请教高手们 帮我做一个简单的均线向上金叉因子的AI策略


(condom325) #1

小白属于新手,连python 都不会,刚在慢慢了解。 请教高手们 帮我做一个简单的均线向上金叉因子的AI策略做参考,布林线的也可以,先谢过了。


(小马哥) #2

如果要做AI策略,其实就是输入特征,均线向上的ai策略,可以这样构造特征:

gold_cross = '(' + '+'.join(['close_%s' % i for i in range(0, 20)]) +')'  + '/' + str(20) + '-' + 'close_0' + '> 0'
f = 'iif(%s ,1,0 )'%gold_cross
features = [ f ]

关于这个独特因子的计算,先求20日均线,然后判断是否为金叉,如果是金叉,则赋值为1,否则为0。然后将这个因子作为单独的一个feature。这里是计算的21日均线(因为包含close_0,所以是最近21个价格的均值),均值减去当日收盘价就是作为金叉死叉的判别标准,这里是做二分类,金叉为0,死叉为1 。

完整的策略如下:

克隆策略
In [9]:
# 基础参数配置
class conf:
    start_date = '2013-01-01'
    end_date='2017-01-01'
    # split_date 之前的数据用于训练,之后的数据用作效果评估
    split_date = '2016-01-01'
    # 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(20)]
    # 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
   
    gold_cross = '(' + '+'.join(['close_%s' % i for i in range(0, 20)]) +')'  + '/' + str(20) + '-' + 'close_0' + '> 0'
    f = 'iif(%s ,1,0 )'%gold_cross
    features = [ f ]

# 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.v5(
    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.v2(training_ds=m4.data, features=conf.features)


## 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=1000000,               # 初始资金
    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}
)
[2017-07-03 20:31:48.566909] INFO: bigquant: fast_auto_labeler.v5 start ..
[2017-07-03 20:31:52.535700] INFO: fast_auto_labeler: load history data: 1703814 rows
[2017-07-03 20:31:52.985048] INFO: fast_auto_labeler: start labeling
[2017-07-03 20:32:38.839486] INFO: bigquant: fast_auto_labeler.v5 end [50.27259s].
[2017-07-03 20:32:38.849935] INFO: bigquant: general_feature_extractor.v5 start ..
[2017-07-03 20:32:58.171369] INFO: general_feature_extractor: year 2013, featurerows=564168
[2017-07-03 20:33:14.771815] INFO: general_feature_extractor: year 2014, featurerows=569948
[2017-07-03 20:33:35.547820] INFO: general_feature_extractor: year 2015, featurerows=569698
[2017-07-03 20:33:53.735242] INFO: general_feature_extractor: year 2016, featurerows=0
[2017-07-03 20:33:53.757584] INFO: general_feature_extractor: total feature rows: 1703814
[2017-07-03 20:33:53.759902] INFO: bigquant: general_feature_extractor.v5 end [74.909897s].
[2017-07-03 20:33:53.767652] INFO: bigquant: transform.v2 start ..
[2017-07-03 20:34:06.347447] INFO: transform: transformed /y_2013, 564140/564168
[2017-07-03 20:34:18.166918] INFO: transform: transformed /y_2014, 567742/569948
[2017-07-03 20:34:26.391976] INFO: transform: transformed /y_2015, 565564/569698
[2017-07-03 20:34:26.414201] INFO: transform: transformed rows: 1697446/1703814
[2017-07-03 20:34:26.441906] INFO: bigquant: transform.v2 end [32.674228s].
[2017-07-03 20:34:26.448060] INFO: bigquant: join.v2 start ..
[2017-07-03 20:34:38.319883] INFO: join: /y_2013, rows=563105/564140, timetaken=10.343971s
[2017-07-03 20:34:51.272002] INFO: join: /y_2014, rows=566243/567742, timetaken=12.914283s
[2017-07-03 20:35:07.947842] INFO: join: /y_2015, rows=542072/565564, timetaken=16.638322s
[2017-07-03 20:35:08.039478] INFO: join: total result rows: 1671420
[2017-07-03 20:35:08.041672] INFO: bigquant: join.v2 end [41.59361s].
[2017-07-03 20:35:08.112976] INFO: bigquant: stock_ranker_train.v2 start ..
[2017-07-03 20:35:20.861993] INFO: df2bin: prepare data: training ..
[2017-07-03 20:35:48.454652] INFO: stock_ranker_train: training: 1671420 rows
[2017-07-03 20:37:03.674103] INFO: bigquant: stock_ranker_train.v2 end [115.561044s].
[2017-07-03 20:37:03.688011] INFO: bigquant: backtest.v6 start ..
[2017-07-03 20:37:05.357535] INFO: bigquant: general_feature_extractor.v5 start ..
[2017-07-03 20:37:30.942563] INFO: general_feature_extractor: year 2016, featurerows=641546
[2017-07-03 20:37:34.974784] INFO: general_feature_extractor: year 2017, featurerows=0
[2017-07-03 20:37:34.985353] INFO: general_feature_extractor: total feature rows: 641546
[2017-07-03 20:37:34.987764] INFO: bigquant: general_feature_extractor.v5 end [29.630265s].
[2017-07-03 20:37:34.992672] INFO: bigquant: transform.v2 start ..
[2017-07-03 20:37:40.549277] INFO: transform: transformed /y_2016, 637338/641546
[2017-07-03 20:37:40.559550] INFO: transform: transformed rows: 637338/641546
[2017-07-03 20:37:40.588726] INFO: bigquant: transform.v2 end [5.596016s].
[2017-07-03 20:37:40.592581] INFO: bigquant: stock_ranker_predict.v2 start ..
[2017-07-03 20:37:40.835888] INFO: df2bin: prepare data: prediction ..
[2017-07-03 20:37:52.156898] INFO: stock_ranker_predict: prediction: 637338 rows
[2017-07-03 20:38:09.495217] INFO: bigquant: stock_ranker_predict.v2 end [28.902597s].
[2017-07-03 20:38:35.891008] INFO: Performance: Simulated 244 trading days out of 244.
[2017-07-03 20:38:35.892152] INFO: Performance: first open: 2016-01-04 14:30:00+00:00
[2017-07-03 20:38:35.893245] INFO: Performance: last close: 2016-12-30 20:00:00+00:00
  • 收益率-4.76%
  • 年化收益率-4.91%
  • 基准收益率-11.28%
  • 阿尔法0.06
  • 贝塔0.96
  • 夏普比率-0.29
  • 收益波动率28.22%
  • 信息比率0.36
  • 最大回撤23.6%
[2017-07-03 20:38:37.311372] INFO: bigquant: backtest.v6 end [93.623355s].

(condom325) #3

非常感谢小马哥提供的例子做参考。 我得好好研究下了。:blush: