ai-自动-一只权0号测试

策略分享
标签: #<Tag:0x00007fc065942410>

(me_robot) #1
克隆策略
In [2]:
# 基础参数配置
class conf:
    start_date = '2006-01-01'
    end_date='2017-07-19'
    
    # split_date 之前的数据用于训练,之后的数据用作效果评估
    split_date = '2012-01-01'
    
    # D.instruments: https://bigquant.com/docs/data_instruments.html
    instruments = D.instruments(start_date, split_date)

    # 机器学习目标标注函数
    # 如下标注函数等价于 min(max((持有期间的收益 * 100), -20), 20) + 20 (后面的M.fast_auto_labeler会做取整操作)
    # 说明:max/min这里将标注分数限定在区间[-20, 20],+20将分数变为非负数 (StockRanker要求标注分数非负整数)
    label_expr = ['return * 100', 'where(label > {0}, {0}, where(label < -{0}, -{0}, label)) + {0}'.format(22)]
    
    # 持有天数,用于计算label_expr中的return值(收益)
    hold_days = 5

    # 特征 https://bigquant.com/docs/data_features.html,你可以通过表达式构造任何特征
    features = [
        'close_5/close_0',  # 5日收益
        'close_10/close_0',  # 10日收益
        'close_20/close_0',  # 20日收益
        'close_30/close_0',  # 30日收益
        'close_60/close_0',  # 60日收益
        'close_120/close_0',  # 120日收益
        
        'avg_amount_0/avg_amount_5',  # 当日/5日平均交易额
        'avg_amount_5/avg_amount_10',  # 5日/20日平均交易额
        'avg_amount_10/avg_amount_20',  # 10日/20日平均交易额
        'avg_amount_20/avg_amount_30',  # 20日/30日平均交易额
        
        'rank_avg_amount_0/rank_avg_amount_5',  # 当日/5日平均交易额排名
        'rank_avg_amount_5/rank_avg_amount_10',  # 5日/10日平均交易额排名
        'rank_avg_amount_10/rank_avg_amount_20',  # 10日/20日平均交易额排名
        'rank_avg_amount20/rank_avg_amount_30',  # 5日/10日平均交易额排名
        
        'rank_return_0',  # 当日收益
        'rank_return_5',  # 5日收益
        'rank_return_10',  # 10日收益
        'rank_return_20',  # 20日收益
        'rank_return_30',  # 30日收益
        'rank_return_60',  # 60日收益
        'rank_return_120',  # 120日收益
        
        'rank_return_0/rank_return_5',  # 当日/5日收益排名
        'rank_return_5/rank_return_10',  # 5日/10日收益排名
        'rank_return_10/rank_return_20',  # 10日/20日收益排名
        'rank_return_20/rank_return_30',  # 20日/30日收益排名
        'rank_return_30/rank_return_60',  # 30日/60日收益排名
        'rank_return_60/rank_return_120',  # 60日/120日收益排名
        
        'pe_ttm_0 < 50',
        'st_status_0 >0',
        'list_days_0 >99',
        
        'pe_ttm_0',
        'st_status_0',
        'list_days_0 > 99',
    ]

# 给数据做标注:给每一行数据(样本)打分,一般分数越高表示越好
m1 = M.fast_auto_labeler.v7(
    instruments=conf.instruments, start_date=conf.start_date, end_date=conf.split_date,
    label_expr=conf.label_expr, hold_days=conf.hold_days,
    benchmark='000300.SHA', sell_at='open', buy_at='open')

# 计算特征数据
m2 = M.general_feature_extractor.v5(
    instruments=conf.instruments, start_date=conf.start_date, end_date=conf.split_date,
    features=conf.features)

# 数据预处理:缺失数据处理,数据规范化,T.get_stock_ranker_default_transforms为StockRanker模型做数据预处理
m3 = M.transform.v2(
    data=m2.data, transforms=T.get_stock_ranker_default_transforms(),
    drop_null=True, astype='int32', except_columns=['date', 'instrument'],
    clip_lower=0, clip_upper=200000000)

