[R-sig-finance] Robust Optimisation - Evening Discussion

[John Marsland]

Any R Finance people based in London might be interested to come along to this
meeting on Thursday this week. There is a small charge for attendance, but
non-member of UKSIP/CFA Institute are most welcome.

Regards,

John Marsland

---------- Forwarded message ----------
From: uksipmail at uksip.org < uksipmail at uksip.org>
Date: Feb 20, 2006 12:44 PM
Subject: Robust Optimisation - Evening Discussion
To: john.marsland at cantab.net

- Hide quoted text -
Robust optimisation - greater certainty through uncertainty?
Quips:
The special interest group for quantitative investment professionals
Thursday 23 February 2006

Speakers: Daniel Mostovoy, Head of
Research at BITA Risk
Professor Steven Satchell, Trinity College,
Cambridge

Robust optimisation tackles the question of how to
achieve a stable portfolio output from unstable or uncertain inputs.
The traditional approach to optimisation - construction of an exact
optimal portfolio - expects that there are exact inputs and can be
highly sensitive to changes in them. Inputs include expected returns,
expected risk, expected transaction costs, risk aversion coefficient,
cost aversion coefficient. These inputs are all estimations and the
general target is to improve the degree of certainty around these so
as to produce a stable result. One approach proposed has been
re-sampling, but this can suffer from un-repeatable results. Another
approach is to accept that there is uncertainty on certain inputs, but
to include that in the optimisation, hence robust optimisation. Our
speakers will present on the premises, solutions and practical
applications of robust optimisation.

To register for this evening
discussion click
(http://www.uksip.org/calendar/event_details.cfm?iEventID=87) for an
application form.

UK Society of Investment Professionals
4th
Floor, 90 Basinghall Street, London EC2V 5AY
T el: 020 7776 7519
-----------------------------------------------
Notice: If received
in error, please notify sender. UKSIP does not waive confidentiality
or privilege, and use is prohibited.