1 (edited by bannedUser 2015-11-13 08:45:48)

Topic: Request for a "mini" optimiser

Hi Popov,

Is there any possibility to create a "mini" optimizer within the strategy development process or after the strategy is done.

This mini optimizer is to search the best results given a set of parameters by changing just the smoothing, signal method and base price.

Currently, I spent tremendous time, manually "optimizing" by running through all the different combination between the smoothing, signal method and base price.  It's very time consuming.  In fact, most of my time is consumed by this process of manually optimizing the 3 combinations.  My strategy effort would have been effortless if I have this "mini optimizer" because most of the time I already know what parameters I want to use.

On my MT4 chart, I use different indicators and parameters (time period) and when I translate it to FSB, I don't know which smoothing, signal and base price yield the best results.  But I don't want to use the FSB optimizer because sometimes it either over optimized (to the point of being "useless") or the parameters were changed in such a way when it becomes "theoretically unsound" eg. opposite direction parameters (if you know what I mean).  I feel that the FSB current optimizer is for those who are figuring out what parameters to use but not very helpful for me who already know what parameters to use (it's just my personal opinion, definitely there are others who think or feel otherwise and I respect that, if it works and serve their purpose that's good.  Whereas for me the current optimizer isn't what I need but I really need the "mini" optimizer).

All I want is to stay as true to my theoretical framework by optimizing the smoothing, signal and price without changing the "structure" of my theory i.e. parameter.  Is it possible to have a button next to the parameters whereby with 1 click it would automatically calculate the highest profit combination for the smoothing, signal and base price?  Each time I add a opening or closing condition, all I need to do is to input the parameters, time period and then click the button and the optimized combination (smoothing, signal and base price) will be set?

HIT

2 (edited by bannedUser 2015-11-13 17:26:20)

Re: Request for a "mini" optimiser

Hi Popov,

This is a very useful feature that all FSB pro users will find it handy and would appreciate and needed it regardless of what strategies they use and whether they are optimizing it or not.

This is also an angle to view optimization from another perspective.  It's like someone already know which radio channel he wants to tune to and hence a digital tuner would be good to fine tune the raido signal for better reception comparing with someone who has no idea which channels he wants or avaliable and hence use the current optimizer to help him search the category of music he likes base on certain criterias etc.  These are two different perspective of optimizing. By having both these optimizer, FSB is able to cater to two different category of traders.  And for experience traders who know what they want...the mini optimizer comes handy but such a great time/life saver.

Do consider this option please because anyone who is in the process of using their own theoretical framework wouldn't want to use the current optimizer to re invent the wheels.  But rather hope to use optimizer to fine tune what he already has in mind.  This mini optimizer is much "simple" compared to the current one but yet straight to the target of what the user need.  All the parameters are determined just need to optimize the smoothing, signal and base price to get the optimal results.

HIT

Re: Request for a "mini" optimiser

In my opinion (which respectfully may defer from your experience or knowledge), why most optimised strategies failed the Montecarlo test and real live trading are:

1.  most but not necessarily all, of the strategies generated or optimised are theoretically unsound, and hence lack consistent results in one currency, how much more improbable it is to work in multi currencies.  If you examine some of the strategies optimised,  you may get opposing directional parameters.  Indicators favour buy/long options and at the same time within the opening conditions,  there are conditions favour short option and therefore such contradictions are theoretical unsound in my opinion.   Sure type of curving fitting conditions are fundamentally wrong. And in the process of optimization,  I've no control to prevent such contradictions from happening.

I would prefer to work on my existing parameters and optimised from there using the mini optimization I suggested ie optimise the smooth, signal and Base price.  In this way I can focus on the EA I'm working at and fine tune them instead of churning endless thousands of strategies that are fundamentally unsound or that  I have no control over the parameters.  I don't want to keep re-inventing the wheels when I already know what parameters I want.  My needs arise when I put these parameters together, I don't know which smoothing, signal or base price to use in order to optimise the results

2. Secondly even if the strategy were to pass Montecar or any test, as long as we are using any time frame above 5 min or 15min, we can forget developing any scalper mode or intra day trading EA.  The "Previous Bar Value" restricted your EA responsiveness to react fast to market changes.  Secondly, it is fundamentally unsound in term's of trading behaviour/style, who trade in such manner? All EA conditions met but don't execute till bar close or open which can range from 1 min of wait or 4 hours of wait.  I'll faint if someone is using weekly time chart, imagine indicators fulfilled but wait till end of the week to enter position!  That's sound absurd to me and fundamentally wrong even though it was meant to achieve statistical stability.  Is such trading behaviour sound?  But that's how FSB EA works, it wait till the next bar open/close and your waiting time depends on which time chart you are using. In order to eliminate any "false" signal, instead of waiting for the end of bar to confirm the crossing etc, I would rather put the challenge on finding the right indicators to eliminate such multiple and yet insignificant price changes than to compromise on my trading responsive with such delay imposed by "previous bar value". PBV

If the removal of PBV doesn't yield any statistical difference in the results generated in FSB, I would really question the reliability of the methods in which the results are calculated.   How can it not affect the results at all?  Put this in real live trading, I can't imagine if I were to behave in similar manner in my manual trading, ie restricted my closing n opening trade only at end of every 4hours bar even if my conditions are fulfilled 3 hours ago, do I expect my EA to yield the same results if I were to respond immediately at the fulfillment of all my trading conditions.   To tell me that even after removing PVB, the statistical results are the same is to tell me that if I were to trade this way manually,  I will get the same results and there is no difference between reacting immediately or waiting 4 hours later at end of the bar.  How can that be logically convincing?  As long as we are tied up to PBV feature, I don't think I will ever get reliable and consistent  results from any of the optimisation tools etc. (Well that's what i experience so far).  As for me, the 1 min time chart is the longest wait I can tolerate and I do believe 1 min wait won't impact my EA performance as badly as 4hours difference.

HIT

Re: Request for a "mini" optimiser

I have 4 words for you: You are absolutely wrong

1.) If you keep numeric parameters, but change the logic rule of an indicator, you change its whole idea. It is a work for the Generator instead for the optimizer.

2.) If I have to choose on feature of FSB that makes it the best tester in the branch, it is the automatic "Use previous bar value" selection.

5 (edited by bannedUser 2015-11-23 12:31:14)

Re: Request for a "mini" optimiser

Hi Popov,

I believe we see "PBV" from different perspectives and I wouldn't argue with a mathematical genius like you.  Probably if I see what you "see", I may agree with you.  Till then I still can't see the logic from my own limited perspective, though I hope you can "see" what I see.  Our differences could have been easily solved if you would let users make their own choices as in the past whereby we can choose to check or uncheck the tick box.  At the end, the onus is in the user's hand and responsibility to trade safe.  I know you are very caring and responsible person, who always remind us to trade safe. 

Most of the time I know which parameter to choose, I just need to know which smoothing, signal and base price to choose.  Can you kindly just add this "mini" feature to automatically "optimize" with the best results each time I input the parameters for the indicator?  This will definitely save me ton of efforts and time. 

Have a "tick" box (next to "PBV") for user to choose whether to enable this feature or not.  So that each time I input the time period for the indicator, it would automatically choose or optimize the best combination of smoothing, signal and base price.  I believe most people would appreciate this features and if they don't, they can just disable this feature by leaving the tick box unchecked.  For those who don't know which time period to choose, definitely least to say, they also don't know which smoothing, signal or base price to choose either.  It would save us lots of effort while trying out different time inputs to see which has the best results without having to manually try out all the various combinations.

Thanks for your kind patience

HIT

Re: Request for a "mini" optimiser

I think you are creating the strategies. and you might find that playing a lot with each parameter will cause you grief as you are looking at past data.

My thinking is that we should keep it simple, decide what general parameter suits and stick with the metrics for that one.

Another thing you can do is make your own indicators each with different sets of parameters..... that do not vary.....  I made some moving average indicators with different parameters as well as some awesome oscillators..

My 'secret' goal is to push EA Studio until I can net 3000 pips per day....

7 (edited by bannedUser 2015-11-26 12:53:02)

Re: Request for a "mini" optimiser

Before one can decide, one need to explore first.  The strength of an EA doesn't depend on simplicity, it depends on how strong the trading theory or framework is, in order to produce consistent and reliable results.

FSB need to cater to different type of traders of different experience, style and knowledge.

Those who didn't know what parameters/period to choose, definitely wouldn't even know which smoothing, signal or base price to choose either.

Those who know what parameter/period to choose, also not necessarily also know which smoothing, signal and base price work best.

Those who know the parameter/period and the smoothing, signal and base price may not necessarily know which combination yield the best results (thus also need optimization tools).

Hence, the "mini optimization" tools is suitable for traders of different level/skills/knowledge.  It help me to zoom in to work on my own parameters or trading theory instead depending on the FSB to generate or optimize for me.  I want to grow in my trading knowledge instead of depending on FSB to optimize for me.  I want to develop my own trading theory EA based on what I've learnt and observe from the market and indicators.  The "mini optimization" tools help me focus on what I want and develop it from there without changing too much of my framework.

The attraction of FSB is that it let traders have the chance to explore and try different parameters to test out their theory or concepts with instant statistical result to show whether the trading logical is workable or not.  That's the beauty of FSB, it allows you to explore and experiment.  Hence, the mini optimize help me to explore different parameters with immediate feedback instead of having to manually change the various combination to know which is optimal.  Most of us aren't mathematician to know which smoothing, signal and base price work best for each particular time period.  Thus the mini optimization tool come in handy.

