自動再平衡

將台股資料匯入，並執行策略，來看最終的績效如何。
其中會因為觸發再平衡條件而自動再平衡

Requirements

python >= 3.8
requirements.txt

Install package

pip install -r requirements.txt

Run

python run.py

Description

Duration

2011/1~今天

Strategy

選股池
台股市值前150

利用回歸模型，計算出各檔股票的統計數據（資料來源為股價日報酬）

def Regression (Stock_data):
    StocksList = list(Stock_data.columns)
    return_Stock_data = np.log(Stock_data/Stock_data.shift())
    return_Stock_data = return_Stock_data.fillna(method='ffill')
    return_Stock_data = return_Stock_data.dropna(how = 'all')
    #=============================================================================
    '''迴歸'''
    XX=sm.add_constant(return_Stock_data['^TWII'])
    
    lr_list=[]
    A = {}
    l_alpha = []
    l_alpha_P = []
    l_beta = []
    l_beta_P = []
    
    for g in StocksList:
        est = sm.OLS(return_Stock_data[g] , XX)
        est = est.fit()
        sol = est.summary()
        lr_list.append(sol.as_csv())
        
        An = pd.DataFrame(est.params)
        An['coeff'] = est.params
        An["p_values"] = est.summary2().tables[1]["P>|t|"]
        An = An.iloc[:, [1, 2]]
        An.rename(index = {'const' : 'Alpha','^TWII' : 'Beta'},inplace = True)
        A[g] = An
        
        l_alpha.append(A[g].iloc[0,0]) #alpha
        l_alpha_P.append(A[g].iloc[0,1]) #alpha P-value
        l_beta.append(A[g].iloc[1,0]) #beta
        l_beta_P.append(A[g].iloc[1,1]) #beta
    ''''''
    #=============================================================================
    ''''''
    mean = return_Stock_data.mean()
    std = return_Stock_data.std()
    skew = pd.Series(si.skew(Stock_data), index = (StocksList))
    kurt = pd.Series(si.kurtosis(Stock_data), index = (StocksList))
    alpha = pd.Series(l_alpha, index = (StocksList))
    alpha_P = pd.Series(l_alpha_P, index = (StocksList))
    beta = pd.Series(l_beta, index = (StocksList))
    beta_P = pd.Series(l_beta_P, index = (StocksList))
    
    Stock_stat = pd.concat([mean, std, skew, kurt, alpha, alpha_P, beta, beta_P],axis = 1)
    Stock_stat.columns = ['mean', 'std', 'skew', 'kurt', 'Alpha', 'alpha_P', 'Beta', 'beta_P']
    ''''''
    #=============================================================================
    return Stock_stat

定義熊市、牛市
購買標的當天的大盤若低於既季均線，則將市場視為熊市，反之則為牛市

if MA_66 > tt:
    sit = 'bearish' # 定義熊市
    choose = Stock_stat.sort_values('Beta',ascending = True).head(5)
    stop_loss = 0.15

if MA_66 <= tt:
    sit = 'bullish' # 定義牛市
    choose = Stock_stat.sort_values('Alpha',ascending = False).head(5)
    stop_loss = 0.2

策略

當市場為熊市時
選擇Beta最小的五黨股票，並用Markowitz Efficient Frontier來取得權重
當市場為牛市時
選擇Alpha最小的五黨股票，並用Markowitz Efficient Frontier來取得權重

# return and volatility functions
def Portfolio_performance(weights, mean_returns, cov_matrix):
    returns = np.sum(mean_returns *weights )
    std = np.sqrt(np.dot(weights.T, np.dot(cov_matrix, weights)))
    return std, returns
# volatility function of portfolio
def Portfolio_volatility(weights, mean_returns, cov_matrix):
    return Portfolio_performance(weights, mean_returns, cov_matrix)[0]
# MV model
def min_variance(mean_returns, cov_matrix):
    num_assets = len(mean_returns)
    args = (mean_returns, cov_matrix)
    constraints = ({'type':'eq', 'fun': lambda x: np.sum(x)-1})
    bound = (0,1)
    bounds = tuple(bound for asset in range(num_assets))
    result = minimize(Portfolio_volatility, num_assets *[1/num_assets , ], args=args ,
                    method= 'SLSQP' , bounds=bounds , constraints=constraints )
    return result
''''''
#=============================================================================

def weights(choose_return_Stock_data):
    mean_return = choose_return_Stock_data.mean()
    cov_matrix = choose_return_Stock_data.cov()
    wT = min_variance(mean_return,cov_matrix)['x']
    vol = Portfolio_volatility(wT,mean_return,cov_matrix)
    return wT ,vol

