curvefitter
Recruit
- Messages
- 2
Hello,
New here.
Long time ago, I was experimenting with a curvefit method to predict forex. Even then I realized it is almost impossible to try predicting forex movements with any form of technical anlaysis (I believe George Soros used no TA!). If anyone could predict forex , it would be Google with their supercomputers, data memory resources and whatever they used to come up with DeepMind which could defeat the world's top Go player (this is a remarkable achievement for those in the know). I do believe no one "yet" in the world has any computer system that could beat the forex market.
I lost a few thousand dollars and stop. My computer then was a 386 that took hours to do one curvefit. Recently, I revived my program as I am curious if I could try again to see if it could work. Now with my intel586[edit intel i5], it takes only less than a minute for me to get a perfect fit of 300 points (correlation coefficient 0.999) with four processors at full throttle. Still, the backtesting shows that the "Holy Grail" is "difficult"" to find!
The chart above shows a curvefit for gbpaud daily data:
data end: 30/sep/2022 (the green is moving average)
moving average = 52 x 6 = 312.
fit range = 10 x 312
backtesting: data end - 360 days.
predicted actual bottom : 30 Sep/2022 + 1.25 years (ma bottom less 1/2 ma) = Jan 2024.
bottom: below 1.46 (now 1.74)
My Opinion: 80% certainty based on curvefit and the pattern of the pair.
This curvefit - even if accurate - is useless for short term traders. The short term volatility would kill! OK to hold long term WITHOUT STOP LOSS, if you are a multi- millionaire !
Best regards.
Last edited: