Question for someone good at mathematics - page 5
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Well that's far more promising than mine ^^. I did some tests with less or more options/calculations, as ubzen stated it is question for maths, so I made test search for huge options math data and I'd put on (no ticks data) PERIOD_M1 and period of 9-10 years, But after a short start expectations were far optimistic from reality, it came out with :: " it would be needed 56-100 years to complete ", LoL. Compared speed with 18:1 vs 8 cores as https://www.mql5.com/en/forum/138211/page3, so that would be 3 to 6 years. That's about my 'perfect' intelligence program and computer ;) ^_^ Of course code can be optimized, but again, it would not go with less than 50% of that. So definitely I am moving to easier math calculations.
Well that's far more promising than mine ^^. I did some tests with less or more options/calculations, as ubzen stated it is question for maths, so I made test search for huge options math data and I'd put on (no ticks data) PERIOD_M1 and period of 9-10 years, But after a short start expectations were far optimistic from reality, it came out with :: " it would be needed 56-100 years to complete ", LoL. Compared speed with 18:1 vs 8 cores as https://www.mql5.com/en/forum/138211/page3, so that would be 3 to 6 years. That's about my 'perfect' intelligence program and computer ;) ^_^ Of course code can be optimized, but again, it would not go with less than 50% of that. So definitely I am moving to easier math calculations.
After what I've been through & seen so far, I think it's better to go the simple way. It always starts simple since complexity builds on simple rules really fast. Imo, complexity is bad for repeat future performance. If it gets complicated too fast, chances are it's garbage. Same goes with too many parameters and conditions. I've traveled the NN/GA route, it's useless with no precise aim of what to look for. The whole EA above took less than 150 lines. M1 open price optimization runs along structurally 'similar' result with tick (tick gives better result). I'm overclocking my AMD APU to achieve 2.4GHz flat clockspeed during backtest. Try to structure the EA from the start, instead of patching things all over the codes.
Flat periods which lasts up-to 3+years are just the nature of a stop-loss system...IMO. You might be better off just accepting it. You can try trading smaller lots, virtual-trading or stop-trading while its in this period. Then monitor it to see when it bounce back again. Also, if you diversify, it might be working on another pair while you wait.
I think I'd be better off taking your advice and optimizing it a bit. Then run several setup variants/agents out of the smoothest running backtest candidates. It never hurts to diversify, even on the same instrument. Thanks.
Changed some calc to adapt to volatility dynamics. Output: higher net profit to max drawdown value. Better distributed flat periods & less DD dip. Enough tweaking for the day.