![Show Menu](styles/mobile-menu.png)
![Page Background](./../common/page-substrates/page0059.png)
INFORMS Nashville – 2016
57
SB41
207C-MCC
Spectral Methods in Finance: Option Pricing
and Econometrics
Sponsored: Financial Services
Sponsored Session
Chair: Lingfei Li, Assistant Professor, The Chinese University of Hong
Kong, 608 William M.W.Mong Engineerng Building, Shatin, N.T., none,
Hong Kong,
lfli@se.cuhk.edu.hk1 - Error Analysis Of Finite Difference And Markov Chain
Approximations For Option Pricing With Non-smooth Payoffs
Gongqiu Zhang, PhD Candidate, The Chinese University of Hong
Kong, ERB 802, Hong Kong,
gqzhang@se.cuhk.edu.hk,Lingfei Li
We provide error analysis for finite difference and Markov Chain approximations
in option pricing when the payoff function is non-smooth. We assume the asset
price is a one-dimensional diffusion or a jump process constructed from a
diffusion by subordination. We show that the spatial discretization error is second
order for call and put-type payoffs and first order for digital-type payoffs.
Furthermore, averaging discontinuous payoffs can restore second order
convergence.
2 - Long Forward Probabilities, Recovery And The Term Structure Of
Bond Risk Premiums
Likuan Qin, Northwestern University, Evanston, IL, 60016,
United States,
likuan.qin@gmail.com,Vadim Linetsky
We show that the martingale component in the long-term factorization of the
stochastic discount factor due to Alvarez and Jermann (2005) and Hansen and
Scheinkman (2009) is highly volatile, produces a downward-sloping term
structure of bond Sharpe ratios, and implies that the long bond is far from growth
optimality. In contrast, the long forward probabilities forecast an upward sloping
term structure of bond Sharpe ratios and implies that the long bond is growth
optimal. Thus, transition independence and degeneracy of the martingale
component are implausible assumptions in the bond market.
3 - Parametric Inference Of Discretely Observed
Subordinate Diffusions
Weiwei Guo, The Chinese University of Hong Kong,
wwguo@se.cuhk.edu.hkWe develop a two-step procedure to estimate a class of jump processes known as
subordinate diffusions from discrete-time data. The first step identifies the
diffusion parameters using estimating functions only involve diffusion parameters
and the second step identifies the subordinator parameters using martingale
estimating functions based on eigenfunctions. Under regularity conditions, our
estimators are consistent and asymptotically normal. Numerical examples show
that our method is statistically and computationally efficient. Analysis of VIX
index indicates that a pure jump subordinate diffusion model significantly
outperforms diffusion models in modeling the VIX index.
SB42
207D-MCC
Quantitative Methods in Finance XIII
Sponsored: Financial Services
Sponsored Session
Chair: Changle Lin, Princeton University, Jersey City, NJ,
United States,
changlel@princeton.eduCo-Chair: Woo Chang Kim, Associate Professor, KAIST, KAIST 291,
Daehak-ro, Yuseong-gu, Daejeon, 34141, Korea, Republic of,
wkim@kaist.ac.kr1 - Goal-based Investment Via Multi-stage Stochastic Programming
For Robo Advisor Service
Woo Chang Kim, KAIST,
wkim@kaist.ac.krRobo-advisors aim to attract non high-net-worth individual investors by signifi-
cantly lowering the entry barrier to professional wealth management industry.
Unfortunately, existing schemes of robo-advisors have not been sophisticated
enough to provide fully personalized investment advices. However, it surely is
challenging to ask clients, who might lack financial literacy, to provide their
detailed financial situation through online platform. Therefore, we propose a
goal-based investment model that only requires the input of wealth, income, and
consumption goals with priorities via multi-stage stochastic programming
approach.
2 - How to Train Your Lawyer
Ephrat Bitton, Future Advisor,
ephratb@gmail.comFor better or worse, Robo Advisors operate in a highly regulated industry. Folks
may roll their eyes at the mention of compliance, but it is a crucial process for
ensuring that we protect the end client. As a mathematician at FutureAdvisor,
one of my greater challenges is adequately describing to our compliance officer
how we manage portfolios using optimization. MILP can be immensely powerful
for solving complex decision problems, but it is notoriously difficult to pinpoint
the reasons for different outcomes. This talk follows my story on automating
portfolio management, ensuring the quality of our results, and finally, explaining
how it all works to someone who reads legal settlements for fun.
3 - Pricing And Hedging Guaranteed Minimum Withdrawal Benefit
With High Water Mark Benefit Base
Peiqi Wang, Princeton University, Princeton, NJ, United States,
peiqiw@princeton.edu, Patrick Cheridito
We consider pricing and hedging of Guaranteed Minimum Withdrawal Benefit
(GMWB) rider on a variable annuity (VA) contract. We price the VA+GMWB
contract by considering the optimal withdrawal strategy of the policyholder. We
show that policyholder’s payoff resulting from the optimal withdrawal strategy
corresponds to the super-replication cost of the contract and we provide a
hedging strategy. Our numerical results show that it is sometimes optimal for the
policyholder to aggressively withdraw and ruin the account. Further analysis on
the numerical results suggests how the insurer should determine the fee structure
and minimal deposit requirement.
4 - About Holistic Robo-advice Engine
Dan Dibartolomeo, Northfield Information,
dan@northinfo.comRobo-advisors aim to attract non high-net-worth individual investors by
significantly lowering the entry barrier to professional wealth management
industry. Unfortunately, existing schemes of robo-advisors have not been
sophisticated enough to provide fully personalized investment advices. However,
it surely is challenging to ask clients, who might lack financial literacy, to provide
their detailed financial situation through online platform. Therefore, we propose a
goal-based investment model that only requires the input of wealth, income, and
consumption goals with priorities via multi-stage stochastic programming
approach.
SB43
208A-MCC
Systems Engineering and Decision Analysis
Sponsored: Decision Analysis
Sponsored Session
Chair: Robert F Bordley, PMP, MBA, Booz Allen Hamilton, Troy, MI,
United States,
bordley_robert@bah.com1 - Value-focused Thinking For Engineering Resilient Systems
Greg Parnell, University of Arkansas,
gparnell@uark.eduDoD’s requirements analysis identifies Key Performance Parameters to meet the
system goals. The acquisition documents identify the thresholds and objectives for
each parameter that are supported by mission analysis that considers mission
needs, technical maturity, affordability, and schedule. Multiple objective decision
analysis and Value-Focused Thinking can provide a mathematical framework for
evaluating the resilience of systems in mission scenarios under uncertainty and
the adaptability of the platform to future missions.
2 - The Systems Engineering Approach To Setting Design Targets
Robert Bordley, Booz Allen Hamilton,
bordley_robert@bah.comSystems engineering is a value-focused process aimed at defining feasible
component-level design targets for a system which, when designed and
assembled, will best meet the needs of system stakeholders. To reach this goal,
systems engineering first defines targets at the system level, then at the subsystem
level, the assembly level etc down to the component level. At each level,
informed trade-offs are made about what is most appealing to stakeholders given
beliefs about what is technically possible. Making these tradeoffs involves
specifying alternative solutions, investigating each solution and constructing an
optimal hybrid of the solutions.
3 - A Bayesian Method For Selecting Elite Varieties Of Soybean
Jack Kloeber, Kromite, LLC, 82 Nelson Drive, Churchville, PA,
18966, United States,
jkloeber@kromite.com, Joseph Byrum,
Tracy Doubler, Greg Doonan, Craig Davis, Peiran Zhao
In agriculture R&D, a new variety’s genetic contribution to higher yield is difficult
to separate from factors of soil, insects, weather, or agronomic practices. Varieties
are grown at multiple locations, downselecting over 4 years. Syngenta developed
a generalizable method which helps soy breeders find the genetic winner using
Bayesian Updating. The increased accuracy leads to better decision-making and
higher yield.
SB43