no_dominated.py

#-- Create random portfolio weights and indexes
#- How many assests in the portfolio
n_assets = 5

mean_variance_pairs = []
#-- Store weights and tickers for labelling
weights_list=[]
tickers_list=[]

for i in tqdm(range(1000)):
    next_i = False
    while True:
        #- Choose assets randomly without replacement
        assets = np.random.choice(tickers, n_assets, replace=False)
        #- Choose weights randomly ensuring they sum to one
        weights = np.random.rand(n_assets)
        weights = weights/sum(weights)

        #-- Loop over asset pairs and compute portfolio return and variance
        portfolio_E_Variance = 0
        portfolio_E_Return = 0
        for i in range(len(assets)):
            portfolio_E_Return += weights[i] * df.mu.loc[assets[i]]
            for j in range(len(assets)):
                portfolio_E_Variance += weights[i] * weights[j] * cov.loc[assets[i], assets[j]]

        #-- Skip over dominated portfolios
        for R,V in mean_variance_pairs:
            if (R > portfolio_E_Return) & (V < portfolio_E_Variance):
                next_i = True
                break
        if next_i:
            break

        #-- Add the mean/variance pairs to a list for plotting
        mean_variance_pairs.append([portfolio_E_Return, portfolio_E_Variance])
        weights_list.append(weights)
        tickers_list.append(assets)
        break