再平衡條件

市場反轉
當牛市變成熊市，或熊市變成牛市時，會先檢查投資組合是否呈現虧損，若為虧損，則觸發自動再平衡
停損
牛市：
停損點為投資期間最高價值，與當日價值之報酬若低於-20%，則觸發停損，執行自動再平衡
熊市：
停損點為投資期間最高價值，與當日價值之報酬若低於-15%，則觸發停損，執行自動再平衡

觸發再平衡條件後，會將所有股票賣出，並重新挑選標的買入。

v_return = (cc - A_value_1)/A_value_1

if v_return < - stop_loss:
    selldate = Assets.index[g]
    print('停損')
    break # 停損再平衡

MA_66 = Stock_data['^TWII'].loc[:Assets.index[g]].rolling(66).mean()[-1]
tt = Stock_data['^TWII'].loc[:Assets.index[g]][-1]
if MA_66 > tt:
    sit_1 = sit_1+1
    if sit_1 !=0:
        sit_2 = 0
    if sit_1 == 10 and sit == 'bullish' and A_value < cc:
        selldate = Assets.index[g]
        print('牛市-->熊市')
        break # 牛市-->熊市 且虧損：再平衡

if MA_66 <= tt:
    sit_2 = sit_2+1
    if sit_2 != 0:
        sit_1 = 0
    if sit_2 == 10 and sit == 'bearish' and A_value < cc:
        selldate = Assets.index[g]
        print('熊市-->牛市')
        break # 熊市-->牛市 且虧損：再平衡

Result

投資組合與大盤之價值比較

	Portfolio	TWII
2011-01-01	1000.0	1000.0
2011-06-29	1043.3140539357225	949.9273577985762
2012-08-17	1223.9606022136643	827.4428246308946
2013-06-26	1551.8476749660604	862.553023613287
2013-09-13	1624.6019486699322	905.0335900525791
2014-08-18	2232.775609721255	1012.8537797478931
2014-09-04	2242.1077051158272	1044.7176670789568
2015-08-20	1793.9409758567822	889.6999558422842
2015-10-21	1824.355214571223	953.8996045498568
2015-12-21	1864.8379053543952	917.6612792666593
2016-03-01	1945.6509743675615	940.2127685589272
2016-11-22	2019.4603608900588	1011.9786324698466
2017-01-13	2032.489335201333	1039.1740938227563
2017-12-18	2422.8753033018156	1164.1229001956635
2018-01-12	2499.778230168822	1205.9433190653017
2018-05-04	2997.5798657191153	1166.6547462835722
2018-05-24	2998.292869332848	1211.8123674713986
2018-07-06	3010.847795184054	1175.4301319656738
2019-01-28	3109.1175747603297	1109.4774836226377
2019-05-27	4348.591907212298	1145.0221183519832
2019-09-19	4478.316326520718	1207.1333375821112
2020-03-02	5093.239818675606	1237.6875336850007
2020-05-18	5224.238999245061	1190.05346643478
2021-05-24	11059.199470723514	1810.283370048474
2021-06-04	10968.039796860177	1899.9339203211296
2022-01-06	10828.822141097882	2035.1664646333052

投資組合與大盤之波動度比較

Resource

yfinance
yfinance資料取得方便且免費，但有許多錯誤資料，導致算出來的績效與實際有些出入。

專案開發期間

2019/12 ~ 2020/1

JT-427 / auto-rebalance-TW

自動再平衡

Requirements

Install package

Run

Description

Duration

Strategy

Result

Resource

專案開發期間

About

Languages