Python - Quantstats量化投资策略绩效统计包 - 详解

使用Quantstats包做量化投资绩效统计的时候因为Pandas、Quantstats版本不匹配踩了一些坑；另外，Quantstats中的绩效统计指标非常全面，因此详细记录一下BUG修复方法、使用说明以及部分指标的内涵示意。

一、Quantstats安装及版本匹配问题

可以在cmd界面分别通过下面代码查询python/pandas/quantstats的版本。

python - version
pip show pandas
pip show quantstats

我使用的是截止到文章发布时点的最新版本：

Python 3.12.8
Pandas 2.2.3
Quantstats 0.0.64

上述版本组合在Quantstats生成绩效统计页面时，因为Quantstats包没及时随着Pandas包的更新，会报两个错，需要修改Quantstats包。第一个是在Quantstats目录下_plotting文件夹下的core.py文件中294-297行要去掉sum函数的传参，因为新的2.2.3版本Pandas这里没有参数。

    if resample:returns = returns.resample(resample)returns = returns.last() if compound is True else returns.sum(axis=0)if isinstance(benchmark, _pd.Series):benchmark = benchmark.resample(resample)benchmark = benchmark.last() if compound is True else benchmark.sum(axis=0)

第二个是把1015-1025行的inplace方法重写成以下形式，新版本Pandas不支持inplace。

    port["Weekly"] = port["Daily"].resample("W-MON").apply(apply_fnc)port["Weekly"] = port["Weekly"].ffill()port["Monthly"] = port["Daily"].resample("ME").apply(apply_fnc)port["Monthly"] = port["Monthly"].ffill()port["Quarterly"] = port["Daily"].resample("QE").apply(apply_fnc)port["Quarterly"] = port["Quarterly"].ffill()port["Yearly"] = port["Daily"].resample("YE").apply(apply_fnc)port["Yearly"] = port["Yearly"].ffill()

上面修订提交了GITHUBGitHub - ranaroussi/quantstats: Portfolio analytics for quants, written in Python

二、Quantstats的使用

QuantStatus由3个主要模块组成：

quantstats.stats-用于计算各种绩效指标，如夏普比率、胜率、波动率等。

quantstats.plots-用于可视化绩效、滚动统计、月度回报等。

quantstats.reports-用于生成指标报告、批量绘图和创建可另存为HTML文件。 

以持有长江电力600900为策略，以上证综指000001为基准，生成reports如下。EXCEL数据附后，没会员下不了附件的可以私我发。

import pandas as pd
import quantstats as qs#read stock data： Seris格式，index为日期，列为return
stock = pd.read_excel('600900.XSHG.xlsx',index_col=0)[['close']].pct_change().dropna().rename({'close':'return'},axis=1)['return'].rename("600900")#read benchmark data:  Seris格式，index为日期，列为return
benchmark  = pd.read_excel('000001.XSHG.xlsx',index_col=0)[['close']].pct_change().dropna().rename({'close':'return'},axis=1)['return'].rename("000001")qs.reports.html(stock,benchmark,output='report.html')

三、指标详解

Quantstats有六个模块：

其中，extend_pandas的功能是可以实现通过Dataframe对象.方法()的方式调用QuantStatsd中的方法，例如：df.sharpe()，实现方式如下：

