def get_priceMom(rebalDate, codes):
past = ut.get_recentBday(rebalDate - datetime.timedelta(365))
recent = ut.get_recentBday(rebalDate - datetime.timedelta(30))
rebalDate = ut.get_recentBday(rebalDate)
prices = ut.get_stock_price(codes, past, rebalDate )
mom_12M = (prices.loc[rebalDate, :] / prices.loc[past, :]) - 1
mom_1M = (prices.loc[rebalDate, :] / prices.loc[recent, :]) - 1
momData = (mom_12M - mom_1M).dropna()
return momData
def get_priceMom(codes, rebalDate):
yearAgo = rebalDate - datetime.timedelta(days=365)
monthAgo = rebalDate - datetime.timedelta(days=30)
yearAgo = util.get_recentBday(yearAgo)
monthAgo = util.get_recentBday(monthAgo)
rebalDate = util.get_recentBday(rebalDate)
prices = util.get_stock_price(codes, yearAgo, rebalDate)
mom_12M = (prices.loc[rebalDate, :] / prices.loc[yearAgo, :]) - 1
mom_1M = (prices.loc[rebalDate, :] / prices.loc[monthAgo, :]) - 1
momData = (mom_12M - mom_1M).dropna()
return momData
def get_portfolio(factor, factorStyle = 'ratio'):
rebalData_long = []
rebalData_short = []
for i in range(len(rebal_sche)):
rebalDate_ = util.get_recentBday(rebal_sche[i], dateFormat = 'datetime')
#print(rebalDate_)
univ_ = getUniverse(marketInfo, mktcap, risk_1, risk_2, rebalDate_)
print(len(univ_))
if factorStyle == 'ratio':
factor = factor.loc[:,univ_]
factor_ = getFactorData(factor, rebalDate_)
mktcap_ = util.get_mktcap(factor_.index.values, rebalDate_, rebalDate_)
ratio_ = get_priceRatio(mktcap_, factor_)
long, short = get_longshort(ratio_, num_group = 10)
longFinal = to_backtestFormat(long, rebalDate_)
shortFinal = to_backtestFormat(short, rebalDate_)
rebalData_long.append(longFinal)
rebalData_short.append(shortFinal)
elif factorStyle == 'Momentum':
factor_mom = get_priceMom(univ_, rebalDate_)
long, short = get_longshort(factor_mom, num_group = 10, asc = False)
longFinal = to_backtestFormat(long, rebalDate_)
shortFinal = to_backtestFormat(short, rebalDate_)
rebalData_long.append(longFinal)
rebalData_short.append(shortFinal)
rebalData_long = pd.concat(rebalData_long).reset_index()[['date','code','weight']]
rebalData_short = pd.concat(rebalData_short).reset_index()[['date','code','weight']]
return rebalData_long, rebalData_short
def get_backtest_history(dollar_inv, rebalData, roundup=False, tradeCost=0.01):
dollar_ongoing = dollar_inv
basket_history = {}
rebal_date = list(set(rebalData.date))
rebal_date.sort()
tradeCost_history = {}
turnover_history = {}
weight_history = {}
inv_money_list = [dollar_inv]
inv_money_date = [rebal_date[0]]
for i in tqdm(range(len(rebal_date))):
if i < len(rebal_date) - 1:
rebalDate = rebal_date[i] # 리밸런싱 시점
stock_list_i = rebalData[rebalData.date ==
rebal_date[i]].code # 기간별 종목 리스트
w_ = rebalData[rebalData.date == rebal_date[i]].weight
weightData = pd.concat([stock_list_i, w_],
axis=1).set_index('code')
if i == 0:
weightData['money'] = weightData[
'weight'] * dollar_inv # 1기인 경우 초기 투자금으로 시작
else:
weightData['money'] = weightData[
'weight'] * dollar_ongoing # 1기 이후는 투자금의 매 기간별 마지막 시점에서의 금액 재투자
stock_price_ = util.get_stock_price(
stock_list_i, util.get_recentBday(rebalDate),
util.get_recentBday(
rebalDate)).transpose() # 해당 일자의 바스켓의 주가 추출
weightData = pd.merge(
weightData,
stock_price_,
how='inner',
left_on=weightData.index,
right_on=stock_price_.index).set_index('key_0')
weightData.columns = ['weight', 'money', 'price']
weightData['n_stocks'] = weightData['money'] / weightData['price']
if roundup == True: # 초기 투입자금으로 투자가능한 종목별 주식수 (정수로 내림)
weightData.n_stocks = np.floor(weightData.n_stocks)
else:
weightData.n_stocks = weightData.n_stocks
basket_price = util.get_basket_history(
weightData.index.values, weightData.n_stocks.values,
util.get_recentBday(rebal_date[i]),
util.get_recentBday(rebal_date[i + 1]))
dollar_ongoing = basket_price.iloc[-1, :].values[
0] # 투자금의 매 기간별 마지막 시점에서의 포트의 가치(금액)
# 투자기간 마지막 시점 (다음리밸) 에서의 가격
last_price = util.get_stock_price(
weightData.index.values,
util.get_recentBday(rebal_date[i + 1]),
util.get_recentBday(rebal_date[i + 1])).transpose()
weightData = pd.merge(weightData,
last_price,
how='inner',
left_on=weightData.index,
right_on=last_price.index).set_index('key_0')
weightData.columns = [
'weight', 'money', 'price', 'n_stocks', 'last_price'
]
weightData['last_weight'] = (
weightData.last_price * weightData.n_stocks) / (
weightData.last_price *
weightData.n_stocks).sum() # 투자기간의 마지막 시점에서의 비중
new_weight = rebalData[rebalData.date == rebal_date[