# 合并标注和特征数据
m4 = M.join.v2(data1=m1.data, data2=m3.data, on=['date', 'instrument'], sort=True)

# StockRanker机器学习训练
m5 = M.stock_ranker_train.v3(training_ds=m4.data, features=conf.features)

# 每一个节点都可以点击展开
m5.plot_model()

## 量化回测 https://bigquant.com/docs/module_trade.html
# 回测引擎:准备数据,只执行一次
def prepare(context):
    # context.start_date / end_date,回测的时候,为trader传入参数;在实盘运行的时候,由系统替换为实盘日期
    n1 = M.general_feature_extractor.v5(
        instruments=D.instruments(),
        start_date=context.start_date, end_date=context.end_date,
        model_id=context.options['model_id'])
    
    n2 = M.transform.v2(
        data=n1.data, transforms=T.get_stock_ranker_default_transforms(),
        drop_null=True, astype='int32', except_columns=['date', 'instrument'],
        clip_lower=0, clip_upper=200000000)
    
    n3 = M.stock_ranker_predict.v2(model_id=context.options['model_id'], data=n2.data)
    
    context.instruments = n3.instruments
    context.options['predictions'] = n3.predictions

# 回测引擎:初始化函数,只执行一次
def initialize(context):

    # 加载预测数据
    context.ranker_prediction = context.options['predictions'].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))
    
    # 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 = 5
    
    # 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 = 0.2

# 回测引擎:每日数据处理函数,每天执行一次
def handle_data(context, data):
    
    # 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')]

    # 1. 资金分配
    # 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
    # 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)
    is_staging = context.trading_day_index < context.options['hold_days'] # 是否在建仓期间(前 hold_days 天)
    cash_avg = context.portfolio.portfolio_value / context.options['hold_days']
    cash_for_buy = min(context.portfolio.cash, (1 if is_staging else 1.5) * cash_avg)
    cash_for_sell = cash_avg - (context.portfolio.cash - cash_for_buy)
    positions = {e.symbol: p.amount * p.last_sale_price
                 for e, p in context.perf_tracker.position_tracker.positions.items()}

    # 2. 生成卖出订单:hold_days天之后才开始卖出;对持仓的股票,按StockRanker预测的排序末位淘汰
    if not is_staging and cash_for_sell > 0:
        equities = {e.symbol: e for e, p in context.perf_tracker.position_tracker.positions.items()}
        instruments = list(reversed(list(ranker_prediction.instrument[ranker_prediction.instrument.apply(
                lambda x: x in equities and not context.has_unfinished_sell_order(equities[x]))])))

        # print('rank order for sell %s' % instruments)
        for instrument in instruments:
            context.order_target(context.symbol(instrument), 0)
            cash_for_sell -= positions[instrument]
            if cash_for_sell <= 0:
                break

    # 3. 生成买入订单:按StockRanker预测的排序,买入前面的stock_count只股票
    buy_cash_weights = context.stock_weights
    buy_instruments = list(ranker_prediction.instrument[:len(buy_cash_weights)])
    max_cash_per_instrument = context.portfolio.portfolio_value * context.max_cash_per_instrument
    for i, instrument in enumerate(buy_instruments):
        cash = cash_for_buy * 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)


# 调用交易引擎
m6 = M.trade.v1(
    instruments=None,
    start_date=conf.split_date,
    end_date=conf.end_date,
    prepare=prepare,
    initialize=initialize,
    handle_data=handle_data,
    order_price_field_buy='open',       # 表示 开盘 时买入
    order_price_field_sell='close',     # 表示 收盘 前卖出
    capital_base=100000,               # 初始资金10万
    benchmark='000300.SHA',             # 比较基准,不影响回测结果
    # 通过 options 参数传递预测数据和参数给回测引擎
    options={'hold_days': conf.hold_days, 'model_id': m5.model_id}
)

