def get_data(assets, start_date):
prices = {}
for asset, ticker in assets.items():
prices[asset] = get_close_price(ticker, start_date)
df = DataFrame(prices)
return df
def Econ_env(YYYY, m, dd):
start_date = datetime.datetime(YYYY, m, dd)
GDP = DataReader('GDP', "fred", start=start_date)
sp500 = DataReader('^GSPC', "yahoo", start=start_date)
Array = DataFrame({'S&P':sp500["Adj Close"]})
return Array
def main():
# define the desired portfolio characteristics
std_max = 100 # maximum standard deviation
MAX_ITERS = 100 # max number of iterations
lam = .94 # exponential decay number
assets = (['GOOGLE', 'APPLE', 'CAT', 'SPDR_GOLD', 'OIL',
'NATURAL_GAS', 'USD', 'GOLDMANSACHS', 'DOMINION'])
# Pull data from Yahoo! Finance
GOOG= Stock_Close('GOOG', 2010, 1, 1)
AAPL = Stock_Close('AAPL', 2010,1, 1)
SP500 = Stock_Close('^GSPC', 2010, 1, 1)
CAT = Stock_Close('CAT', 2010, 1, 1)
GOLD = Stock_Close('GLD', 2010, 1, 1)
GAS = Stock_Close('GAZ', 2010, 1, 1)
OIL = Stock_Close('OIL', 2010, 1, 1)
GS = Stock_Close('GS', 2010, 1, 1)
DOM = Stock_Close('D', 2010, 1, 1)
# FX currency
USD = Stock_Close('UUP', 2010, 1, 1)
# create a dataframe housing the above
X = DataFrame({'GOOGLE':GOOG, 'APPLE':AAPL, 'CAT':CAT, 'SPDR_GOLD':GOLD,
'OIL':OIL, 'NATURAL_GAS':GAS, 'USD':USD, 'GOLDMANSACHS': GS, 'DOMINION':DOM})
# define weights of each asset held
# weights = {'GOOGLE':.2,
# 'APPLE':.1,
# 'CAT':.1,
# 'SPDR GOLD':.05,
# 'OIL':.1,
# 'NATURAL GAS':.1,
# 'USD':.05,
# 'DOMINION':.1,
# 'GOLDMANSACHS':.2}
best = zeros(((2+len(assets)), MAX_ITERS))
for i in range(1, MAX_ITERS):
# check to make sure sum weights = 1
numWeight = DirichletDistro(len(assets),1)
if int(sum(numWeight)) < 1.01:
#create dictionary of weights for each of the assets
weights = dict(zip(assets, numWeight))
profit, STD, EWMA = Stock_stats(X, weights, assets, lam)
# store trial profit, stdev in column of best
best[0:2, i] = [ profit, STD]
best[2::, i] = numWeight
else:
print('Sum of weights does not equal 1')
# now that we have the monte carlo simulation setup, eliminate
# all trials that do not meet critera
for i in range(1, MAX_ITERS):
if best[1, i] >std_max:
best[:, i] = zeros(( (2+len(assets)), 1)) # drop the trial
# add more critera here...
else:
pass
print( max(sum(best, 0)) ) # print the maximum portfolio
return X, weights, STD, profit, numWeight, best, EWMA
def main():
# define the desired portfolio characteristics
std_max = .2 # maximum standard deviation
MAX_ITERS = 200 # max number of iterations
lam = .94 # exponential decay number
exit_date = 12 # when you sell stocks (in months)
window_begin = 6 # how far back you want to window reg. (in months)
beta = .6 # personal risk adversion level
delta = .99 # discount factor
assets = (['GOOGLE', 'APPLE', 'CAT', 'SPDR_GOLD', 'OIL',
'NATURAL_GAS', 'USD', 'GOLDMANSACHS', 'DOMINION'])
print('Pulling data from Yahoo! Finance')
# Pull data from Yahoo! Finance
GOOG= Stock_Close('GOOG', 2010, 1, 1)
AAPL = Stock_Close('AAPL', 2010,1, 1)
SP500 = Stock_Close('^GSPC', 2010, 1, 1)
CAT = Stock_Close('CAT', 2010, 1, 1)
GOLD = Stock_Close('GLD', 2010, 1, 1)
GAS = Stock_Close('GAZ', 2010, 1, 1)
OIL = Stock_Close('OIL', 2010, 1, 1)
GS = Stock_Close('GS', 2010, 1, 1)
DOM = Stock_Close('D', 2010, 1, 1)
# FX currency
USD = Stock_Close('UUP', 2010, 1, 1)
# create a dataframe housing the above
X = DataFrame({'GOOGLE':GOOG, 'APPLE':AAPL, 'CAT':CAT, 'SPDR_GOLD':GOLD,
'OIL':OIL, 'NATURAL_GAS':GAS, 'USD':USD, 'GOLDMANSACHS': GS, 'DOMINION':DOM})
best = zeros(((4+len(assets)), MAX_ITERS))
print('Running monte carlo simulation')
for i in range(1, MAX_ITERS):
print('Percent Done: \t' + str(float(i)/float(MAX_ITERS)*100)+' %')
numWeight = DirichletDistro(len(assets),1)
EU = {}
# check to make sure sum weights = 1
if int(sum(numWeight)) < 1.01: # account for floating point err
#create dictionary of weights for each of the assets
weights = dict(zip(assets, numWeight))
STD, EWMA = Stock_stats(X, weights, assets, lam)
# calculate price at future period
regr, window, projection, profit = regression(X, assets, window_begin, exit_date)
# calculate expected utility
for ass in assets:
profit[ass] = profit[ass] * weights[ass]
EU[ass] = utl(profit[ass], EWMA[ass], beta, delta, exit_date*30)
# calculate the variance of the portfolio
var_portfolio = var(X, weights, assets)
# # check if the EWMA is above the specified limit
if var_portfolio == std_max:
best[:, i] = zeros(( (4+len(assets)) )) # drop the trial
# add more critera here...
else: # store trial profit, EWMA, STD in column of best
best[0:4, i] = [sum(EU.values()), var_portfolio, sum(EWMA.values()), sum(profit.values())]
sorted_assest = sorted(weights.keys())
print(sum(profit.values()))
j = 4
for ass in sorted_assest:
best[j, i] = weights[ass]
j = j+1
else:
print('Sum of weights does not equal 1')
maxP = max(best[0,:])
print( "The maximum profit to be made is: %f") % maxP
# find the column where the sum is equal to the max
opt = where(best[0,:] == maxP)
# OptimalAllocation = dict(zip(assets, [float(xx) for xx in best[2:,opt]]))
print('\n The optimal asset allocation in the portfolio is:')
# print(OptimalAllocation)
return X, best, assets, EU, regr, profit