def optimize():
market = Market()
asset_returns = compute_market_returns(market)
min_lower = -10
valid_portfolios = []
for i in range(0, 100):
asset_weights = {}
left = 1
for asset in market.get_assets():
weight = random() * left
asset_weights[asset] = weight
left = left - weight
if left > 0:
asset_weights[market.get_assets()[:1][0]] += left
portfolio = Portfolio('Portfolio' + str(i), asset_weights)
portfolio_returns = evaluate_portfolio(market, portfolio, asset_returns)
if np.percentile(portfolio_returns, 25) > min_lower:
valid_portfolios.append((portfolio, portfolio_returns))
valid_portfolios.sort(key=lambda p: np.mean(p[1]), reverse=True)
best_portfolio = valid_portfolios[0]
comparisons = [compute()]
to_display = comparisons.append(best_portfolio)
return to_display
def build_random_portfolio():
asset_count = random.randint(1, len(MARKET.assets))
new_portfolio_values = {}
for asset in random.sample(MARKET.assets, asset_count):
new_portfolio_values[asset] = random.random()
portfolio = Portfolio(random_name(), adjust_weight(new_portfolio_values))
return portfolio
def add_random_asset(portfolio):
new_portfolio_values = portfolio.asset_and_weights.copy()
if len(new_portfolio_values) < len(MARKET.get_assets()):
asset_to_add = random.choice(
list(set(MARKET.get_assets()) - set(new_portfolio_values.keys())))
weight = random.random()
new_portfolio_values[asset_to_add] = weight
return Portfolio(random_name(), adjust_weight(new_portfolio_values))
def change_random_asset(portfolio):
new_portfolio_values = portfolio.asset_and_weights.copy()
asset_to_adjust = random.choice(list(portfolio.asset_and_weights.keys()))
value_to_add = random.random() - 0.5
new_value = portfolio.weight(asset_to_adjust) + value_to_add
if new_value <= 0:
new_portfolio_values.pop(asset_to_adjust)
else:
new_portfolio_values[asset_to_adjust] = new_value
return Portfolio(random_name(), adjust_weight(new_portfolio_values))
def test_evaluate_portfolio_return(self):
single_asset_portfolio = Portfolio('US Stocks only',
{MARKET.real_estate: 1})
portfolio_return = MARKET.evaluate_portfolio_return(
single_asset_portfolio)
self.assertAlmostEqual(portfolio_return.mean,
MARKET.real_estate.mean,
delta=0.5)
def crossover(first_portfolio, other_portfolio):
new_portfolio_values = {}
all_assets = set(first_portfolio.asset_and_weights.keys()).union(
other_portfolio.asset_and_weights.keys())
for asset in all_assets:
pick_from = first_portfolio if random.random(
) > 0.5 else other_portfolio
if pick_from.weight(asset) > 0:
new_portfolio_values[asset] = pick_from.weight(asset)
if sum(new_portfolio_values.values()) == 0:
new_portfolio_values[list(all_assets)[0]] = 1
return Portfolio(random_name(), adjust_weight(new_portfolio_values))
def test_portfolios(self):
portfolio_0_100 = Portfolio('Pure Bond', {
MARKET.eu_stocks: 0,
MARKET.agg_bonds: 1,
MARKET.gold: 0
})
portfolio_50_50 = Portfolio('50/50', {
MARKET.eu_stocks: 0.5,
MARKET.agg_bonds: 0.5,
MARKET.gold: 0
})
portfolio_60_40 = Portfolio('60/40', {
MARKET.eu_stocks: 0.6,
MARKET.agg_bonds: 0.4,
MARKET.gold: 0
})
