week_evaluation.py 9.36 KB
Newer Older
李宗熹's avatar
李宗熹 committed
1 2 3 4 5 6 7 8 9 10 11 12 13
# coding: utf-8

"""
计算各个指标的方法

"""
import pandas as pd
import numpy as np
import datetime
import calendar
import math


14 15 16 17 18 19 20
dict_substrategy = {1010: '主观多头', 1020: '股票多空', 1030: '量化多头', 2010: '宏观策略', 3010: '主观趋势', 3020: '主观套利',
                            3030: '量化趋势', 3040: '量化套利', 3050: 'CTA策略', 4010: '并购重组', 4020: '定向增发', 4030: '大宗交易',
                            4040: '事件驱动复合', 5010: '市场中性', 5020: '套利策略', 5030: '相对价值复合', 6010: '纯债策略', 6020: '强债策略',
                            6030: '债券策略', 7010: 'MOM', 7020: 'FOF', 8010: '主观多策略', 8020: '量化多策略', -1: '其他策略'}
BANK_RATE = 0.015


李宗熹's avatar
李宗熹 committed
21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51
def simple_return(net_worth):
    """
    简单收益率
    net_worth:净值或指数数据
    """
    d = net_worth / net_worth.shift(1) - 1
    d.iloc[0] = 0
    return d


def excess_return(returns, bank_rate, n):
    """
    超额收益率
    returns:简单收益率
    bank_rate: 银行收益率, 是已经除过的无风险收益。也可以是其他的基准收益
    n: 数据类型, 周(52), 月(12), 日(250)
    """
    d = returns.mean() - bank_rate / n
    # print(pd.Series(d*np.ones(len(returns))))
    return d
    # pd.Series(d*np.ones(len(returns)))


def sharpe_ratio(excess_return, simple_return, n):
    """
    夏普比率
    excess_return: 超额收益率
    simple_return: 简单收益率
    n: 数据类型, 周(52), 月(12), 日(250)
    """
    import math
52
    d = math.sqrt(n) * excess_return / simple_return.std(ddof=1)
李宗熹's avatar
李宗熹 committed
53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266
    return d


def volatility(simple_return, n):
    """
    波动率
    :param simple_return:
    :param n:数据类型, 周(52), 月(12), 日(250)
    :return:
    """
    d = math.sqrt(n) * simple_return.std(ddof=1)
    return d


def IR(excess_return, n):
    """
    excess_return: 收益减去基准收益率
    """
    d = math.sqrt(n) * excess_return.mean() / excess_return.std(ddof=1)
    return d


def max_drawdown(return_list):
    """
    最大回撤
    return_list:净值或指数数据的列表
    返回最大回撤值,以及开始位置,和结束位置值
    """
    i = np.argmax((np.maximum.accumulate(return_list) - return_list) / np.maximum.accumulate(return_list))  # 结束位置
    if i == 0:
        return 0, 0, 0  # 没有回撤
    j = np.argmax(return_list[:i])  # 开始位置
    return (return_list[j] - return_list[i]) / (return_list[j]), j, i


def month_differ(x, y):
    """
    计算月份相差
    只根据month,year计算相差月份, 没有考虑day
    :param x: datetime.datetime
    :param y:
    :return:
    """
    m_d = abs((x.year - y.year) * 12 + (x.month - y.month) * 1)
    return m_d


def downside_risk(r, bank_rate, n):
    """
    下行风险
    r: 简单收益率
    """
    _r = r.map(lambda x: x / 100)
    # mean = _r.mean()
    r_adjust = -r.map(lambda x: min(x - bank_rate / n, 0))
    risk = np.sqrt((r_adjust ** 2).mean() * len(r_adjust) / (len(r_adjust) - 1))
    return risk


def sortino_ratio(excess_return, downside, n):
    """
    索提诺比率
    df: 净值或指数数据
    """
    import math
    sortino_ratio = math.sqrt(n) * excess_return.mean() / downside
    return sortino_ratio


def month_minus(date, n):
    """
    计算对标的前几个月份,如2020,3的前三个月是2019.12
    输入datetime格式
    注意:二月份没有30,31号的,而且3月31号,的前几个月有的是没有31号的。
    :return:
    """
    # day = date.day
    if date.month > n:
        month = date.month - n
        year = date.year
    else:
        month = date.month + 12 - n
        year = date.year - 1
    # print('month////',month)
    try:
        pre_date = datetime.datetime(year, month, date.day)
    except:
        pre_date = datetime.datetime(year, month, calendar.monthrange(year, month)[1])
    return pre_date


def is_exsits(a, b):
    """
    判断日期是否存在, 将日期与基金最开始的时间对比, 如果存在,返回日期, 不存在,返回None
    :param a: 基金初始时间
    :param b: 需要计算收益的起始时间
    :return:
    """
    if a < b:
        return True
    else:
        return False


def year_minus(date, n):
    """
    计算对标的前几个年份,如2020.3的前1年是2019.3
    输入datetime格式
    :return:
    """
    day = date.day
    month = date.month
    year = date.year - n
    pre_date = datetime.datetime(year, month, day)
    return pre_date


