轮动交易

轮动交易涉及购买表现最好的资产，同时出售表现不佳的资产。PyBroker 可用于回测此类策略。

[1]:

import pybroker
from pybroker import ExecContext, Strategy, YFinance

我们的策略将涉及对价格涨幅（ROC）最高的股票进行排名和购买。首先，我们将使用 TA-Lib 定义一个 20 天的 ROC 指标：

[2]:

import talib as ta

roc_20 = pybroker.indicator(
    "roc_20", lambda data: ta.ROC(data.adj_close, timeperiod=20)
)

接下来，让我们定义策略的规则：

购买 20 天涨幅（ROC）最高的两只股票。
将我们的资本的 50% 分配给每只股票。
如果其中一只股票不再位于前五名的 20 天涨幅（ROC）中，则我们将清盘该股票。
每天交易这些规则。

为了实现该策略，我们编写一个 rotate 函数，将每只股票的 long_score 设置为其 20 天涨幅（ROC）。然后 PyBroker 会根据 long_score 对这些股票进行降序排名。

[3]:

def rotate(ctx: ExecContext):
    ctx.long_score = ctx.indicator("roc_20")[-1]

既然我们已经有了一种根据 ROC 对股票评分的方法，就可以使用 enable_rotation 方法来启用轮动交易。

我们通过 set_max_long_positions 将多头仓位数量上限设置为 2。将 worst_rank_held 设置为 5 会清盘任何跌出前五名 20 天涨幅（ROC）排名之外的当前持仓股票。否则，PyBroker 会买入排名前两位的股票，将 50% 的资金分配给每只股票。这次回测将使用由 10 只股票组成的股票池：

[4]:

strategy = Strategy(YFinance(), start_date="1/1/2018", end_date="1/1/2023")
strategy.set_max_long_positions(2)
strategy.enable_rotation(worst_rank_held=5)
strategy.add_execution(
    rotate,
    [
        "TSLA",
        "NFLX",
        "AAPL",
        "NVDA",
        "AMZN",
        "MSFT",
        "GOOG",
        "AMD",
        "INTC",
        "META",
    ],
    indicators=roc_20,
)
result = strategy.backtest(warmup=20)

Backtesting: 2018-01-01 00:00:00 to 2023-01-01 00:00:00

Loading bar data...

[*********************100%***********************]  10 of 10 completed

Loaded bar data: 0:00:01

Computing indicators...

100% (10 of 10) |########################| Elapsed Time: 0:00:00 Time:  0:00:00


Test split: 2018-01-02 00:00:00 to 2022-12-30 00:00:00

100% (1259 of 1259) |####################| Elapsed Time: 0:00:00 Time:  0:00:00


Finished backtest: 0:00:01

[5]:

result.orders

[5]:

	type	symbol	date	created	order_type	intent	shares	limit_price	market_price	fill_price	fees
id
1	buy	NFLX	2018-02-01	2018-01-31	market	buy_to_open	1849	NaN	26.77	26.77	0.0
2	buy	AMD	2018-02-01	2018-01-31	market	buy_to_open	3639	NaN	13.53	13.53	0.0
3	sell	AMD	2018-02-05	2018-02-02	market	sell_to_close	3639	NaN	11.56	11.56	0.0
4	buy	AMZN	2018-02-05	2018-02-02	market	buy_to_open	623	NaN	69.49	69.49	0.0
5	sell	AMZN	2018-04-03	2018-04-02	market	sell_to_close	623	NaN	69.23	69.23	0.0
...	...	...	...	...	...	...	...	...	...	...	...
256	buy	AMD	2022-11-21	2022-11-18	market	buy_to_open	3591	NaN	72.28	72.28	0.0
257	sell	AMD	2022-12-14	2022-12-13	market	sell_to_close	3591	NaN	70.16	70.16	0.0
258	buy	NFLX	2022-12-14	2022-12-13	market	buy_to_open	8822	NaN	31.96	31.96	0.0
259	sell	NVDA	2022-12-28	2022-12-27	market	sell_to_close	14918	NaN	14.07	14.07	0.0
260	buy	META	2022-12-28	2022-12-27	market	buy_to_open	1868	NaN	116.83	116.83	0.0

260 rows × 11 columns

自定义仓位配置

默认情况下，PyBroker 会将我们的资金平均分配到各个仓位。若要自定义此行为，我们可以向 enable_rotation 传入一个 sizer 函数。在轮动决定要买入哪些股票之后，会使用 RotationContext 调用 sizer，使我们能够覆盖每笔入场的仓位规模。long_ranks 属性包含每只股票的排名，其中 1 表示排名最高。

让我们重用相同的策略，但这次将 70% 的资金分配给排名第一的股票，30% 分配给排名第二的股票：

[6]:

from pybroker import RotationContext


def size_by_rank(rotation: RotationContext):
    for symbol, ctx in rotation.ctxs.items():
        if ctx.buy_shares is not None:
            rank = rotation.long_ranks[symbol]
            match rank:
                case 1:
                    ctx.buy_shares = ctx.calc_target_shares(0.7)
                case 2:
                    ctx.buy_shares = ctx.calc_target_shares(0.3)


strategy.enable_rotation(worst_rank_held=5, sizer=size_by_rank)
result = strategy.backtest(warmup=20)
result.orders

Backtesting: 2018-01-01 00:00:00 to 2023-01-01 00:00:00

Loading bar data...

[*********************100%***********************]  10 of 10 completed

Loaded bar data: 0:00:00

Computing indicators...

100% (10 of 10) |########################| Elapsed Time: 0:00:00 Time:  0:00:00


Test split: 2018-01-02 00:00:00 to 2022-12-30 00:00:00

100% (1259 of 1259) |####################| Elapsed Time: 0:00:00 Time:  0:00:00


Finished backtest: 0:00:00

[6]:

	type	symbol	date	created	order_type	intent	shares	limit_price	market_price	fill_price	fees
id
1	buy	NFLX	2018-02-01	2018-01-31	market	buy_to_open	2589	NaN	26.77	26.77	0.0
2	buy	AMD	2018-02-01	2018-01-31	market	buy_to_open	2183	NaN	13.53	13.53	0.0
3	sell	AMD	2018-02-05	2018-02-02	market	sell_to_close	2183	NaN	11.56	11.56	0.0
4	buy	AMZN	2018-02-05	2018-02-02	market	buy_to_open	379	NaN	69.49	69.49	0.0
5	sell	AMZN	2018-04-03	2018-04-02	market	sell_to_close	379	NaN	69.23	69.23	0.0
...	...	...	...	...	...	...	...	...	...	...	...
256	buy	AMD	2022-11-21	2022-11-18	market	buy_to_open	4701	NaN	72.28	72.28	0.0
257	sell	AMD	2022-12-14	2022-12-13	market	sell_to_close	4701	NaN	70.16	70.16	0.0
258	buy	NFLX	2022-12-14	2022-12-13	market	buy_to_open	4790	NaN	31.96	31.96	0.0
259	sell	NVDA	2022-12-28	2022-12-27	market	sell_to_close	7975	NaN	14.07	14.07	0.0
260	buy	META	2022-12-28	2022-12-27	market	buy_to_open	2731	NaN	116.83	116.83	0.0

260 rows × 11 columns