portfolio_100_0 = Portfolio('Pure Equity', {
MARKET.eu_stocks: 1,
MARKET.agg_bonds: 0,
MARKET.gold: 0
})
golden_butterfly = Portfolio('Golden Butterfly', {
MARKET.eu_stocks: 0.4,
MARKET.agg_bonds: 0.4,
MARKET.gold: 0.2
})
varx = Portfolio(
'VarX', {
MARKET.ch_stocks: 0.175,
MARKET.us_stocks: 0.175,
MARKET.eu_stocks: 0.20,
MARKET.agg_bonds: 0.15,
MARKET.real_estate: 0.05,
MARKET.cash: 0.15,
MARKET.gold: 0.1
})
varx_plus = Portfolio(
'VarX+', {
MARKET.ch_stocks: 0.185,
MARKET.us_stocks: 0.185,
MARKET.eu_stocks: 0.20,
MARKET.agg_bonds: 0.15,
MARKET.real_estate: 0.05,
MARKET.cash: 0.13,
MARKET.gold: 0.10
})
current = Portfolio(
'current', {
MARKET.ch_stocks: 0.055,
MARKET.us_stocks: 0.12,
MARKET.eu_stocks: 0.285,
MARKET.agg_bonds: 0.105,
MARKET.real_estate: 0.005,
MARKET.cash: 0.375,
MARKET.gold: 0.055
})
dalio_curr = Portfolio(
'Bridgewater', {
MARKET.eu_stocks: 0.05,
MARKET.ch_stocks: 0.3,
MARKET.us_stocks: 0.4,
MARKET.agg_bonds: 0.05,
MARKET.gold: 0.2
})
portfolios = [
portfolio_0_100, portfolio_50_50, portfolio_60_40, portfolio_100_0,
golden_butterfly, current, varx, varx_plus, dalio_curr
]
for p in portfolios:
print(MARKET.evaluate_portfolio_return(p))
def remove_random_asset(portfolio):
new_portfolio_values = portfolio.asset_and_weights.copy()
if len(portfolio.asset_and_weights.items()) > 1:
removed_item = random.choice(list(portfolio.asset_and_weights.items()))
new_portfolio_values.pop(removed_item[0])
return Portfolio(random_name(), adjust_weight(new_portfolio_values))
def compute():
market = Market()
portfolio_0_100 = Portfolio('Pure Bond', {
market.eu_stocks: 0,
market.agg_bonds: 1,
market.gold: 0
})
portfolio_50_50 = Portfolio('50/50', {
market.eu_stocks: 0.5,
market.agg_bonds: 0.5,
market.gold: 0
})
portfolio_60_40 = Portfolio('60/40', {
market.eu_stocks: 0.6,
market.agg_bonds: 0.4,
market.gold: 0
})
portfolio_100_0 = Portfolio('Pure Equity', {
market.eu_stocks: 1,
market.agg_bonds: 0,
market.gold: 0
})
golden_butterfly = Portfolio('Golden Butterfly', {
market.eu_stocks: 0.4,
market.agg_bonds: 0.4,
market.gold: 0.2
})
varx = Portfolio('VarX', {
market.ch_stocks: 0.175,
market.us_stocks: 0.175,
market.eu_stocks: 0.20,
market.agg_bonds: 0.15,
market.real_estate: 0.05,
market.cash: 0.15,
market.gold: 0.1
})
current = Portfolio('current', {
market.ch_stocks: 0.055,
market.us_stocks: 0.12,
market.eu_stocks: 0.285,
market.agg_bonds: 0.105,
market.real_estate: 0.005,
market.cash: 0.375,
market.gold: 0.055
})
dalio_curr = Portfolio('Bridgewater', {
market.eu_stocks: 0.05,
market.ch_stocks: 0.3,
market.us_stocks: 0.4,
market.agg_bonds: 0.05,
market.gold: 0.2
})
portfolios = [portfolio_0_100,
portfolio_50_50,
portfolio_60_40,
portfolio_100_0,
golden_butterfly,
current,
varx,
dalio_curr]
asset_returns = compute_market_returns(market)
all_returns = [evaluate_portfolio(market, p, asset_returns) for p in portfolios]
fig = plt.figure()
ax = fig.add_subplot()
plt.yticks(range(-20, 20)[::2])
ax.grid(b=True, which='both', axis='y')
ax.boxplot(all_returns, sym='')
ax.set_xticklabels([p.name for p in portfolios])
plt.show()
return zip(portfolios, all_returns)