Mean-Variance Optimization and the Total Portfolio Approach

In 1952, Harry Markowitz was a 25-year-old PhD student at the University of Chicago when he wrote a 14-page paper titled "Portfolio Selection." His advisor, Milton Friedman, reportedly told the dissertation committee the work was not economics, not mathematics, not business administration, and not anything they could identify. The committee passed him anyway. Thirty-eight years later, Markowitz won the Nobel Prize.

Why MVO exists. Before Markowitz, allocation was stock picking stacked on top of stock picking. Markowitz reframed the problem: a portfolio's risk is not the average of individual risks, it is driven by how the holdings move together. Modern portfolio theory says less than perfectly correlated assets can be combined to maximize return for any given level of risk. The curriculum starts not from a frontier picture but from how an investor trades expected return against volatility, using utility.

Utility, expected utility, and the utility function. In an investments context, utility is a measurement of the satisfaction an investor receives from investment wealth or return.

Common mistakes

• Minimizing variance with no regard for risk aversion. The curriculum's MVO maximizes $\mu -(\lambda /2){\sigma }^2$, and the optimal weight scales with $1/\lambda$. A question that asks how the allocation to the risky asset changes when $\lambda$ rises is testing that risky weights fall. Trap: treating MVO as pure variance minimization with no $\lambda$ term.
• Forgetting that a negative-beta asset can have a negative hurdle rate. The hurdle is $E[R_{New}]-R_f>(E[R_p]-R_f){\beta }_{New}$. With $\beta =-0.3$, an optimal portfolio at 16 percent and a 1 percent riskless rate yields a hurdle of $-3.5$ percent, so an asset can lose money and still improve the portfolio.
• Using reported correlations and volatilities for alternatives without unsmoothing. Smoothed appraisal returns understate variance and covariance, so the optimizer over-allocates. Anson (2016) found unsmoothed volatility roughly double the smoothed figure and optimal allocations 20 percent lower for PE and 30 percent lower for VC.

Bottom line

• MVO maximizes mean-variance utility $E[U(W)]=\mu -(\lambda /2){\sigma }^2$; concave utility means risk averse, linear means risk neutral, and $\lambda$ is the risk-aversion coefficient.
• Risk aversion can be recovered from a chosen portfolio via $\lambda =(E[R_p]-R_f)/{\sigma }_p^2$.
• Closed-form weights scale inversely with $\lambda$; for one risky asset $w=(1/\lambda )(E[R-R_0]/{\sigma }^2)$, and a positive liability covariance adds $L(\delta /{\sigma }^2)$.
• Unconstrained MVO produces shorts, weights above 100 percent, and leverage; nonnegativity constraints zero out the worst asset but yield less-optimal portfolios.

Exam shortcut

The CAIA Level 2 question bank leans on two fulcrums: how $\lambda$ and the closed-form weights drive allocations, and the four-dimension TPA framework versus SAA. Memorize: $\lambda =(E[R_p]-R_f)/{\sigma }_p^2$ backs risk aversion out of a portfolio; $w=(1/\lambda )(E[R-R_0]/{\sigma }^2)$ for one risky asset; unsmoothing roughly doubles alternative volatility and cuts PE/VC allocations 20/30 percent; the four TPA dimensions are Governance, Factor lens, Competition for capital, and Culture.

The full lesson (about 6,334 words, 42 min read) adds 2 worked examples, all 6 common mistakes, a self-check, free in the app.

Learning objectives

• mvo process
• mvo implementation
• mvo multiple risky
• mvo issues
• mvo adjustments
• mvo estimation error
• tpa overview
• tpa defining
• tpa governance
• tpa factor lens
• tpa competition capital
• tpa culture
• tpa implementing
• core satellite
• top down bottom up
• risk budgeting
• factor risk budgeting
• risk parity
• other quant strategies
• taa
• taa process
• cash commitments illiquidity
• liquid alternatives
• lp direct investment
• co investments
• co investment returns
• secondary market pe
• gp led secondaries
• rebal buy hold constant mix
• rebal directional
• rebal cppi
• rebal obpi
• rebal dynamic illiquid
• rebal costs

Browse all free CAIA Level II lessons or jump into free CAIA Level II practice questions.