Why the Average Family Office Portfolio Is Structurally Inefficient
UBS Global Family Office report
UBS published their 2025 Global Family Office Report, surveying over 300 family offices about their portfolios.
On average, FOs in this survey manage USD 1.1 billion.
Data from clients’ investments was gathered to form a strategic portfolio, representing the average asset allocation amongst the following classes: Equities, Private equity, Fixed income, Real estate, Private debt, Cash, Hedge funds, Gold / Precious metals, Commodities, Art and antiques, Infrastructure.
This asset allocation raises the following questions: what are the expected return, risk, of such a portfolio, and are there superior portfolios that invest in the same asset classes?
Portfolio analysis using Markowitz’ model
The Markowitz model seeks to find the best portfolios, given a set of asset classes to invest in. By letting the weights associated with the assets vary, one obtains a portfolio with some level of expected return and risk.
Each one of these portfolios can be plotted in a figure, yielding a convex cloud of points. The efficient frontier is the upper boundary of this cloud — the red line on the figure.
Portfolios on the efficient frontier are the best ones in the following sense: for a given level of returns, there is no other portfolio with a lower risk. Similarly, for a given level of risk, there are no portfolios with a better expected return.
Asset classes performance
Before plotting the cloud of portfolios, we need past data to get a good estimate of the expected returns and risk of each asset class. Public sources such as ETFs, Indices, and Preqin, have been used to gather quarterly data from 2003 up to the second quarter of 2025.
We have omitted Cash, as it represents the liquid part of the portfolio, and usually doesn’t yield returns exceeding the risk-free rate. Commodities and Art / antiques have been removed as well, since their weight is negligible in the portfolio; moreover, it is difficult to find a good proxy of these asset classes’ performance.
Using a 10 year horizon, we compute the annualized returns and risk of each asset, weighted according to the average family office portfolio. These weights add up to 91%, and by normalizing back to 100%, we find an expected return of 10.37%.
Family office portfolio’s position in the cloud
When plotting the strategic portfolio according to the assets performances shown above, we find an expected return of 10.37% as computed above.
The risk (standard deviation) is calculated using the covariance matrix of the asset classes, yielding an annualized risk of 8.13%. This leads to a Sharpe ratio of 0.78 (where we have subtracted the 4.00% risk-free rate from the returns of the portfolio).
While it has a decent Sharpe ratio, this portfolio’s position in the cloud is far from the efficient frontier, and thus severely underperforms many other portfolios, with the best one scoring a Sharpe of 1.31.
The max Sharpe portfolio suggests a different weight allocation.
The following is a theoretical exercise intended to illustrate what an optimal portfolio could look like for a very long-term–oriented investor. In practice, such a portfolio would be difficult to implement. Liquidity risk is high. Concentration risk is significant. Public equity exposure is unusually low. Real estate is underrepresented.
The portfolio has a high concentration in assets such as Private equity, Infrastructure, and Fixed income. It corresponds to the tangency portfolio, whose name derives from the Capital Allocation Line joining the origin (0%, 0%) to the cloud, in a tangent fashion.
This figure shows the effect of reducing the allocation of the assets, individually. Each reduction in weight is proportionally added back to the other assets. The pink dots on the bottom left represent a reduction of equities, impacting the returns by approximately -0.04%, while reducing the risk by almost 1.2%, demonstrating that a strong investment in equities is not generally beneficial for the portfolio.
Similarly, a reduction of private equity leads to a higher risk for a lower return; which goes to say that private equity, with its past performance shown above, is a great asset to invest in.
Conclusion
This analysis highlights a structural inefficiency in the average family office portfolio. Despite access to sophisticated asset classes and long investment horizons, the observed allocation is far from optimal in a mean–variance framework.
Under the theoretical constraints of investing only in the reported asset classes, significantly better portfolios exist. These portfolios deliver higher risk-adjusted returns without increasing overall risk. The gap is not driven by asset availability, but by suboptimal weighting.
The theoretical optimal portfolio presented is not meant to be directly implemented. Liquidity constraints, governance, capital calls, and operational considerations limit practical feasibility. However, it provides a clear benchmark. It shows what is achievable when diversification, correlations, and long-term return drivers are fully exploited.
The key insight is that portfolio construction itself is the dominant source of underperformance. Improving allocation discipline, even marginally, could meaningfully enhance long-term outcomes without changing the investment universe.