benchmark怎么填如果是用的us数据？

[boundles]

如上，谢谢。

[zhupr]

@boundles Hi, SP500 or NASDAQ100. You can also get more: https://github.com/microsoft/qlib/tree/main/scripts/data_collector/us_index

[boundles]

@boundles Hi, SP500 or NASDAQ100. You can also get more: https://github.com/microsoft/qlib/tree/main/scripts/data_collector/us_index

File "/Users/didi/opt/miniconda3/lib/python3.8/site-packages/pyqlib-0.7.2.99-py3.8-macosx-10.9-x86_64.egg/qlib/backtest/report.py", line 73, in init self.init_bench(freq=freq, benchmark_config=benchmark_config) File "/Users/didi/opt/miniconda3/lib/python3.8/site-packages/pyqlib-0.7.2.99-py3.8-macosx-10.9-x86_64.egg/qlib/backtest/report.py", line 91, in init_bench self.bench = self._cal_benchmark(self.benchmark_config, self.freq) File "/Users/didi/opt/miniconda3/lib/python3.8/site-packages/pyqlib-0.7.2.99-py3.8-macosx-10.9-x86_64.egg/qlib/backtest/report.py", line 112, in _cal_benchmark raise ValueError(f"The benchmark {_codes} does not exist. Please provide the right benchmark") ValueError: The benchmark ['NASDAQ-100'] does not exist. Please provide the right benchmark

改成NASDAQ-100之后报了这个错误，回测配置如下： "backtest": { "start_time": "2017-01-01", "end_time": "2020-08-01", "account": 100000000, "benchmark": "NASDAQ-100", "exchange_kwargs": { "freq": "day", "limit_threshold": 0.095, "deal_price": "close", "open_cost": 0.0005, "close_cost": 0.0015, "min_cost": 5, }, },

[zhupr]

Sorry, benchmark should be written: ^gspc or ^ndx or ^dji; the correspondence between benchmarket and market is:

• market=SP500, benchmarket=^gspc
• marekt=NASDAQ100, benchmark=^ndx

[boundles]

"open_cost": 0.0005, "close_cost": 0.0015, "min_cost": 5,

谢谢，可以跑通了，再问一下，open cost和close cost还有min cost一般是怎么设置的呢？

[boundles]

Sorry, benchmark should be written: ^gspc or ^ndx or ^dji; the correspondence between benchmarket and market is:

• market=SP500, benchmarket=^gspc
• marekt=NASDAQ100, benchmark=^ndx

`# Copyright (c) Microsoft Corporation.

Licensed under the MIT License.

import qlib from qlib.data import D from qlib.config import REG_CN, REG_US from qlib.workflow import R from qlib.workflow.record_temp import SignalRecord, PortAnaRecord from qlib.utils import init_instance_by_config from qlib.contrib.data.handler import DataHandlerLP from qlib.contrib.model import DEnsembleModel from qlib.tests.data import GetData from qlib.tests.config import CSI300_BENCH, CSI300_GBDT_TASK

NASDAQ100_MARKET = "nasdaq100"

def get_data_handler_config( start_time="2008-01-01", end_time="2020-08-01", fit_start_time="2008-01-01", fit_end_time="2014-12-31", instruments=NASDAQ100_MARKET, ): return { "start_time": start_time, "end_time": end_time, "fit_start_time": fit_start_time, "fit_end_time": fit_end_time, "instruments": instruments, }

def get_dataset_config( dataset_class="Alpha158", train=("2008-01-01", "2014-12-31"), valid=("2015-01-01", "2016-12-31"), test=("2017-01-01", "2020-08-01"), handler_kwargs={"instruments": NASDAQ100_MARKET}, ): return { "class": "DatasetH", "module_path": "qlib.data.dataset", "kwargs": { "handler": { "class": dataset_class, "module_path": "qlib.contrib.data.handler", "kwargs": get_data_handler_config(**handler_kwargs), }, "segments": { "train": train, "valid": valid, "test": test, }, }, }

def get_gbdt_dataset(dataset_kwargs={}, handler_kwargs={"instruments": NASDAQ100_MARKET}): return get_dataset_config(**dataset_kwargs, handler_kwargs=handler_kwargs)

if name == "main":

use us data

provider_uri = "~/.qlib/qlib_data/us_data"  # target_dir
# GetData().qlib_data(target_dir=provider_uri, region=REG_US, exists_skip=True)
qlib.init(provider_uri=provider_uri, region=REG_US)

# define dataset
dataset = init_instance_by_config(get_gbdt_dataset())
df_train, df_valid = dataset.prepare(
    ["train", "valid"], col_set=["feature", "label"], data_key=DataHandlerLP.DK_L
)
x_train, y_train = df_train["feature"], df_train["label"]

