Derivative pricing and asset allocation appear to solve very different problems. One asks what a financial claim is worth. The other asks how capital should be invested.
Yet the quantitative methods in both fields have evolved along a strikingly similar path. Both began with elegant models that made difficult problems tractable. As practitioners asked more realistic questions, time mattered, decisions became dynamic, and outcomes depended not just on where you ended up, but on how you got there. Analytical solutions increasingly gave way to numerical and simulation-based methods.
Option pricing worked through that transition first. Its evolution offers a useful perspective on the continuing evolution of portfolio construction.
Black-Scholes transformed option pricing by showing that, under a set of simplifying assumptions, the value of a European option could be calculated analytically.
Markowitz did something similarly transformational for portfolio construction. Expected returns, volatilities, and correlations could be brought together in a systematic framework to identify portfolios that balance expected return against risk.
Both remain foundational. And both draw much of their power from the same source: a deliberate decision about what to leave out.
For a European option, only the value of the underlying at maturity determines the payoff. How the asset arrived there is irrelevant. Mean-variance optimization makes a comparable simplification. The investor selects an allocation based on expected return and risk over a specified horizon, while much of what may happen between today and that horizon is compressed into a relatively small set of summary assumptions.
For many problems, that is exactly the right trade-off. But not for all of them.
Consider the move from vanilla to exotic options. For a European call, the terminal price is sufficient. But an exotic payoff may depend on whether a barrier was crossed before maturity, on the highest or lowest price observed along the way, or on the average price over the life of the contract. Once those features appear, the final price alone no longer determines the outcome.
Long-term asset allocation faces the same challenge, and the magnitude of the difference can be easy to underestimate. Consider a single multi-asset portfolio over a 20-year horizon and measure its risk in two ways. In one example, the cumulative probability of ending the horizon with a loss falls to roughly 0.1%. At the same time, the probability of experiencing a loss in an individual intermediate period remains between roughly 26% and 38%.
Same portfolio. Same inputs. Two very different views of risk.
Both are meaningful. The terminal measure describes the chance of finishing below the starting point; the within-horizon measure captures losses an investor may experience along the journey. FactSet's earlier work on multi-horizon risk makes the same point: different risk measures can all be correct, but the appropriate measure depends on how the investor experiences risk and how frequently the portfolio is evaluated (Sui, 2022b). As Kritzman and Rich emphasized, exposure during the investment period can differ materially from exposure measured only at its conclusion.
This is why within-horizon measures such as drawdown, probability of insolvency, or probability of breaching a funding threshold differ in kind from terminal return and volatility. They also challenge the comfortable statement that time diversifies risk. The probability of a terminal loss may decline with horizon, while the probability of experiencing a large drawdown somewhere along the way can increase.
For an investor drawing income, a severe loss may force asset sales at an unfavorable time. A pension plan may breach a funding threshold. An investment committee may abandon a strategy after a drawdown even if it would eventually recover. Identical terminal wealth does not necessarily imply an identical investment experience.
Here the analogy becomes deeper. Dynamic trading rules introduce the same ideas of convexity and path dependence that are familiar from derivatives.
Perold and Sharpe showed that different rebalancing disciplines generate distinct payoff profiles relative to the market. Buy-and-hold produces a broadly linear relationship. Constant-mix—selling winners and buying losers to restore target weights—creates a concave profile that tends to benefit from oscillating markets and suffer in strongly trending ones. Constant Proportion Portfolio Insurance (CPPI), which increases exposure to risky assets as wealth rises above a floor, creates a convex profile with the opposite behavior.
Concavity, convexity, floors, and participation rates are familiar option concepts because dynamic rebalancing can create option-like payoff profiles without the investor explicitly trading an option. The derivatives literature reaches related conclusions from the opposite direction: path-dependent quantities such as maximum drawdown can themselves be studied as contingent claims.
The implication for asset allocation is important. A single-period optimization does not model rebalancing within the optimization horizon. Investors can, of course, rerun a single-period optimization periodically, but that is different from jointly modeling how today's allocation and future rebalancing decisions interact across the full horizon. That distinction matters when turnover limits, changing assumptions, cash flows, or path-dependent risk connect decisions across time.
This reframes the role of scenario-based optimization.
In derivative pricing, we can generate many possible paths for the underlying variables, compute the payoff associated with each path, and aggregate those outcomes to estimate value:
Market paths → Path-dependent payoff → Value
In asset allocation, scenarios are used similarly, but the economic question is different. For each candidate strategy, we evaluate how wealth evolves across many possible futures. Along those paths we can measure return, drawdown, probability of achieving a goal, probability of insolvency, the effect of cash flows, and other outcomes that matter to the investor. We then optimize the strategy itself:
Market scenarios → Portfolio wealth paths → Investor outcomes → Optimization
In derivatives, the payoff is given and the problem is to price it. In asset allocation, the objectives are given and the problem is to choose the policy that generates the most suitable payoff profile.
That policy might target a required return while controlling drawdown, maximize the probability of funding future withdrawals subject to turnover and liquidity limits, or determine how an allocation should evolve as assumptions change through time.
This is the key difference between simply simulating a portfolio and scenario-based optimization: the scenarios are not used only to describe what might happen. They become the environment in which the investment strategy is selected.
There is a mathematical reason richer problems require different tools. Once portfolio decisions interact through time, wealth compounds through rebalancing and objectives such as drawdown depend on the full path. The resulting optimization problem generally loses much of the convenient structure of single-period mean-variance optimization.