i + 1]].set_index('code')['weight'] # 새로운 비중
new_weight = pd.Series(new_weight, name='new_weight')
weightData = pd.concat([weightData, new_weight], axis=1)
weightData['new_weight'] = weightData['new_weight'].fillna(0)
weightData['weight'] = weightData['weight'].fillna(0)
tradingCost = np.abs(weightData.new_weight - weightData.last_weight
).sum() * tradeCost * dollar_ongoing
dollar_ongoing -= tradingCost
basket_history[rebalDate] = basket_price
inv_money_list += list(basket_price.values.squeeze()[1:-1])
inv_money_list += [dollar_ongoing]
inv_money_date += list(basket_price.index[1:])
tradeCost_history[''.join([
x for x in str(rebal_date[i + 1])[:10] if x != '-'
])] = tradingCost
turnover_history[''.join([
x for x in str(rebal_date[i + 1])[:10] if x != '-'
])] = tradingCost / dollar_ongoing
weight_history[''.join([
x for x in str(rebal_date[i + 1])[:10] if x != '-'
])] = weightData
elif i == len(rebal_date) - 1:
rebalDate = rebal_date[i] # 리밸런싱 시점
#print(rebalDate)
stock_list_i = rebalData[rebalData.date ==
rebal_date[i]].code # 기간별 종목 리스트
lastDay = datetime.datetime.today() - datetime.timedelta(2)
w_ = rebalData[rebalData.date == rebal_date[i]].weight
weightData = pd.concat([stock_list_i, w_],
axis=1).set_index('code')
if i == 0:
weightData['money'] = weightData[
'weight'] * dollar_inv # 1기인 경우 초기 투자금으로 시작
else:
weightData['money'] = weightData[
'weight'] * dollar_ongoing # 1기 이후는 투자금의 매 기간별 마지막 시점에서의 금액 재투자
stock_price_ = util.get_stock_price(
stock_list_i, util.get_recentBday(rebalDate),
util.get_recentBday(
rebalDate)).transpose() # 해당 일자의 바스켓의 주가 추출
weightData = pd.merge(
weightData,
stock_price_,
how='inner',
left_on=weightData.index,
right_on=stock_price_.index).set_index('key_0')
weightData.columns = ['weight', 'money', 'price']
weightData['n_stocks'] = weightData['money'] / weightData['price']
if roundup == True: # 초기 투입자금으로 투자가능한 종목별 주식수 (정수로 내림)
weightData.n_stocks = np.floor(weightData.n_stocks)
else:
weightData.n_stocks = weightData.n_stocks
basket_price = util.get_basket_history(
weightData.index.values, weightData.n_stocks.values,
util.get_recentBday(rebal_date[i]),
util.get_recentBday(lastDay))
dollar_ongoing = basket_price.iloc[-1, :].values[
0] # 투자금의 매 기간별 마지막 시점에서의 포트의 가치(금액)
#print(inv_money_list)
# 투자기간 마지막 시점 (다음리밸) 에서의 가격
last_price = util.get_stock_price(
weightData.index.values, util.get_recentBday(lastDay),
util.get_recentBday(lastDay)).transpose()
weightData = pd.merge(weightData,
last_price,
how='inner',
left_on=weightData.index,
right_on=last_price.index).set_index('key_0')
weightData.columns = [
'weight', 'money', 'price', 'n_stocks', 'last_price'
]
weightData['last_weight'] = (
weightData.last_price * weightData.n_stocks) / (
weightData.last_price *
weightData.n_stocks).sum() # 투자기간의 마지막 시점에서의 비중
weightData['new_weight'] = 0
tradingCost = 0
dollar_ongoing -= tradingCost
basket_history[rebalDate] = basket_price
inv_money_list += list(basket_price.values.squeeze()[1:-1])
inv_money_list += [dollar_ongoing]
inv_money_date += list(basket_price.index[1:])
tradeCost_history[''.join(
[x for x in str(lastDay)[:10] if x != '-'])] = tradingCost
turnover_history[''.join([
x for x in str(lastDay)[:10] if x != '-'
])] = tradingCost / dollar_ongoing
weight_history[''.join([x for x in str(lastDay)[:10]
if x != '-'])] = weightData
inv_money_history = pd.DataFrame(inv_money_list, index=inv_money_date)
return inv_money_history, tradeCost_history, turnover_history, weight_history
df = pd.DataFrame(index=range(len(codes)),
columns=['date', 'code', 'weight'])
df['code'] = codes
df['weight'] = np.ones(len(codes)) / len(codes)
df['date'] = rebalDate
return df
rebalData_long = []
rebalData_short = []
num_group = 5
method = 'integrated'
for i in tqdm(range(len(rebal_sche))):
date_spot = util.get_recentBday(rebal_sche[i], dateFormat='datetime')
univ_ = util.getUniverse(marketInfo, mktcap, risk_1, risk_2, date_spot)
psr_spot = util.getFinancialData(factor_PSR, date_spot)[univ_]
pbr_spot = util.getFinancialData(factor_PBR, date_spot)[univ_]
# I-1. Integrated Method (Signal Blend, 개별 팩터 스코어를 모두 더해 한번에 주식을 뽑는 방법)
if method == 'integrated':
factorName = ['sales', 'book']
# factorName = ['book']
multifactor_df = pd.concat([psr_spot, pbr_spot], axis=1, sort=False)
# multifactor_df = pd.concat([pbr_spot], axis = 1, sort = False)
multifactor_df.columns = factorName
mktcaps = util.get_mktcap(multifactor_df.index.values, date_spot,
date_spot)
