# Efficient frontier

Efficient frontier is the set of portfolios offering the highest estimated expected return for each level of volatility, or the lowest volatility for each estimated expected return, under specified assumptions.

## What it means

The efficient frontier describes a trade-off between a portfolio’s estimated return and its risk. In the mean-variance framework associated with Markowitz (1952), risk is measured by the standard deviation of portfolio returns. A portfolio is *efficient* if no other feasible portfolio offers a higher estimated expected return at the same or lower volatility.

The frontier is not a list of securities to trade. It is the upper boundary of feasible risk-return combinations for a specified set of assets, constraints and estimates. Change those inputs, and the frontier can change. A portfolio below the frontier is inefficient *within that model*: another feasible mix has a better estimated risk-return combination.

## How it is calculated

For a portfolio of assets, calculate estimated expected return and volatility from asset weights, estimated returns and covariances:

- `E(Rp) = Σ wi × E(Ri)`
- `σp = √(Σi Σj wi × wj × Cov(Ri, Rj))`

Here, `E(Rp)` is estimated expected portfolio return; `wi` is asset `i`’s portfolio weight, expressed as a decimal; `E(Ri)` is its estimated expected return; `σp` is portfolio return standard deviation; and `Cov(Ri, Rj)` is the covariance of returns for assets `i` and `j`. Returns and covariances must use consistent periods: annual returns with annual covariances, for example. Weights sum to 1 for a fully invested portfolio.

An optimizer varies the weights subject to rules such as no short selling (`wi ≥ 0`). For each target expected return, it finds the feasible portfolio with the lowest variance, `σp²`. The efficient frontier is the higher-return portion of that minimum-variance boundary, beginning at the global minimum-variance portfolio. Without a risk-free asset, a minimum-variance portfolio is not necessarily the highest-return portfolio.

## Worked example

These are **hypothetical round numbers**, not forecasts. Suppose Stock A has an estimated annual return of 8% and annual volatility of 10%; Stock B has 12% and 20%, respectively. Assume their annual return correlation is zero, no short selling and no transaction costs.

1. Put 50% in each: `E(Rp) = 0.5 × 8% + 0.5 × 12% = 10%` per year.
2. With zero correlation, covariance is zero. Portfolio variance is `0.5² × 10%² + 0.5² × 20%² = 0.0125` in decimal-return-squared units. Volatility is `√0.0125 ≈ 11.18%` per year.
3. Compare an 80% A / 20% B mix: estimated return is `0.8 × 8% + 0.2 × 12% = 8.8%`; volatility is `√(0.8² × 0.10² + 0.2² × 0.20²) ≈ 8.94%` per year.

Under these assumptions, the second mix has lower estimated return and lower volatility. That comparison alone does not make either mix efficient: identifying the frontier requires checking *all* permitted weights. For these two assets and assumptions, the 50/50 mix is on the efficient portion; the 80/20 mix is not, because a feasible mix offers the same estimated return at lower volatility.

## Why it matters before you trade

A proposed trade changes portfolio weights, so it can change estimated return, volatility and diversification. Checking the resulting portfolio—not just the individual stock—helps distinguish an intended change in risk from an unintended concentration.

The frontier omits practical costs unless they are explicitly modeled. Bid-ask spread, market impact, commissions and taxes can alter the economics of moving between portfolios. Historical estimates can also shift, and standard deviation does not capture every risk, including liquidity problems or severe losses.

## How PreTrAIde uses it

The product facts do not state that PreTrAIde calculates or optimizes an efficient frontier. Its pre-trade analysis checks a proposed stock basket before an order is placed. It provides portfolio breakdowns, a Herfindahl-Hirschman concentration index, volatility and liquidity scores, estimated execution costs and a risk assessment. Those checks can add practical context to a portfolio allocation considered through a frontier model; a PreTrAIde verdict is decision support, not an investment recommendation.

## Common mistakes

- Treating estimated returns, correlations or the frontier itself as certain future outcomes.
- Calling any low-volatility portfolio efficient without comparing it with feasible alternatives.
- Assuming an efficient portfolio is suitable for every investor: constraints and risk tolerance differ.
- Ignoring trading costs, concentration and liquidity when translating theoretical weights into orders.

---
Source: https://pretraide.com/glossary/efficient-frontier