Neuro-forecasting of financial series (based on one article)
Good afternoon!
Subj: http://etd.ohiolink.edu/send-pdf.cgi/Lakshminarayanan%20Sriram.pdf?ohiou1127333497&dl=y
In this article, the researcher achieved a daily closing price prediction accuracy of around 94% for exchange-traded instruments. His test sample size was: 158 days.
He built a predictive neural network, fed a number of standard technical indicators as inputs, then made pruning redundant inputs. As he writes, prediction accuracy rose from 57% to 94% by using an Elliott Waves indicator which he applied to the data in some devilishly devious way, using fuzzy logic.
I suggest you read on and think about the validity of the Indian researcher's results.
There seems to be a lot of water in there. Where do you start reading from?
You could start on page 69. Before that he reaches an accuracy of 57% (not very interesting, could be pure chance). And then using Elliott Waves raises the forecast to 94%.
94%? It's not Indian, it looks like a British scientist.)
)) In fact, I'll tell you more: American.
94%? It's not Indian, it's a British scientist.)
He's not a scientist at all, he's just a student... He's got a lot on his plate... 94% in 2005 and we still don't see him on the Forbes list?
Actually, the pruff is there. The question is how trustworthy it is :)
There are also big doubts about the reality of the figures, because all the charts are clearly skewed.
In this article, the researcher achieved an accuracy of 94% in predicting daily closing prices for exchange-traded instruments.
94% is too small a percentage, indeed, unless it is a student or high school one. Adult starts somewhere between 155-160%, not to mention professor with its lower bound around 300%.
94% is too small a percentage.
Not funny. Pruf is on page 3.
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
Good afternoon!
Subj: http://etd.ohiolink.edu/send-pdf.cgi/Lakshminarayanan%20Sriram.pdf?ohiou1127333497&dl=y
In this article, the researcher achieved a daily closing price prediction accuracy of around 94% for exchange-traded instruments. His test sample size was: 158 days.
He built a predictive neural network, fed a number of standard technical indicators as inputs, then made pruning redundant inputs. As he writes, prediction accuracy rose from 57% to 94% by using an Elliott Waves indicator which he applied to the data in some devilishly devious way, using fuzzy logic.
I suggest you read on and think about the validity of the Indian researcher's results.