Portfolio optimizer: a step-by-step guide

QuantMundi is a free portfolio optimizer that works with more than 1,300 stocks, ETFs and bonds from around the world, and calculates everything in your currency. This guide takes you from zero to your first portfolio in a few minutes, and explains what each part of the results means.

One note before starting: QuantMundi is an educational tool. The results are what a mathematical model estimates from historical data, not investment advice.

Step 1: choose your assets

There are three ways to build your list:

• Search directly. In the optimizer, type a name or ticker in the "Your list" search box: Apple, Nestlé, EWZ… You can add up to 25 assets.
• Use the explorer. In the explorer you filter by region, country, type, sector, index, currency or exchange. Quick picks like "Top Latin America", "Country ETFs" or "World bonds" give you ready-made lists. When you have your selection, press "Optimize my list".
• Start from an example. If you don't know where to begin, try the "Global diversified" example, with stocks from several regions, bonds and gold.

A tip: mix assets that move differently. Ten Latin American banks don't diversify as much as five assets from different regions and types. We explain it in correlation and diversification.

Step 2: set up the calculation

In the Settings section:

• Method: how the money is split. To start, Maximum Sharpe looks for the best return per unit of risk, and Minimum variance looks for the most stable portfolio. There are also Target return, Target risk, Hierarchical Risk Parity (HRP) and Equal weights.
• Historical period: how many years of data to use. More years include more crises and give more robust results.
• Your currency: everything is converted to your currency using each day's exchange rate, so currency risk is included.
• Risk-free rate: what you'd earn without risk in your currency. It's used for the Sharpe ratio. It comes with an approximate value you can adjust.

Advanced options

Under "Constraints and advanced options" you can:

• Set a minimum and maximum weight per asset. A maximum of 20% or 25% stops the model from putting almost everything into one or two assets.
• Add limits by region, country, sector or currency, for example "at most 30% in Latin America".
• Choose the return frequency. In automatic mode, the model uses weekly returns if you mix Asia with the Americas, because exchanges close at different times.
• Choose what to compare the backtest against (World, S&P 500, Latin America, Emerging markets or Equal weights) and the rebalancing (none, monthly or quarterly). More on this in how to rebalance a portfolio.

Press "Calculate optimal portfolio". In a few seconds you'll see the results.

Step 3: understand the results

The key metrics

At the top you'll see expected annual return, volatility, Sharpe ratio, maximum drawdown, VaR and CVaR. Hover over each one for a short explanation. We explain the loss measures in VaR, CVaR and maximum drawdown.

Why these weights?

A plain-language explanation of why the model chose each asset: for its stability, for its return-to-risk trade-off or because it diversifies well.

Efficient frontier

The chart of all the best possible risk-return combinations with your assets. Your portfolio is marked on the curve. We explain it in what is the efficient frontier.

Geographic exposure

A map and charts showing where your money is: by country, region, sector and currency. If more than half is in a single place, you'll see a concentration warning.

Backtest

How much the portfolio would have been worth since the start of the period, compared with a benchmark index. Careful: it uses the same data the portfolio was optimized on, so it tends to look better than the future would be.

Correlations

The matrix showing how much your assets move together. The colors help you quickly spot the assets that diversify the most.

How it was calculated

The technical details: prices adjusted for dividends and splits, covariance with Ledoit-Wolf shrinkage and expected returns with the Bayes-Stein estimator, which reduces the weight of extreme past returns.

Step 4: compare and adjust

• Compare methods: turn on "Compare methods" and pick up to 3. You'll see, for example, Maximum Sharpe, Minimum variance and HRP side by side. If two methods give similar results but one is much more concentrated, the more diversified one tends to be more robust.
• Change one thing at a time: the period, the maximum weight or an asset. That way you understand what drives the result.
• Be wary of extremes: if the model puts 60% in a single asset, it's probably because that asset had an exceptional period. Cap the maximum weight and calculate again.