This echoes the evolution of derivative pricing. As products became more complex and path dependent, closed-form solutions increasingly gave way to numerical methods. Asset allocation faces a similar trade-off: a more faithful representation of the investment problem often requires more computation.
Path dependence is only one example. Capital market assumptions are routinely produced for multiple horizons, and those forecasts can differ materially—sometimes even in sign. FactSet's earlier work on dynamic asset allocation (Stepping Into Dynamic Asset Allocation) illustrates the problem using 5-, 10-, and 20-year return forecasts: a single-period investor with a 20-year horizon must compress that term structure into one set of inputs (Sui, 2022a). Long-term assumptions may ignore near-term conditions; short-term assumptions neglect the later horizon; averaging can dilute information at both ends.
A multi-period framework can instead use the full term structure and determine how the allocation should evolve as the assumptions change. This is one of the central benefits highlighted in the dynamic asset-allocation framework: weights can change period by period to reflect the information contained in the full forecast surface, while turnover constraints can smooth the transition between allocations (Sui, 2022a). The optimal portfolio today may therefore depend on where the investor expects the portfolio to move tomorrow.
The same issue arises with illiquid assets, where ramp-up and lock-up periods are naturally multi-period, and with contributions, withdrawals, and goals occurring at different dates. Dynamic allocation also makes it possible to represent implementation rules triggered by portfolio conditions—such as CPPI-style management rules—rather than treating each horizon as an isolated allocation problem (Sui, 2022a).
No more than Monte Carlo made Black-Scholes obsolete.
A vanilla European option does not require the most sophisticated numerical pricing engine available. If an analytical solution answers the question, additional machinery may add cost and opacity without adding insight. The same principle holds in portfolio construction. For a straightforward allocation problem defined by expected return, volatility, and correlation, mean-variance optimization remains effective, fast, and transparent.
The question is not whether scenario-based optimization should replace Markowitz. It is: Does this investment problem contain characteristics that a single-period model was never designed to represent?
Where the answer involves dynamic decisions, a term structure of assumptions, intermediate cash flows, illiquidity, implementation constraints, or path-dependent risks such as drawdown, a multi-period approach can represent the actual problem more faithfully. Where it does not, the simpler tool may be the better one.
The analogy should not be pushed further than it holds. Derivative pricing asks what an uncertain payoff is worth. Asset allocation asks which decisions best serve an investor's objectives under uncertainty. These remain different economic problems.
But their methodological evolution shares something important. As problems become more realistic, fewer simplifying assumptions can safely be made. Static models give way to dynamic ones. Terminal outcomes give way to full paths. And numerical methods become increasingly valuable; not because sophistication is a goal in itself, but because they allow more of the real problem to be represented directly.
The destination matters. But for long-term investors, so does the path taken to get there.
The ideas discussed here build on a sequence of FactSet research and Insight publications. Multi-Period Portfolio Optimization (2020) develops scenario-based multi-period optimization. Stepping into Dynamic Asset Allocation (2022) shows how changing weights can use the full term structure of capital market assumptions and incorporate implementation considerations such as turnover and lock-ups. What Is the Right Risk Measure for Long-Term Investors? (2022) focuses on multi-horizon and path-dependent risk. Building Long-Term Asset Allocation: A Flexible Multi-Period Simulation Framework (2023) extends the framework to decision rules, within-horizon risk measures, cash-flow planning, and goal-based investing.
FactSet Asset Allocation 3 supports traditional asset-allocation workflows alongside Multi-Period Scenario-Based Optimization, allowing investors to match the methodology to the characteristics of the investment problem. The framework supports forward-looking scenarios, dynamic allocations, changing assumptions, cash flows, implementation constraints, and within-horizon risks such as drawdown—moving the conversation beyond what portfolio looks optimal today toward what investment strategy is designed to perform appropriately across the journey ahead.
Bilarev, T., Mitov, G., Racheva-Iotova, B., and Stefanov, I. (2020). Multi-Period Portfolio Optimization. FactSet White Paper.
Bilarev, T., Mitov, G., Racheva-Iotova, B., Stefanov, I., and Sui, C. (2023). Building Long-Term Asset Allocation: A Flexible Multi-Period Simulation Framework. FactSet White Paper.
Kritzman, M. and Rich, D. (2002). The Mismeasurement of Risk. Financial Analysts Journal, 58(3), 91-99.
Markowitz, H. M. (1952). Portfolio Selection. Journal of Finance, 7(1), 77-91.
Perold, A. F. and Sharpe, W. F. (1988). Dynamic Strategies for Asset Allocation. Financial Analysts Journal, 44(1), 16-27. Reprinted in Financial Analysts Journal, 51(1), 149-160 (1995).
Sui, C. (2022a). Stepping into Dynamic Asset Allocation. FactSet Insight, March 21, 2022.
Sui, C. (2022b). What Is the Right Risk Measure for Long-Term Investors?. FactSet Insight, August 4, 2022.
Vecer, J. (2006). Maximum Drawdown and Directional Trading. Risk, 19(12), 88-92.
This blog post is for informational purposes only. The information contained in this blog post is not legal, tax, or investment advice. FactSet does not endorse or recommend any investments and assumes no liability for any consequence relating directly or indirectly to any action or inaction taken based on the information contained in this article.