We continue working on improvement of Artificial
Foreteller for better finance prognostics.
To increase data processing speed we are working
on splitting process of forecasting into two parallel processes:
- indicator and other
criteria calculations based on tick data flow;
- prediction of
future trade period.
We also working on API
interface to feed Artificial Foreteller with data flow through Internet.
We also want to improve training process. For that
we want to implement additional category of train data. Periods with small
gain or small loss (periods with not very well defined trends) we will mark
as "neutral". In that case Associative self-building neural
network will be trained on positive, negative, and neutral trade periods.
A demo version of Artificial Foreteller is
available at Test
Result. and full version is at Products.
We are on the final stage of development of two
new systems that will be used for prediction of oil and mineral deposits
and solar flare prediction. We are going to finish these systems to the mid
of 2014. Each stage of development process will be provided for discussion.
Interested parties can contact us at:e-maiL
|
