파이썬으로 Maximum Drawdown (MDD) 확인하기

import numpy as np
def get_mdd(x):
    """
    MDD(Maximum Draw-Down)
    :return: (peak_upper, peak_lower, mdd rate)
    """
    arr_v = np.array(x)
    peak_lower = np.argmax(np.maximum.accumulate(arr_v) - arr_v)
    peak_upper = np.argmax(arr_v[:peak_lower])
    return peak_upper, peak_lower, (arr_v[peak_lower] - arr_v[peak_upper]) / arr_v[peak_upper]

data = [['20190219', 125000.0, 127500.0, 123500.0, 126000.0, 57757], ['20190220', 125000.0, 127000.0, 124500.0, 126000.0, 68453], ['20190221', 125500.0, 126000.0, 124000.0, 125000.0, 43961], ['20190222', 125000.0, 125000.0, 123500.0, 125000.0, 31065], ['20190225', 125500.0, 126000.0, 124000.0, 125500.0, 45852], ['20190226', 125000.0, 127000.0, 124500.0, 126500.0, 37404], ['20190227', 126500.0, 127000.0, 124500.0, 126000.0, 36131], ['20190228', 126500.0, 127000.0, 124500.0, 125000.0, 69474], ['20190304', 124500.0, 125500.0, 122500.0, 123500.0, 65517], ['20190305', 123000.0, 125000.0, 122000.0, 124500.0, 35186], ['20190306', 124000.0, 124500.0, 122500.0, 123500.0, 35449], ['20190307', 123500.0, 124000.0, 122000.0, 123500.0, 34768], ['20190308', 122500.0, 122500.0, 120500.0, 121500.0, 35118], ['20190311', 121500.0, 122500.0, 120000.0, 122500.0, 39576], ['20190312', 123000.0, 124000.0, 122000.0, 123500.0, 24117], ['20190313', 123000.0, 123500.0, 121500.0, 123500.0, 37649], ['20190314', 123000.0, 124000.0, 122000.0, 123500.0, 95132], ['20190315', 123000.0, 128000.0, 123000.0, 126500.0, 107246], ['20190318', 127000.0, 131000.0, 126500.0, 131000.0, 74644], ['20190319', 130000.0, 134000.0, 129500.0, 133000.0, 68348], ['20190320', 132000.0, 133000.0, 129500.0, 131000.0, 42697], ['20190321', 130000.0, 132000.0, 127500.0, 128500.0, 54018], ['20190322', 127500.0, 129500.0, 126500.0, 127000.0, 32380], ['20190325', 125500.0, 126500.0, 124000.0, 124500.0, 37185], ['20190326', 124500.0, 125500.0, 123500.0, 124000.0, 45161], ['20190327', 124000.0, 125000.0, 123500.0, 124000.0, 34336], ['20190328', 124000.0, 124500.0, 120500.0, 121500.0, 43518], ['20190329', 122500.0, 125000.0, 122000.0, 124500.0, 39035], ['20190401', 124000.0, 126500.0, 124000.0, 126500.0, 22463], ['20190402', 125500.0, 126500.0, 123500.0, 126000.0, 31754], ['20190403', 124500.0, 128000.0, 124500.0, 128000.0, 36250], ['20190404', 128500.0, 128500.0, 126000.0, 128000.0, 34854], ['20190405', 127500.0, 129000.0, 126000.0, 127500.0, 33513], ['20190408', 127500.0, 128000.0, 126000.0, 128000.0, 39005], ['20190409', 128000.0, 129000.0, 127500.0, 128500.0, 33266], ['20190410', 128000.0, 129000.0, 126500.0, 128000.0, 64476], ['20190411', 128500.0, 129000.0, 125000.0, 125000.0, 84802], ['20190412', 126000.0, 127500.0, 125500.0, 127000.0, 39663], ['20190415', 126500.0, 128500.0, 126000.0, 127000.0, 61140], ['20190416', 127000.0, 129000.0, 126500.0, 128500.0, 40123], ['20190417', 128500.0, 129000.0, 127000.0, 128000.0, 30846], ['20190418', 128000.0, 128500.0, 124000.0, 124500.0, 55346], ['20190419', 124500.0, 125000.0, 122000.0, 123000.0, 53439], ['20190422', 123000.0, 123500.0, 121000.0, 122500.0, 30421], ['20190423', 122500.0, 123500.0, 120500.0, 121500.0, 54997], ['20190424', 122500.0, 122500.0, 119500.0, 120000.0, 63486], ['20190425', 121000.0, 121000.0, 118500.0, 119000.0, 36046], ['20190426', 118500.0, 119500.0, 117000.0, 119000.0, 43749], ['20190429', 119000.0, 119500.0, 117000.0, 119500.0, 33516], ['20190430', 122000.0, 123000.0, 119000.0, 119500.0, 94118], ['20190502', 118500.0, 122000.0, 118000.0, 121000.0, 56723], ['20190503', 121000.0, 122000.0, 119500.0, 120000.0, 35240], ['20190507', 118500.0, 119500.0, 117000.0, 117500.0, 44453], ['20190508', 116500.0, 117000.0, 115000.0, 116500.0, 52805], ['20190509', 117000.0, 117000.0, 113000.0, 113000.0, 116012], ['20190510', 113000.0, 114500.0, 110000.0, 111500.0, 86072], ['20190513', 110500.0, 110500.0, 107500.0, 108500.0, 70847], ['20190514', 107500.0, 108000.0, 105000.0, 106500.0, 92820], ['20190515', 107000.0, 107500.0, 105000.0, 107000.0, 68937], ['20190516', 107000.0, 107500.0, 104000.0, 105000.0, 64047]]
import pandas as pd
df = pd.DataFrame(data, columns=['date', 'open', 'high', 'low', 'close', 'volume'])
mdd = get_mdd(df['close'])

(19, 59, -0.21052631578947367)

import matplotlib.pyplot as plt
import matplotlib.gridspec as gridspec
fig = plt.figure(figsize=(8, 5))
fig.set_facecolor('w')
gs = gridspec.GridSpec(2, 1, height_ratios=[3, 1])
axes = []
axes.append(plt.subplot(gs[0]))
axes.append(plt.subplot(gs[1], sharex=axes[0]))
axes[0].get_xaxis().set_visible(False)

from mpl_finance import candlestick_ohlc

x = np.arange(len(df.index))
ohlc = df[['open', 'high', 'low', 'close']].astype(int).values
dohlc = np.hstack((np.reshape(x, (-1, 1)), ohlc))

# 봉차트
candlestick_ohlc(axes[0], dohlc, width=0.5, colorup='r', colordown='b')

# 거래량 차트
axes[1].bar(x, df.volume, color='k', width=0.6, align='center')

import datetime
_xticks = []
_xlabels = []
_wd_prev = 0
for _x, d in zip(x, df.date.values):
    weekday = datetime.datetime.strptime(str(d), '%Y%m%d').weekday()
    if weekday <= _wd_prev:
        _xticks.append(_x)
        _xlabels.append(datetime.datetime.strptime(str(d), '%Y%m%d').strftime('%m/%d'))
    _wd_prev = weekday
axes[1].set_xticks(_xticks)
axes[1].set_xticklabels(_xlabels, rotation=45, minor=False)

# MDD 그리기
axes[0].plot(mdd[:2], df.loc[mdd[:2], 'close'], 'k')

plt.tight_layout()
plt.show()