I see the mini optimization tool more essential and a basic need.  What are the stages of developing an EA? 

Don't we first create the logical conditions first, with various parameters input?  Isn't here where we need the mini optimizer the most and foremost before we ever go further into any other stage such as optimization or generator?

Imagine you input the parameters and have been using the smoothing, signal and base price in it's default combination.  And one day as you try out the different combination, you realised that the results can be significantly different if only you knew which combination to use.  And you wonder all these while, did you miss out a good workable parameters because you happened to use the "wrong" combination because you didn't know which yield the best or optimal results?

HIT

Re: Request for a "mini" optimiser

When I build a strategy, I usually put it into the generator with the chain and let it generate for a few hours to see what options it likes.....

I lock some and leave some with the chain and then switch them so as to get a good idea as to what seems best.

I usually assemble the thing.. then use the generator to improve it, then perhaps take it apart and start over again.

Sometimes use the optimizer with several things locked.

I find that going back and forth between editor and generator and optimizer allows me to find what I want. only a few mouse clicks and the machine does all the work.

I am not so sure that mini optimizer would be much of an improvement re time and effort.

My 'secret' goal is to push EA Studio until I can net 3000 pips per day....

9 (edited by bannedUser 2015-11-26 16:01:29)

Re: Request for a "mini" optimiser

Hi Dave,

Thanks for sharing your style and some may be benefited by your steps and process of developing your EA.

As for me, I'm currently working on a "super" sensitive EA whereby accuracy is the crux of the matter.  This framework used very specific sets of time period to eliminate side way trending and enter trade only when it is "confirmed" whereby all the specific sets of time period are met.  This new EAs I'm developing is able to capture both short and long trends.  If the trend is short, it is able to exit on it's own with small pips and if the trend is long, a couple thousands pips.  It is so accurate that these EA can enter/exit at where the breakout happens or within a couple of minutes before/after breakouts.

My believe is that with these specific set of parameters, it provides consistency and reliability and hence it would work with whatever market/currencies because the market situations will influence the indicators and hence determine when to enter and exit.

Because I already have a "strict sets of parameters to work on, I don't need the generator nor the optimizer to help me find out other options.  I want the mini optimizer to help me focus on what I'm already have in mind, i.e. fine tuning it, not searching for new options.

Hence, that's why I feel FSB need to be "flexible" with different features to cater to different type, style and needs of traders.  And a min optimizer is just what I needed the most (the next wish list is dynamic closing, such as trailing stops with other closing options etc, not the current either or option).

HIT

Re: Request for a "mini" optimiser

the difficulty, as I see, it, is adding a routine, that would be compatible to the existing software and also of benefit to others,

He would probably need a very detailed and specific logic flow in order that he could determine the validity of your request.

He has to be able to see that it is easy to implement and from the descriptions you have provided there is not a detailed, step by step that is easy to code.

Perhaps try to break down into small steps....... very objective precisely what you think is needed.... where each step will fit in so that he can picture exactly what you feel to be of benefit.

It has to be simply laid out so he can visualize the routines and code he would have to use.

I am working on something that i would like to see him add.. and I am going to be very specific so that he will either accept it or reject it with one message.

My 'secret' goal is to push EA Studio until I can net 3000 pips per day....

11 (edited by bannedUser 2015-11-26 16:20:37)

Re: Request for a "mini" optimiser

Thanks for your advice.

Yes, I agree.  Sometimes in the midst of so many request and queries, I  sometimes wonder whether Popov manage to "see" what I'm trying to say.  But I don't blame him if he doesn't because there are so many request he has to put up with and Popov is very kind to accommodate most of them that are useful. 

I don't really know how to put my request in such a way that is aligned with the coding method etc cos I'm not a programmer to start with.  I'm just a lay person, end user.

As for the mini optimizer...the goal is, as an end user, all I need to do is to input the time period and the other options such as the smoothing, signal and base price will be automatically optimized for you based on the best profit factor etc.

Secondly, user can choose to enable or disable this function by checking or unchecking the box at the bottom of the page, next to "Previous Bar Value" where there is a box next to it (unfortunately, this option to disable it, is now not available).

HIT

Re: Request for a "mini" optimiser

There are many people who have the opinion that to be a trader, one has to have some programming skills....  I studied programming at college many years ago and learned to write a few different languages...... the new languages are vastly different so I am depending on Popov. But I still retain some knowledge and I do write a few ea's in MT4 and MT5 (and ask my son when I get stuck hahaha)

I have an idea that you want to optimize indicators as you add them, I think that may be as complex as using the optimizer after you have put the strategy together.

we do not have to use the generator before the optimizer, we can do whatever in whatever order, as long as the strategy is profitable. The optimizer refuses to work on non profitable strategies.

Maybe there is something you can change in your process to accommodate what u need.

My 'secret' goal is to push EA Studio until I can net 3000 pips per day....