# define model
model = DEnsembleModel(base_model="gbm",
                       loss="mse",
                       num_models=6,
                       enable_sr=True,
                       enable_fs=True,
                       alpha1=1.0,
                       alpha2=1.0,
                       bins_sr=10,
                       bins_fs=5,
                       decay=0.5,
                       sample_ratios=[0.8, 0.7, 0.6, 0.5, 0.4],
                       sub_weights=[1, 0.2, 0.2, 0.2, 0.2, 0.2],
                       epochs=28,
                       colsample_bytree=0.8879,
                       learning_rate=0.2,
                       subsample=0.8789,
                       lambda_l1=205.6999,
                       lambda_l2=580.9768,
                       max_depth=8,
                       num_leaves=210,
                       num_threads=20)

port_analysis_config = {
    "executor": {
        "class": "SimulatorExecutor",
        "module_path": "qlib.backtest.executor",
        "kwargs": {
            "time_per_step": "day",
            "generate_portfolio_metrics": True,
        },
    },
    "strategy": {
        "class": "TopkDropoutStrategy",
        "module_path": "qlib.contrib.strategy.signal_strategy",
        "kwargs": {
            "signal": (model, dataset),
            "topk": 50,
            "n_drop": 5,
        },
    },
    "backtest": {
        "start_time": "2017-01-01",
        "end_time": "2020-08-01",
        "account": 100000000,
        "benchmark": "^ndx",
        "exchange_kwargs": {
            "freq": "day",
            "limit_threshold": 0.095,
            "deal_price": "close",
            "open_cost": 0.0005,
            "close_cost": 0.0015,
            "min_cost": 5,
        },
    },
}

# start exp
with R.start(experiment_name="workflow"):
    model.fit(dataset)
    R.save_objects(**{"params.pkl": model})

    # prediction
    recorder = R.get_recorder()
    sr = SignalRecord(model, dataset, recorder)
    sr.generate()

    # backtest. If users want to use backtest based on their own prediction,
    # please refer to https://qlib.readthedocs.io/en/latest/component/recorder.html#record-template.
    par = PortAnaRecord(recorder, port_analysis_config, "day")
    par.generate()

`

Result： 'The following are analysis results of benchmark return(1day).' risk mean 0.000987 std 0.014722 annualized_return 0.234893 information_ratio 1.034224 max_drawdown -0.297210 'The following are analysis results of the excess return without cost(1day).' risk mean -0.000064 std 0.004405 annualized_return -0.015222 information_ratio -0.224014 max_drawdown -0.121643 'The following are analysis results of the excess return with cost(1day).' risk mean -0.000262 std 0.004405 annualized_return -0.062459 information_ratio -0.919101 max_drawdown -0.263510 'The following are analysis results of indicators(1day).' value ffr 1.0 pa 0.0 pos 0.0 [89836:MainThread](2021-11-30 19:38:46,403) INFO - qlib.timer - [log.py:113] - Time cost: 0.014s | waiting async_log Done

效果好像不行，跟benchmark差别比较大，请问代码有问题吗？或者你们有在us data上的例子吗，谢谢

[zhupr]

@boundles Hi, benchmark示例中的配置是针对A股的，比如A股有涨跌停限制(limit_threshold=0.095)等；而美股没有涨跌停(limit_threshold应该设置为None)，并且model的参数也是针对A股的，如果用在美股，参数都需要重新设置，各个参数的释义在这里：https://qlib.readthedocs.io/en/latest/component/workflow.html

[boundles]

@boundles Hi, _benchmark_示例中的配置是针对A股的，比如A股有涨跌停限制(limit_threshold=0.095)等；而美股没有涨跌停(limit_threshold应该设置为None)，并且model的参数也是针对A股的，如果用在美股，参数都需要重新设置，各个参数的释义在这里：https://qlib.readthedocs.io/en/latest/component/workflow.html

好的，谢谢

[boundles]

@boundles Hi, _benchmark_示例中的配置是针对A股的，比如A股有涨跌停限制(limit_threshold=0.095)等；而美股没有涨跌停(limit_threshold应该设置为None)，并且model的参数也是针对A股的，如果用在美股，参数都需要重新设置，各个参数的释义在这里：https://qlib.readthedocs.io/en/latest/component/workflow.html

hi，你们有在美股数据上完整的示例吗？我试了几个模型参数效果都不怎么好，不确定是哪里设置错了还是模型效果本身就是这样？

[zhupr]

不好意思，美股的示例暂时还没有

[boundles]

不好意思，美股的示例暂时还没有

我调了下训练数据结果大概正常了些，再想问下下面这个benchmark结果的计算逻辑是什么？ 'The following are analysis results of benchmark return(1day).' risk mean 0.002877 std 0.014513 annualized_return 0.684689 information_ratio 3.058170 max_drawdown -0.050106

[boundles]

不好意思，美股的示例暂时还没有

我调了下训练数据结果大概正常了些，再想问下下面这个benchmark结果的计算逻辑是什么？

'The following are analysis results of benchmark return(1day).' risk mean 0.002877 std 0.014513 annualized_return 0.684689 information_ratio 3.058170 max_drawdown -0.050106

[zhupr]

benchmark的计算逻辑在这里: https://github.com/microsoft/qlib/blob/main/qlib/contrib/evaluate.py#L24

[boundles]

benchmark的计算逻辑在这里: https://github.com/microsoft/qlib/blob/main/qlib/contrib/evaluate.py#L24

好像没看到benchmark这个策略的逻辑是什么？

[zhupr]

是这个指这个策略吗：https://github.com/microsoft/qlib/blob/main/qlib/contrib/strategy/signal_strategy.py#L20

[github-actions[bot]]

This issue is stale because it has been open for three months with no activity. Remove the stale label or comment on the issue otherwise this will be closed in 5 days

[tAnGTaNgT]

NASDAQ-

麻烦问下，为什么是这个写法呢

[quant2008]

请问你怎么调的？我的美股绩效也是很差

[l0ngc]

请问你怎么调的？我的美股绩效也是很差

老哥，我美股的绩效也很差，请问您调出来了么，方便分享一下吗