Financial Applications of Neural Networks

1
Financial Applications of Neural Networks

• Lecture 3
• Some General Principles and a STT Analysis Example

2
Trading/Investing

• Long term (investing)
• Medium term (investing/trading)
• Short term (trading)
• No-nos
• Fees
• Commissions
• Slippage
• Taxes

3
Investing Strategies

• Know the primary trend of the marketIf the
  market or a stock is in a downtrend, it will be
  difficult to make money.
• Diversity. Consider at least 3-4 different areas.
• Cut losses early. Use mental stops rather than
  automatic stop losses.

4
Investing Strategies

• Let profits run. If a stock rises, sell if it
  falls back a predetermined percentage.
• Don t be emotional about about investments. Set
  stop points and stick to them.

5
STT Analysis Example
The following picture shows a graph of AMAT
(Applied Materials Corporation) taken from a
typical quote screen. This graph is from
Quote.com. The graph shows price for the period,
as well as three technical indicators which are
useful in predicting price movement in the short
term (days to weeks). The technical indicators
are OBV, MACD and momentum, and are standard
indicators used by most analysts. Definitions and
a discussion of the use of these indicators can
be found in many places. See, e.g.,
www.bigcharts.com. These data are some of the
inputs for the computer analysis we use in our
automatic data analysis programs using neural
networks and fuzzy sets. Take a look at the graph
and then go to the following slide.
6
STT Analysis Example
7
AMAT Graph Discussion
Lets focus on the price movement at the end of
August. The price weakness is signaled by the
downward crossing of the fast (blue) moving
average and the slow (red) moving average in the
MACD graph. It is also signaled by the drop in
the momentum graph, although this signal is a bit
late. However the MACD signal occurred early
enough to avoid the subsequent loss in the value
of the stock. Because AMAT is a strong company
it might have been prudent to buy it back at the
following MACD crossing in early October. But
remember, in early October we could not know that
there was not going to be another terrorist
attack. In this case, strict adherence to a MACD
crossover trading policy would have been OK, but
notice how current events can interfere with
automatic trading rules.
8
AMAT Graph Discussion
Lets consider the timing issue more closely.
The graph shows that the downward crossover
occurs after a peak, and at that time you might
have thought that the stock would not drop
further. This behavior is typical. One of the
most interesting challenges of automatic trading
programs is to attempt to factor other factors
into the analysis, in an attempt to anticipate
price movement. In fact, this will be one of the
major design goals in our work in this course.