Optimal Portfolios Under Worst-Case Scenarios Jit Seng Chen January 28, 2016 Optimal Portfolios Under Worst-Case Scenarios • Joint work with Carole Bernard (Grenoble School of Management) and Steven Vanduffel (Vrije Universiteit Brussel) • Optimal portfolios under worst-case scenarios. Quantitative Finance, 14(4):657-671, 2014 2 Stewardship of Finance 3 Outline 1. 2. 3. 4. 5. 6. 7. Motivation Setting Traditional Diversification Strategies Optimal Tail Diversification Examples Conclusion Q&A 4 1. Motivation Standard Portfolio Theories • Expected Utility Theory (EUT) - dominant theory for making decisions under risk but inconsistent with observed behaviour • Many alternatives have been proposed to capture systematic behavioural departures from EUT • Most portfolio theories assume that investors only care about the terminal distribution of wealth and prefer more to less 6 Law-Invariant and Increasing 1 𝑋 =� 0 1 0 𝑋 =� 1 2 50% 𝑐𝑐𝑐𝑐𝑐𝑐 𝑜𝑜 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 50% 𝑐𝑐𝑐𝑐𝑐𝑐 𝑜𝑜 𝑛𝑛 − 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 50% 𝑐𝑐𝑐𝑐𝑐𝑐 𝑜𝑜 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 50% 𝑐𝑐𝑐𝑐𝑐𝑐 𝑜𝑜 𝑛𝑛 − 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 Expected utility of contract 1 = Expected utility of contract 2 7 Standard Portfolio Theories • For an investor with a fixed horizon without intermediate consumption, optimal portfolios are positively correlated with a diversified market portfolio • No downside protection • Seek security in specific states of the economy 8 SP/A Theory • Security-Potential/Aspiration • Maximize expected consumption subject to budget constraint and minimum probability that consumption meets an aspiration level • Explains the use of capital guaranteed products, CPPI strategies, and other structured products • Optimal portfolios still positively correlated with the market i.e. perform poorly in a crisis 9 Framework • Consider investors who target some distribution for their final wealth and seek security in specific states of the economy • Crisis → market index takes values in the worst 5% of its distribution • Allow investors to specify the tail dependence of strategies with the market under crisis regimes 10 Contribution • Construct strategy that • Delivers a prescribed wealth distribution • Preserves a desired dependence with the market during a crisis • Comes at the cheapest possible price 11 2. Setting Financial Market • Classical multi-dimensional Black-Scholes with deterministic risk free rate • Complete, frictionless and arbitrage free • Variance-covariance matrix is positive definite 13 Growth Optimal Portfolio • Constant-mix portfolios with in the risk-free asset in and invested • Value of constant-mix portfolio is where is normal • The growth optimal portfolio (GOP) is a constant-mix strategy that maximizes the expected growth rate with proportions 14 Growth Optimal Portfolio • The price of a pay-off equals the average of all its possible outcomes expressed in units of the GOP i.e. GOP appears as a numeraire portfolio • Define a market crisis as the event where 15 3. Traditional Diversification Strategies Buy-and-Hold • Initial amount is the bank account and , where is invested in is invested in the ith stock • No rebalancing. • Final wealth 17 Constant Mix • Need dynamic rebalancing to preserve initial allocation • Initial investment of , at maturity, grows to • GOP is a constant-mix portfolio 18 Performance During a Crisis • An optimal strategy must be the cheapest possible one (costefficient) that provides some distribution F for terminal wealth • A strategy with distribution F is cost efficient if and only if • Cost-efficient strategies are monotonic in the GOP → they provide their worst outcomes in downturns 19 Performance During a Crisis • Consider two payoffs, and with the same distribution F, but only the former is non-decreasing in the GOP • Can be shown that all investors who prefer more to less will choose I over • Investors who seek value when GOP is low tend to prefer the latter payoff 20 4. Optimal Tail Diversification Optimal Tail Diversification • Cheapest strategies that exhibit a fixed distribution and satisfy some constraints are called “constrained-cost efficient” • Provides the cheapest payoff for an investor who wants to achieve a fixed distribution F for her final wealth at investment horizon T, as well as a desired interaction with the market under a crisis regime • Key result in this paper 22 Definitions • Let F be the investor’s target distribution of terminal wealth • Investor chooses joint distribution of final wealth with the market when there is a crisis • This constraint can be expressed using a copula 23 Result • • is any random variable with distributed. , , and continuously can be calculated explicitly 24 Result • Construction of constrained optimal strategy requires another source of uncertainty • • is the cheapest strategy with distribution F that verifies the constraints on the dependency in the tail is not unique, depends largely on the choice of 25 Applications • Can find explicit expression for the optimal strategy using independence, Gaussian, Frank, and Clayton copulas • Can identify with another asset in the market at maturity (path-independent), or the value of the GOP at an earlier time t<T (path dependent) 26 5. Examples Parameters • Some numerical results in a two-dimensional Black-Scholes market • μ1 = 0.07, σ1 = 0.20; μ1 = 0.08, σ1 = 0.30; ρ12 = 0.25; α = 0.05; r = 0.05; T = 1 28 Standard Diversification Strategies • Strategy 1: Invest fully in the GOP • Strategy 2: Equal proportions in risky assets and risk free account, no rebalancing • Strategy 3: Equal proportions in risky assets and risk free account, with rebalancing 29 Optimal Diversification Strategies • Same distribution as GOP at maturity, different tail dependence during crisis • Strategy 4: Independent tail • Strategy 5: Gaussian tail with negative dependence • Strategy 6: Clayton tail with negative dependence 30 Strategies 1-3 vs GOP 31 Strategies 4-6 vs GOP 32 Measures • Cost • Sharpe ratio • Conditional probabilities related to the following events 33 Cost and Sharpe Ratio Strategy Cost Sharpe 1 – Invest fully in GOP 100 0.125 2 – Buy-and-hold 100 0.121 3 – Constant-mix 100 0.123 4 – Independent tail 100.30 0.103 5 – Gaussian tail 100.56 0.083 6 – Clayton tail 100.41 0.095 34 Conditional Probabilities Strategy P (A|C) P (B|C) 1 – Invest fully in GOP 1.00 0.19 2 – Buy-and-hold 1.00 0.16 3 – Constant-mix 1.00 0.27 4 – Independent tail 0.48 0.01 5 – Gaussian tail 0.12 0.000 6 – Clayton tail 0.24 0.000 35 6. Conclusion Conclusion • Optimal strategies that are non-decreasing in the bear market incur their worst outcomes in bear markets • Construct the cheapest possible strategy that • Delivers a given wealth distribution • Preserves a desired tail dependence with the market during a crisis • Trade-off between performance vs protection 37 7. Q & A
© Copyright 2025 Paperzz