First download your free copy of GoldenGem, also click on the data file proof.txt. It contains
sin(i/10)^3+.001, cos(i/10), sin(i/10), (sin(i/10)^3+.001)^2+ 6 cos(4 cos(i/10)), and sin(5 (sin(i/10)^3)cos(3(cos(i/10))
for i=1 to 400, and fictitious tickers x,y,z,w. Look at the screenshot for the result. It predicts them perfectly. And you can try it yourself for a very convincing proof of why GoldenGem is the one which was chosen. More details are at http://www.goldengem.co.uk
Here are the first few lines of your test data file
Just enter "x", "y", "z", "w" in the ticker window and browse to the file.
You can also, of course, enter real tickers and press 'Load from Internet' and GoldenGem will load with real stock data of your choice, and IF there is a mathematical relationship, GoldenGem is most likely to find it.
The number of days to look ahead is selected on a horizontal slider, the sensitivity on a vertical slider. To predict these math functions, we left the sensitivity slider in the default position, but as you see it took dozens of iterations to train GoldenGem.
The green trace is the neural net's attempt to match the blue curve, but using all loaded prices and volumes from the earlier time indicated by the red trace. Raise the sensitivity until the green and blue curves begin to match, and leave it running.
After a while, you can lower the sensitivity. If the green curve still matches the blue curve, this means the neural net can match the advanced blue curve based on loaded prices and volumes from the earlier time. It can continue this calculation for an interval of time an equal number of days into the future, and therefore give an opinion about the selected ticker at that time.
Further information about this program is at http://www.goldengem.co.uk
Click here to download Proof that GoldenGem works
Rating: 2.0/5 (6 votes cast)
BBCode is enabled .