derived_feature_extractor.v3 开始运行..

[2019-10-12 09:24:16.275716] INFO: derived_feature_extractor: 提取完成 (close_0/close_1-1)*10, 0.001s

[2019-10-12 09:24:16.279564] INFO: derived_feature_extractor: 提取完成 (high_0/high_1-1)*10, 0.002s

[2019-10-12 09:24:16.282840] INFO: derived_feature_extractor: 提取完成 (low_0/low_1-1)*10, 0.001s

[2019-10-12 09:24:16.286369] INFO: derived_feature_extractor: 提取完成 (open_0/open_1-1)*10, 0.001s

[2019-10-12 09:24:16.290287] INFO: derived_feature_extractor: 提取完成 (volume_0/volume_1-1)*10, 0.001s

[2019-10-12 09:24:16.357164] INFO: derived_feature_extractor: /y_2016, 22

[2019-10-12 09:24:16.393812] INFO: derived_feature_extractor: /y_2017, 244

[2019-10-12 09:24:16.434004] INFO: derived_feature_extractor: /y_2018, 243

[2019-10-12 09:24:16.551389] INFO: bigquant: derived_feature_extractor.v3 运行完成[0.429059s].

[2019-10-12 09:24:16.554769] INFO: bigquant: dropnan.v1 开始运行..

[2019-10-12 09:24:16.660816] INFO: dropnan: /y_2016, 22/22

[2019-10-12 09:24:16.689288] INFO: dropnan: /y_2017, 244/244

[2019-10-12 09:24:16.717598] INFO: dropnan: /y_2018, 243/243

[2019-10-12 09:24:16.826991] INFO: dropnan: 行数: 509/509

[2019-10-12 09:24:16.830246] INFO: bigquant: dropnan.v1 运行完成[0.275474s].

[2019-10-12 09:24:16.874946] INFO: bigquant: dl_convert_to_bin.v2 开始运行..

[2019-10-12 09:24:17.299778] INFO: bigquant: dl_convert_to_bin.v2 运行完成[0.424834s].

[2019-10-12 09:24:17.311071] INFO: bigquant: dl_model_predict.v1 开始运行..

[2019-10-12 09:24:18.701740] INFO: bigquant: dl_model_predict.v1 运行完成[1.390687s].

[2019-10-12 09:24:18.707861] INFO: bigquant: cached.v3 开始运行..

[2019-10-12 09:24:18.932672] INFO: bigquant: cached.v3 运行完成[0.224799s].

[2019-10-12 09:24:18.998790] INFO: bigquant: backtest.v8 开始运行..

[2019-10-12 09:24:19.004759] INFO: bigquant: biglearning backtest:V8.2.16

[2019-10-12 09:24:19.006832] INFO: bigquant: product_type:stock by specified

[2019-10-12 09:24:19.229838] INFO: bigquant: cached.v2 开始运行..

[2019-10-12 09:24:28.216680] INFO: bigquant: 读取股票行情完成:695

[2019-10-12 09:24:28.554987] INFO: bigquant: cached.v2 运行完成[9.325144s].

[2019-10-12 09:24:28.588720] INFO: algo: TradingAlgorithm V1.5.9

[2019-10-12 09:24:28.716281] INFO: algo: trading transform...

[2019-10-12 09:24:28.999478] INFO: algo: handle_splits get splits [dt:2017-08-24 00:00:00+00:00] [asset:Equity(0 [600009.SHA]), ratio:0.9882099126332605]

[2019-10-12 09:24:29.001557] INFO: Position: position stock handle split[sid:0, orig_amount:36200, new_amount:36631.0, orig_cost:27.61057956776991, new_cost:27.29, ratio:0.9882099126332605, last_sale_price:36.87999363789555]

[2019-10-12 09:24:29.003464] INFO: Position: after split: PositionStock(asset:Equity(0 [600009.SHA]), amount:36631.0, cost_basis:27.29, last_sale_price:37.31999969482422)

[2019-10-12 09:24:29.005225] INFO: Position: returning cash: 32.94

[2019-10-12 09:24:29.552385] INFO: algo: handle_splits get splits [dt:2018-08-23 00:00:00+00:00] [asset:Equity(0 [600009.SHA]), ratio:0.9900703412737303]

[2019-10-12 09:24:29.554974] INFO: Position: position stock handle split[sid:0, orig_amount:36631.0, new_amount:36998.0, orig_cost:27.29, new_cost:27.02, ratio:0.9900703412737303, last_sale_price:57.83000848272584]

[2019-10-12 09:24:29.556595] INFO: Position: after split: PositionStock(asset:Equity(0 [600009.SHA]), amount:36998.0, cost_basis:27.02, last_sale_price:58.40999984741211)

[2019-10-12 09:24:29.557921] INFO: Position: returning cash: 22.05

[2019-10-12 09:24:29.784782] INFO: Performance: Simulated 451 trading days out of 451.

[2019-10-12 09:24:29.786911] INFO: Performance: first open: 2017-03-01 09:30:00+00:00

[2019-10-12 09:24:29.788554] INFO: Performance: last close: 2018-12-28 15:00:00+00:00

[2019-10-12 09:24:31.831398] INFO: bigquant: backtest.v8 运行完成[12.832555s].

Epoch 1/10
525/525 [==============================] - 1s 2ms/step - loss: 1.2100 - acc: 0.4781
Epoch 2/10
525/525 [==============================] - 0s 195us/step - loss: 1.3012 - acc: 0.4686
Epoch 3/10
525/525 [==============================] - 0s 216us/step - loss: 1.2714 - acc: 0.4724
Epoch 4/10
525/525 [==============================] - 0s 230us/step - loss: 1.0524 - acc: 0.5524
Epoch 5/10
525/525 [==============================] - 0s 255us/step - loss: 1.1205 - acc: 0.4971
Epoch 6/10
525/525 [==============================] - 0s 198us/step - loss: 1.0773 - acc: 0.5162
Epoch 7/10
525/525 [==============================] - 0s 210us/step - loss: 1.1179 - acc: 0.5029
Epoch 8/10
525/525 [==============================] - 0s 197us/step - loss: 1.1792 - acc: 0.4648
Epoch 9/10
525/525 [==============================] - 0s 239us/step - loss: 1.1691 - acc: 0.5048
Epoch 10/10
525/525 [==============================] - 0s 224us/step - loss: 1.1771 - acc: 0.4876

DataSource(b0388ce384c94f4490a0adf7a10aa5a0T, v3)

2017-03-01 15:00:00+00:00 买入！

收益率87.83%
年化收益率42.22%
基准收益率-12.81%
阿尔法0.46
贝塔0.87
夏普比率1.08
胜率1.0
盈亏比0.0
收益波动率35.59%
信息比率0.09
最大回撤29.6%

bigcharts-data-start/{"__id":"bigchart-62baa074616f4e63833dc6c88fc773fb","__type":"tabs"}/bigcharts-data-end

读取模型训练结果¶

In [ ]:

model = m6.load_model()
for i,layer in enumerate(model.layers):
    print(i,layer.name,layer.output_shape)

构建中间层模型¶

In [ ]:

from keras.models import Model
layer_model = Model(input=model.input, output=model.get_layer('L3').output)

读取输入input数据¶

In [ ]:

x=m27.data.read()['x']

获取中间层预测结果¶

In [ ]:

layer_result = layer_model.predict(x)

查看中间层结果¶

In [ ]:

layer_result[0:1]