Our motivation was striking, but, soon we found that, people have been trying do this all along, and a lot of money rolls in this area. To predict whether there is going to be huge selling or buying tommrrow, based on the economic, political, climatic, geographical and many more intricate details. The Dow, or the Nasdaq, predict the market's orientation on by adding the "bull" and "bear" definitions, and has been an excellent predictor in the 20th centuary.
The "Stock Predictor" softwares are an advanced stock charting and investment strategy performance analysis software for financial markets information. It allows to display several technical indicators for a single security on the same chart, maintains predefined lists of securities and to test our own investment strategies. Stock Predictor has over a dozen built-in technical indicators and over four hundred combined investment strategies. One-click buy and sell feature gives you an unique opportunity to apply a trading strategy to a given security and view a cumulative and summarized performance of your investment for a selected period of time.
To add more the work that is being done, the "Neural Stock Predictor"
is more interesting. It calculates projected price motion for a stable
set of company stocks,
for the next trading day. These predictions are based upon a totally
autonomous neural networt system patented by IEI, the Self-Training Artificial
Neural Network Object (STANNO). Projections are updated each trading day
at 6:00 PM EST.
Although great strides have been made in the area of neural networks,
these computational devices remain passive, simply serving in a reactive
learning role. The newly patented "Creativity Machine Paradigm", in contrast,
provides a means to make these systems proactive, in effect granting them
the free will and the flexibility to escape the knowledge contained within
their learning and to produce novel concepts and courses of action. The
effect these virtual machines are based upon is exceedingly simple and
straightforwardly controllable: A normal neural network that has been exposed
to any knowledge domain and then repeatedly subjected to mild internal
disturbances, tends to produce a mixture of both intact memories and unusual
juxtapositions of those memories that are unprecedented in the net’s experience.
In effect, the network is mildly hallucinating; manufacturing novelties
derived from its own unique microcosm. By adding a second ‘critic’ network,
operating in the standard role of pattern recognition, any emerging notions
offering potential value or utility may be extracted, archived, or utilized
in real time. This second net may likewise serve to focus the ‘dreaming’
net, via feedback connections, to accelerate the search for promising concepts.
The ensuing internal dialog between these two neural networks may exceed
mere computation to produce undeniably appealing works of art and striking
emotional statements. It is our prediction that the Creativity Machine
Paradigm will have immense impact not only in the building of a truly creative
artificial intelligence, but in our
understanding of how the human brain arrives at seminal discovery, invention,
and artistic achievement.
http://www.imagination-engines.com/CyberStocks/CyberStocks.htm
These are some of the existing systems, and the new revoltionary ideas entering the market. Though it would be hard to compete with these products in terms of quality or the content wise, we have made a very good stand here, by collecting the data for various years from the Nasdaq, and running them through the classifiers we have learnt in the class and also tried our best, within the given time, to experiment on the variours features with this, and drawn our conclusion based on this. Following this we would like to show our work, and then the inference drawn from this, and also how it is appicable to real world scenario.