Optimize Indicators ?
What tpye of optimization.
Speed-Optimization?
Results-Optimization?
Realtime-Self Optimization?
Hi Ubzen,
The later two: Results in increased accuracy and resulting improved profitability such as is done with the Strategy Tester.
Eventually I want to start including accuracy and profitability self Optimization in both EAs & Indicators when they are running.
Yep you got the first part right with the use of Strategy Tester. For the later part, try taking a look at this.
Hi Ubzen,
The first one is more of an immediate concern with the later one later when I'm more proficient.
How could I do Optimization on indicators with the strategy tester? How can I access them as they are not included in the EA nor are their any external variables available to access by the strategy tester. I can make the indicator's variables external. One thought that comes to me is to actually temporarily include the indicators into the EA itself to access these to Optimize otherwise the Indicator Variables would not be accessible to the Optimizer would they? Have something else in mind. If so, what ?
Regards
" One thought that comes to me is to actually temporarily include the indicators into the EA itself to access these to Optimize "
Yeah, thats the only way I can think of too. Another option is building a virtual_visual optimizer but for now this goes beyond my programming capabilities and it'll be similar to programming your own visual_tester. I don't have any links or tutorial on this, it's just a concept I've taught of when thinking about multi-currency back-testing.
Yes, optimize the EA.
Optimizing indicators is meaningless.
Don't need to include the indicators into the EA. Add the indicator's externs to the EA, and pass them to the indicator in the iCustom. Optimize the EA.
Good Idea WH. Thanks, I'll give this a go and it will be my first experience with iCustom. Though this will be limited to public domain EAs and I have quite a few commercial ones that frankly for the most part anything would be an improvement in them. I don't think that I can encapsulate an existing compiled EA into another EA with the iCustom functions in them. If the indicators are public domain with the source code available, I still don't see how I can get the info into the compiled EAs? Not sure if I could utilize proxy buffers or ??? etc and somehow redirect the original EA to take the optimizing altering settings. I wouldn't know what the compiled EA is using for descriptors, variables and functions etc. Also some EAs suppress and hide the indicators that it is actually using. Any ideas and insights into these things ?
Yes, optimize the EA.
Optimizing indicators is meaningless.
Hardly.
Indicators ARE by and large what we use as entrance signals; and often to a lesser extend, productive and profitable exit strategies and points as well. Finding the right combination of indicators working well together with optimal settings for these that shifts the balance of entering into many more winning vs losing trades has very significant chance of improving things substantially.
Getting into the highest possible percentage of winning trades is VITAL to (long term) profitability and robustness. Most EA have a VERY low accuracy rate. One can have the best EA and exit strategy in the world, but if your EA is getting into mostly losing trades to start with, then it is a loser. Period. Trying to optimize such EAs and utilize filtering it can improve the profitability somewhat, but it is never going to amount to much and be good, profitable and robust EA over any significant period of time. How can it if it has mostly losers to start with? Many try to make this work by setting the SL up to 25 TIMES higher than the TP ! This is just a nightmare that is going to increase losses even more. In my mind, such EAs, and their are many of them including most of the commercial ones I have bought, all do this; do not even merit being called an EA at ll. Any trading with this type of system, if these can even really be classified as a legitimate 'FX trading system and strategy' at all. is little more than guessing and do not have good enough entrance strategies that will ever be profitable over the long haul, if at all in the first place. With such high SL/TP ratios, one such HUGE bad trade can wipe out the profits from dozens or even hundreds of profitable trades and then some and eat into your account balance substantially more than all of the winning trades combined than this.
I have one such EA that for a period of several months in a row had a 100% profitable trades. It is supposed to trade on 3 different pairs. The problems: it only makes 5 ~ trades a month on one pair and like most EA has the SL set WAY higher than the TP. Eventually one losing trade wiped out many months of 100% profitable trades and then some by a wide margin.
Thus it would seem that one has to have everything and ALL aspects of the EA working very well and productively or it is all for naught. It has to be the 'whole package' or it's just GIGO. All to many ForEx EA traders are unfortuantely all to familiar with these things, especially trading with 'static' EAs. I rather suspect that this is the 'norm' and the converse is very rare and of very small proportions indeed.
The big problems with this is adding yet more variables to optimize. As it is these are very time consuming and resource hogs. Anyone have unlimited access to a Mini, MainFrame or Super Computer or clouds of many computers available? (< 8) It would be great to have a BOINC project to do these to access a LOT more computing power. But as this is not their intended purpose and application this is only a whimsical dream.
Tip: The optimizer is unable to manipulate Boolean TRUE or FALSE settings. I suggest using a 'translation' function as a work around for this by making a 'bi_Boolean_T_or_F' function where 'bi' stands for 'Boolean Integer'. This will translate integers of 0 into a false or a 1 into a true for the Boolean logic and assign it to the Boolean. In this manner these can be optimized as well.
One profitable manual technical trading system that I was tutored in used ate least 8 or 9 indicators all at once at all times, mostly public ones, and even then these required adjustments for various market conditions and still called for human judgement and interpretation based on confidence through KNOWLEDGE and that builds from successful experience. Though it was and is consistent enough to be profitable in the long haul IF one knows and understands it and all of the dynamics in play and applies it properly and consistently. In the early days of this extensive, intensive and very expensive tutoring I quite often 'got my knuckles raped' and was often admonished by my frustrated tutor that I was 'just guessing yet 'AGAIN!'' For the first several months I had a perfect record of almost 100%! WRONG! LoL )< 8( I often said to my tutor over these early months that if I took my 'conclusions' and did the exact opposite I would do just great )< 8) I also got admonished for using a safety net of StopLosses in my practice trading with this system. In spite of being discouraged and with perseverance I later tracked 3 months of consecutive trades and analysis in Excel at the end of this period showed that I had ~ 80% accuracy rate over these 3 months. No one was more (pleasantly) surprised than I was. (< 8) Also the system incorporates proper and balanced TP & SL that resulted profitability if one was trading it properly. These levels were also ascertained by indicators. Including the usage of many simple, manually placed Support and Resistance levels on top of sometimes over a dozen other indicators that one that to know, understand and including the influences, implications of these and their interactions and incorporate them all properly and correctly in order to achieve the desired results. Programming an automated trading system such as this would be a signifcant challenge indeed and well beyond my current quite limited capabilities.
Yep you got the first part right with the use of Strategy Tester. For the later part, try taking a look at this.
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
Is it possible to Optimize Indicators?
If so, how ?