import pandas as pd
import pandas_datareader.data as web
import datetime
import numpy as np
%matplotlib inline
import matplotlib.pyplot as plt
import math
import pyfolio as pf
import quantstats
plt.rcParams["figure.figsize"] = (10, 6) # (w, h)
import sys
from scipy.stats import rankdata
from scipy.stats import stats
from scipy.optimize import minimize

Monthly_Return = pd.read_excel('MonthlyAssetClassReturn.xlsx')
Monthly_Return.head()

Monthly_Return = Monthly_Return.set_index('Data Index')
Mon_Return = Monthly_Return.loc[:, ['S&P 500 Total return','MSCI EAFE','EEM','US 10 YR','US 30 YR','US Corp Bond Return Index','TIPS Series','GOLD','GSCI']]
Mon_Return = Mon_Return.loc[Mon_Return.index >= '1973-02-28']
Mon_Return.head()

covmat = pd.DataFrame.cov(Mon_Return)
covmat

def RC(weight, covmat) :
    weight = np.array(weight)
    # @: 행렬곱
    variance = weight.T @ covmat @ weight
    sigma = variance ** 0.5
    mrc = 1/sigma * (covmat @ weight)
    rc = weight * mrc
    rc = rc / rc.sum()
    return(rc)

def RiskParityObjective(x) :
    
    variance = x.T @ covmat @ x
    sigma = variance ** 0.5
    mrc = 1/sigma * (covmat @ x)
    rc = x * mrc
    a = np.reshape(rc.to_numpy(), (len(rc), 1))
    risk_diffs = a - a.T
    # np.ravel: 다차원 배열 -> 1차원 배열
    # 1차원으로 풀어준 후에 제곱 -> 합 -> 목적함수
    sum_risk_diffs_squared = np.sum(np.square(np.ravel(risk_diffs)))
    return (sum_risk_diffs_squared)

# 제약 조건: 비중 합 1
def SumConstraint(x):
    return (x.sum()-1.0)

# 제약 조건: 비중 0 이상
def LongOnly(x):
    return(x)

def RiskParity(covmat) :
    
    x0 = np.repeat(1/covmat.shape[1], covmat.shape[1]) 
    constraints = ({'type': 'eq', 'fun': SumConstraint},
                  {'type': 'ineq', 'fun': LongOnly})
    options = {'ftol': 1e-20, 'maxiter': 2000}
    
    result = minimize(fun = RiskParityObjective,
                      x0 = x0,
                      method = 'SLSQP',
                      constraints = constraints,
                      options = options)
    return(result.x)

RiskParity(covmat)

weight_rp = RiskParity(covmat)
RC(weight_rp, covmat)

weight_equal = np.repeat(1/Mon_Return.shape[1], Mon_Return.shape[1])
rc_equal = RC(weight_equal, covmat)

rc_equal

covmat_6040 = pd.DataFrame.cov(Mon_Return[['S&P 500 Total return', 'US 10 YR']])
weight_6040 = np.array([0.6, 0.4])
rc_6040 = RC(weight_6040, covmat_6040)

rc_6040

RP_Mon_12 = Mon_Return
RP_Weight = []

for i in range(len(RP_Mon_12)):
    # use previous 12 months
    #print(i)
    if i < 12:
        RP_Weight.append('')
    else:
        covmat = pd.DataFrame.cov(RP_Mon_12.iloc[i-12 : i-1])
        temp = []
        temp.append(RiskParity(covmat))
        RP_Weight.append(temp)

RP_Mon_12['RP_Weight'] = RP_Weight    
RP_Mon_12

RP_Mon_12 = RP_Mon_12.loc[RP_Mon_12.index >= '1974-02-28']
RP_Mon_12

RP_Return = []
ret = 0
for i in range(len(RP_Mon_12)):
    # 자산군 수익률 * 자산군 비중. 9개 자산군
    for j in range(9):
        ret += RP_Mon_12.iloc[i,j]*RP_Mon_12['RP_Weight'][i][0][j]

    RP_Return.append(ret)
    ret = 0

RP_Mon_12['RP_Return'] = RP_Return
RP_Mon_12

quantstats.stats.sharpe(RP_Mon_12['RP_Return'])/math.sqrt(252/12)

quantstats.reports.plots(RP_Mon_12['RP_Return'], mode='basic')

quantstats.reports.metrics(RP_Mon_12['RP_Return'], mode='full')