# 调用风险分析
m6.risk_analyze()      
    
[2017-07-21 17:28:48.857329] INFO: bigquant: fast_auto_labeler.v7 start ..
[2017-07-21 17:28:48.869702] INFO: bigquant: hit cache
[2017-07-21 17:28:48.885747] INFO: bigquant: fast_auto_labeler.v7 end [0.028408s].
[2017-07-21 17:28:48.902168] INFO: bigquant: general_feature_extractor.v5 start ..
[2017-07-21 17:28:48.906013] INFO: bigquant: hit cache
[2017-07-21 17:28:48.911699] INFO: bigquant: general_feature_extractor.v5 end [0.009547s].
[2017-07-21 17:28:48.943617] INFO: bigquant: transform.v2 start ..
[2017-07-21 17:28:48.948747] INFO: bigquant: hit cache
[2017-07-21 17:28:48.954018] INFO: bigquant: transform.v2 end [0.010407s].
[2017-07-21 17:28:48.979850] INFO: bigquant: join.v2 start ..
[2017-07-21 17:28:48.988797] INFO: bigquant: hit cache
[2017-07-21 17:28:48.990943] INFO: bigquant: join.v2 end [0.011116s].
[2017-07-21 17:28:49.033084] INFO: bigquant: stock_ranker_train.v3 start ..
[2017-07-21 17:28:49.044840] INFO: bigquant: hit cache
[2017-07-21 17:28:49.052518] INFO: bigquant: stock_ranker_train.v3 end [0.019436s].
[2017-07-21 17:28:49.182029] INFO: bigquant: backtest.v6 start ..
[2017-07-21 17:28:49.495158] INFO: bigquant: general_feature_extractor.v5 start ..
[2017-07-21 17:29:08.294680] INFO: general_feature_extractor: year 2012, featurerows=565675
[2017-07-21 17:29:39.686384] INFO: general_feature_extractor: year 2013, featurerows=564168
[2017-07-21 17:30:13.096025] INFO: general_feature_extractor: year 2014, featurerows=569948
[2017-07-21 17:30:41.782595] INFO: general_feature_extractor: year 2015, featurerows=569698
[2017-07-21 17:31:09.182378] INFO: general_feature_extractor: year 2016, featurerows=641546
[2017-07-21 17:31:26.686571] INFO: general_feature_extractor: year 2017, featurerows=388437
[2017-07-21 17:31:26.702310] INFO: general_feature_extractor: total feature rows: 3299472
[2017-07-21 17:31:26.710580] INFO: bigquant: general_feature_extractor.v5 end [157.21541s].
[2017-07-21 17:31:26.723589] INFO: bigquant: transform.v2 start ..
[2017-07-21 17:31:33.760937] INFO: transform: transformed /y_2012, 541917/565675
[2017-07-21 17:31:39.645461] INFO: transform: transformed /y_2013, 562660/564168
[2017-07-21 17:31:45.480195] INFO: transform: transformed /y_2014, 559457/569948
[2017-07-21 17:31:51.342031] INFO: transform: transformed /y_2015, 541527/569698
[2017-07-21 17:31:57.706345] INFO: transform: transformed /y_2016, 622280/641546
[2017-07-21 17:32:02.700660] INFO: transform: transformed /y_2017, 357763/388437
[2017-07-21 17:32:02.726165] INFO: transform: transformed rows: 3185604/3299472
[2017-07-21 17:32:02.761094] INFO: bigquant: transform.v2 end [36.037516s].
[2017-07-21 17:32:02.770371] INFO: bigquant: stock_ranker_predict.v2 start ..
[2017-07-21 17:32:06.559653] INFO: df2bin: prepare data: prediction ..
[2017-07-21 17:33:09.546235] INFO: stock_ranker_predict: prediction: 3185604 rows
[2017-07-21 17:34:31.189525] INFO: bigquant: stock_ranker_predict.v2 end [148.419163s].
[2017-07-21 17:35:36.633891] INFO: Performance: Simulated 1346 trading days out of 1346.
[2017-07-21 17:35:36.635703] INFO: Performance: first open: 2012-01-04 14:30:00+00:00
[2017-07-21 17:35:36.636863] INFO: Performance: last close: 2017-07-19 19:00:00+00:00
[注意] 有 547 笔卖出是在多天内完成的。当日卖出股票超过了当日股票交易的2.5%会出现这种情况。
  • 收益率4031.94%
  • 年化收益率100.71%
  • 基准收益率59.0%
  • 阿尔法0.92
  • 贝塔0.86
  • 夏普比率2.89
  • 收益波动率33.3%
  • 信息比率3.53
  • 最大回撤44.71%
[2017-07-21 17:35:42.609953] INFO: bigquant: backtest.v6 end [413.427921s].

(shinefuture) #2

风险暴露 的意义是什么,大小分别对应什么?


(小Q) #3

风险暴露简单理解为:该组合在哪些因子上分布如何。
比如你买小市值股票,那么你的组合从市值这个维度来看,数值就偏小。
比如你买最近趋势很强的股票,那么你的组合从动量这个维度来看,数值就很大。
一旦数值太大或者太小都表明你的股票组合在该因子上的分布有潜在的风险。
BigQuant的风险分析模块就是分析组合在各个因子上是否均衡。