Re: Portfolio Theory

Robin B. Lake (rbl@hal.EPBI.CWRU.Edu)
Sat, 28 Mar 1998 23:30:06 +0100 (MET)

It was asked on this list:

>"I am working as a financial markets analyst and I would be interested in
using fuzzy theory in portfolio selection and optimizing. Any experiences
or ideas, suggested reading etc.?

Thank You

Tommi Salminen
Merita Bank
+358 9 1655 9038
"
=====

I tried to reply to Mr. Salminen privately, but the reply address came out
as tommi@dbai.tuwien.ac.at and the message was refused.

I have just finished a portfolio price prediction system, using fuzzy
concepts. There was no book or reading behind this, just 25 years of
working with fuzziness and someone to ask the question "Would this
work on the stock market?"

The answer appears to be "Yes, indeed!" After debugging the system for
several months, I put it into action 7 trading days ago. My gain is
about 12.5% in that time and I've made about 10 good trades and only
one minor losing one.

The system does the following:

- Based on fuzzy matching, it provides an "overnight" model whose result
is a minute-by-minute price tick-chart for the next trading day. This is
for the portfolio I'm testing, the Standard & Poors 100.

- Then, during the trading day, the System runs models on three portfolios
every 15 minutes. These models quantitatively predict the change in
portfolio prices 60-minutes and 90-minutes in the future. On the basis
of these models and the price charts from the "overnight" model, I trade
index options.

This posting is not at all intended as a commercial for this system. It
is to encourage others to explore the opportunities for fuzzy applications
in financial markets.

It also seems to disprove the theory that public markets are "efficient",
in that everyone has the same information at the same time. They don't
or I couldn't do this.

Cheers,
Robin Lake, Ph.D.
Environmental Modeling Inc.
rbl@po.cwru.edu