Kyle Wisniewski

Every result here is reproducible: open-source code and 229 automated checks, from theory-anchored quantitative tests to publishing checks. Browse the source on GitHub →

Quantitative model · Portfolio construction

Efficient-frontier geometry under uncertain inputs

Expected returns, volatility, correlations, and constraints decide which portfolios look efficient. Every input here is illustrative, not a forecast. Illustrative inputs

Decision summary

An efficient portfolio is only as stable as its inputs

Interpretation

The efficient frontier depends on expected returns and covariances that are not known in advance. A precise optimizer can amplify small estimation errors into concentrated allocations.

Decision context

Choose an allocation method according to confidence in the inputs. In the lab’s locked test, risk-based portfolios reduced out-of-sample volatility and drawdown. They did not reliably deliver higher returns.

Intended analytical use

Asset allocators, investment committees, model validators, and analysts use it to compare optimization with risk budgeting.

Principal limitation

This explorer uses illustrative capital-market assumptions. The empirical result comes from one frozen 2015–2026 ETF sample and one sealed evaluation window.

Universe

Parameters

Highlighted portfolios

Performance statistics for highlighted portfolios
PortfolioE[r]σSharpe
Min variance
Max Sharpe
Equal weight

Efficient frontier risk-return plot

Long-only random portfolios colored by Sharpe ratio. Diamonds are individual assets. A circle marks minimum variance, an X maximum Sharpe, and an open triangle equal weight. The dotted line from the risk-free rate through the max-Sharpe portfolio is the capital market line. Exact highlighted values appear in the table beside the chart.

Highlighted portfolio weights

Weights of the highlighted portfolios use both color and distinct fill patterns. Under these inputs, the maximum-Sharpe solution is more concentrated than the minimum-variance solution.
View portfolio weight data

Inputs

Illustrative capital market assumptions

The explorer uses the annualized expected returns, volatilities, and correlations below. They are illustrative: plausible in magnitude and ordering, chosen for pedagogy, not estimated from any particular sample. That is precisely the point. Change them by amounts well within estimation error and the "optimal" portfolio can change dramatically. Expected returns are the least estimable quantity in finance. Mean-variance optimization is maximally sensitive to them.

Research question

How does portfolio construction change when the objective moves from classical return optimization to uncertainty-aware risk allocation?

Three answers to the same problem

Mean-variance optimization (Markowitz, 1952) solves \( \min_w \; w^\top \Sigma w \) subject to a target return \( w^\top \mu = \bar\mu \). It is the intellectual foundation of the comparison. The solution is highly sensitive to estimated expected returns. Small input changes can produce materially different weights.

Risk parity abandons return forecasts entirely and equalizes each asset's contribution to portfolio risk, \( w_i (\Sigma w)_i = w_j (\Sigma w)_j \). It answers uncertainty about \(\mu\) by removing it from the objective. That implicitly assumes risk premia per unit of risk are roughly comparable across assets. Reaching equity-like returns then typically requires leverage on low-volatility assets.

Hierarchical risk parity (López de Prado, 2016) attacks a different weakness: inverting an estimated covariance matrix. HRP clusters assets by correlation distance, then allocates down the resulting tree by inverse variance — no matrix inversion, no return forecasts. It trades in-sample optimality for greater stability under covariance-estimation error.

MVO, risk parity, and HRP encode different assumptions about expected returns and covariance estimates. Their allocations therefore differ in concentration, sensitivity to estimation error, leverage requirements, and dependence on the realized market regime.

On the sampling method Weights are drawn from a symmetric Dirichlet distribution (normalized unit-rate exponentials). That distribution is uniform on the simplex, so every long-only, fully invested portfolio is equally likely. The frontier is then the empirical upper envelope. A quadratic-programming solver would find it exactly. The cloud instead makes the geometry, and the sparseness of near-optimal portfolios, visible.