백테스트 - SMA 1시간봉

data = df.copy()
data = data[['close']]
data['SMA1'] = data['close'].rolling(42).mean()
data['SMA2'] = data['close'].rolling(180).mean()
data.plot(title="2years BTC", figsize=(10,4))​

import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

import pyupbit
df = pyupbit.get_ohlcv("KRW-BTC","hour1", count=24*30*24)

data['position'] = np.where(data['SMA1']>data['SMA2'],1,0)
data.dropna(inplace=True)
data['position'].plot(ylim=[-1.1, 1.1], title='market positioning', figsize=(10,4))
;

# 로그수익률
data['strategy'] = data['position'].shift(1)*data['returns']
data[['returns','strategy']].sum()

data[['returns','strategy']].sum().apply(np.exp)

data[['returns','strategy']].cumsum().apply(np.exp).plot(figsize=(10,4));

 

매도조건 변경 : SMA1 > SMA2 & CLOSE < SMA1

data = df.copy()
data = data[['close']]
data['SMA1'] = data['close'].rolling(42).mean()
data['SMA2'] = data['close'].rolling(180).mean()
data.dropna(inplace=True)
#cond = data['SMA1']>data['SMA2']
#data.loc[cond, 'position'] = 1
data['position'] = np.where(data['SMA1']>data['SMA2'],1,np.NaN)
data['position'].plot(ylim=[-0.1, 1.1], title='market positioning', figsize=(10,4))
;

cond = ( data['position']==1 ) & ( data['close'] < data['SMA1'])
data.loc[cond,'position'] = 0
data['position'].plot(ylim=[-0.1, 1.1], title='market positioning', figsize=(10,4))
;

data['position'] = data['position'].fillna(method='ffill')
data['position'].plot(ylim=[-0.1, 1.1], title='market positioning', figsize=(10,4))
;

data['returns'] = np.log(data['close']/data['close'].shift(1))
data['strategy'] = data['position'].shift(1)*data['returns']
data[['returns','strategy']].sum()

data[['returns','strategy']].sum().apply(np.exp)

data[['returns','strategy']].cumsum().apply(np.exp).plot(figsize=(10,4));

▶ 매도조건을 바꾸니 조금 더 수익률이 MDD가 낮아지는 것 같다. but....

data['cumret'] = data['strategy'].cumsum().apply(np.exp)
data['cummax'] = data['cumret'].cummax()
data[['cumret','cummax']].dropna().plot(figsize=(10,4));

 

# strategy의 MDD 
drawdown = data['cummax'] - data['cumret']
drawdown.max()

 

data['cumret_returns'] = data['returns'].cumsum().apply(np.exp)
data['cummax_returns'] = data['cumret_returns'].cummax()
data[['cumret_returns','cummax_returns']].dropna().plot(figsize=(10,4));

# 보유 시 MDD 
drawdown = data['cummax_returns'] - data['cumret_returns']
drawdown.max()

보유시 MDD 217% 보다 전략 MDD 109%가 작게 느껴지지만...........