#!/usr/bin/env python
# -*- coding: UTF-8 -*-
#
# QuantStats: Portfolio analytics for quants
# https://github.com/ranaroussi/quantstats
#
# Copyright 2019-2024 Ran Aroussi
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.from . import version__version__ = version.version
__author__ = "Ran Aroussi"from . import stats, utils, plots, reports__all__ = ["stats", "plots", "reports", "utils", "extend_pandas"]# try automatic matplotlib inline
utils._in_notebook(matplotlib_inline=True)def extend_pandas():"""Extends pandas by exposing methods to be used like:df.sharpe(), df.best('day'), ..."""from pandas.core.base import PandasObject as _po_po.compsum = stats.compsum_po.comp = stats.comp_po.expected_return = stats.expected_return_po.geometric_mean = stats.geometric_mean_po.ghpr = stats.ghpr_po.outliers = stats.outliers_po.remove_outliers = stats.remove_outliers_po.best = stats.best_po.worst = stats.worst_po.consecutive_wins = stats.consecutive_wins_po.consecutive_losses = stats.consecutive_losses_po.exposure = stats.exposure_po.win_rate = stats.win_rate_po.avg_return = stats.avg_return_po.avg_win = stats.avg_win_po.avg_loss = stats.avg_loss_po.volatility = stats.volatility_po.rolling_volatility = stats.rolling_volatility_po.implied_volatility = stats.implied_volatility_po.sharpe = stats.sharpe_po.smart_sharpe = stats.smart_sharpe_po.rolling_sharpe = stats.rolling_sharpe_po.sortino = stats.sortino_po.smart_sortino = stats.smart_sortino_po.adjusted_sortino = stats.adjusted_sortino_po.rolling_sortino = stats.rolling_sortino_po.omega = stats.omega_po.cagr = stats.cagr_po.rar = stats.rar_po.skew = stats.skew_po.kurtosis = stats.kurtosis_po.calmar = stats.calmar_po.ulcer_index = stats.ulcer_index_po.ulcer_performance_index = stats.ulcer_performance_index_po.upi = stats.upi_po.serenity_index = stats.serenity_index_po.risk_of_ruin = stats.risk_of_ruin_po.ror = stats.ror_po.value_at_risk = stats.value_at_risk_po.var = stats.var_po.conditional_value_at_risk = stats.conditional_value_at_risk_po.cvar = stats.cvar_po.expected_shortfall = stats.expected_shortfall_po.tail_ratio = stats.tail_ratio_po.payoff_ratio = stats.payoff_ratio_po.win_loss_ratio = stats.win_loss_ratio_po.profit_ratio = stats.profit_ratio_po.profit_factor = stats.profit_factor_po.gain_to_pain_ratio = stats.gain_to_pain_ratio_po.cpc_index = stats.cpc_index_po.common_sense_ratio = stats.common_sense_ratio_po.outlier_win_ratio = stats.outlier_win_ratio_po.outlier_loss_ratio = stats.outlier_loss_ratio_po.recovery_factor = stats.recovery_factor_po.risk_return_ratio = stats.risk_return_ratio_po.max_drawdown = stats.max_drawdown_po.to_drawdown_series = stats.to_drawdown_series_po.kelly_criterion = stats.kelly_criterion_po.monthly_returns = stats.monthly_returns_po.pct_rank = stats.pct_rank_po.treynor_ratio = stats.treynor_ratio_po.probabilistic_sharpe_ratio = stats.probabilistic_sharpe_ratio_po.probabilistic_sortino_ratio = stats.probabilistic_sortino_ratio_po.probabilistic_adjusted_sortino_ratio = (stats.probabilistic_adjusted_sortino_ratio)# methods from utils_po.to_returns = utils.to_returns_po.to_prices = utils.to_prices_po.to_log_returns = utils.to_log_returns_po.log_returns = utils.log_returns_po.exponential_stdev = utils.exponential_stdev_po.rebase = utils.rebase_po.aggregate_returns = utils.aggregate_returns_po.to_excess_returns = utils.to_excess_returns_po.multi_shift = utils.multi_shift_po.curr_month = utils._pandas_current_month_po.date = utils._pandas_date_po.mtd = utils._mtd_po.qtd = utils._qtd_po.ytd = utils._ytd# methods that requires benchmark stats_po.r_squared = stats.r_squared_po.r2 = stats.r2_po.information_ratio = stats.information_ratio_po.greeks = stats.greeks_po.rolling_greeks = stats.rolling_greeks_po.compare = stats.compare# plotting methods_po.plot_snapshot = plots.snapshot_po.plot_earnings = plots.earnings_po.plot_daily_returns = plots.daily_returns_po.plot_distribution = plots.distribution_po.plot_drawdown = plots.drawdown_po.plot_drawdowns_periods = plots.drawdowns_periods_po.plot_histogram = plots.histogram_po.plot_log_returns = plots.log_returns_po.plot_returns = plots.returns_po.plot_rolling_beta = plots.rolling_beta_po.plot_rolling_sharpe = plots.rolling_sharpe_po.plot_rolling_sortino = plots.rolling_sortino_po.plot_rolling_volatility = plots.rolling_volatility_po.plot_yearly_returns = plots.yearly_returns_po.plot_monthly_heatmap = plots.monthly_heatmap_po.metrics = reports.metrics# extend_pandas()

...正在更新