def range_return(end_df, begin_df):
    """
    区间收益
    """
    d = end_df / begin_df - 1
    return d


def annual_return(range_return, df, n):
    """
    年化收益
    """
    d = (1 + range_return) ** (n / len(df)) - 1
    return d


def gain_loss_ratio(simple_return):
    """
    盈亏比
    """
    pos = simple_return[simple_return >= 0].sum()
    neg = simple_return[simple_return < 0].sum()
    d = - pos / neg
    return d


def alpha_beta(simple_return, b_simple_return, n):
    """
    alpha, beta
    """
    df = pd.DataFrame()
    from sklearn.linear_model import LinearRegression
    linreg = LinearRegression()
    l = len(simple_return)
    df['returns'] = simple_return
    df['b_returns'] = b_simple_return
    X = np.array(df[['b_returns']][:l - 1])
    y = np.array(df[['returns']][:l - 1])
    linreg.fit(X, y)

    beta = linreg.coef_[0][0]
    alpha = linreg.intercept_[0] * n
    return alpha, beta


def win_rate(simple_return, b_simple_return):
    """
    胜率
    """
    df = pd.DataFrame()
    df['diff'] = simple_return - b_simple_return
    d = df[df['diff'] >= 0]['diff'].count() / df['diff'].count()
    return d


def lpm(returns, threshold, order):
    """
    下偏距, 一阶和二阶
    order: 是一阶和二阶的设定
    threshold: 是期望收益率
    """
    # This method returns a lower partial moment of the returns
    # Create an array he same length as returns containing the minimum return threshold
    threshold_array = np.empty(len(returns))
    threshold_array.fill(threshold)
    # Calculate the difference between the threshold and the returns
    diff = threshold_array - returns
    # Set the minimum of each to 0
    diff = diff.clip(min=0)
    # Return the sum of the different to the power of order
    return np.sum(diff ** order) / len(returns)


def var(returns, alpha):
    """
    计算var值,历史收益率方法, 将历史收益率由小到大排序,去置信区间的分位点, alpha是置信区间
    """
    # This method calculates the historical simulation var of the returns
    sorted_returns = np.sort(returns)
    # Calculate the index associated with alpha
    index = int(alpha * len(sorted_returns))
    # VaR should be positive
    return abs(sorted_returns[index])


def cvar(returns, alpha):
    # This method calculates the condition VaR of the returns
    sorted_returns = np.sort(returns)
    # Calculate the index associated with alpha
    index = int(alpha * len(sorted_returns))
    # Calculate the total VaR beyond alpha
    sum_var = sorted_returns[0]
    for i in range(1, index):
        sum_var += sorted_returns[i]
    # Return the average VaR
    # CVaR should be positive
    return abs(sum_var / index)
267 268 269 270 271 272 273 274 275 276 277 278 279 280 281


def freq_days(fav_freq):
    if fav_freq == 1:
        N = 250  # 日更新
    elif fav_freq == 2:
        N = 52  # 周更新
    elif fav_freq == 3:
        N = 24  # 半周更新
    elif fav_freq == 4:
        N = 12  # 月更新
    elif fav_freq == 5:
        N = 3  # 季度更新
    else:
        N = 250  # 没有设置freq默认是天更
282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321
    return N


def resample(df, trading_cal, freq):
    """对基金净值表进行粒度不同的重采样,并剔除不在交易日中的结果

    Args:
        df ([DataFrame]): [原始基金净值表]
        trading_cal ([type]): [上交所交易日表]
        freq ([type]): [重采样频率: 1:工作日,2:周, 3:月, 4:半月, 5:季度]

    Returns:
        [DataFrame]: [重采样后剔除不在交易日历中的净值表和交易日历以净值日期为索引的合表]
    """
    freq_dict = {1: 'B', 2: 'W-FRI', 3: 'M', 4: 'SM', 5: 'Q'}
    resample_freq = freq_dict[freq]
    # 按采样频率进行重采样并进行净值的前向填充
    df = df.resample(rule=resample_freq).ffill()

    # 根据采样频率确定最大日期偏移量(保证偏移后的日期与重采样的日期在同一周,同一月,同一季度等)
    timeoffset_dict = {1: 1, 2: 5, 3: 30, 4: 15, 5: 120}
    timeoffetmax = timeoffset_dict[freq]

    # Dataframe不允许直接修改index,新建一份index的复制并转为list
    new_index = list(df.index)
    # 遍历重采样后的日期
    for idx, date in enumerate(df.index):
        # 如果重采样后的日期不在交易日历中
        if date not in trading_cal.index:
            # 对重采样后的日期进行偏移
            for time_offset in range(1, timeoffetmax):
                # 如果偏移后的日期在交易日历中,保留偏移后的日期
                if date - datetime.timedelta(days=time_offset) in trading_cal.index:
                    new_index[idx] = date - datetime.timedelta(days=time_offset)
                    # 任意一天满足立即退出循环
                    break

    # 更改净值表的日期索引为重采样后且在交易日内的日期
    df.index = pd.Series(new_index)
    return df