Jump to content

Impossible EAs to code?


Recommended Posts

Hello everyone,

 

I've read some posts on this forum by experienced traders that seem pessimistic when it comes to translate their strategies into logical code for an EA. I don't understand the reasoning behind this because trading is a 100% logical endeavor. The market can only do three things, go up, go down, or go sideways. Comparing indicators to determine a trade setup isn't difficult. I don't think we need a masters degree in physics to write some simple code. How hard can it be?

 

Regards

Link to comment
Share on other sites

Hello everyone,

 

I've read some posts on this forum by experienced traders that seem pessimistic when it comes to translate their strategies into logical code for an EA. I don't understand the reasoning behind this because trading is a 100% logical endeavor. The market can only do three things, go up, go down, or go sideways. Comparing indicators to determine a trade setup isn't difficult. I don't think we need a masters degree in physics to write some simple code. How hard can it be?

 

Regards

 

There's never a problem to translate trading strategy to an EA. The problem is that most traders who trade manually, and have little or no experience of computer programming, fail to translate their strategy to a logic structure that a programmer can use. In most cases, the strategy is vague and full of empty holes. That's the problem.

Link to comment
Share on other sites

Coding - easy

Translating manual trading to formal rules can be hard if the manual trader doesn't follow a complete strict set of rules. I guess many manual traders have trading systems following a simple set of rules. However, that might or might not realize that they in practice use an additional set of ad-hoc rules based on "current market condition". The process of formalizing these additional and crucial rules is probably the hardest part.

 

Should be rewarding for a manual trader to do this as he can learn exactly how the manual trading strategy rules works formally.

Link to comment
Share on other sites

Thanks for everyone's input. It makes sense that some traders cannot translate their ideas 100% logically or "simple" sets of rules are actually complex to implement. If the trading rules and filters can be expressed in a logical formula, then it is very doable. As far as ad-hoc filters due "current market conditions", it can be easily done by performing a query on a trusted and organized source of information to prevent such bad trades. Thanks for all the input.
Link to comment
Share on other sites

I see you are thinking logical as a programmer :)

 

Just a comment on your last statement. The ad-hoc rules based on "current market conditions" I was referring to could be the manual traders understanding of the current market based on any number of inputs and consequently very hard to formalize. If not formalized, it would be close to impossible to retrieve that information. However, as you said, if he/she evaluates current market depending on some external information that could be formalized then we would be able to query that.

 

And taking the subject further for the fun of it...if we had a large trading history of the manual trader and as much formal rules a possible we could have the ea adapt to the trading "style" of the manual trader. Ie, using artificial intelligence, the ea could learn to trade like the manual trader and consequently fill in the missing informal rules. But now its not exactly simple ea coding any more :)

Link to comment
Share on other sites

From my experience, a simple manual system is fairly easy to program into an EA. A really good comprehensive manual system, that seems straight forward to a human trader, is wickedly complex to put into a hard-coded set of rules. Add to that the fact that the logic must continually be "updated" to adapt to current market conditions.
Link to comment
Share on other sites

In my opinion it's not worth the effort to think about "ad hoc filters"

The big players spent tons of money for their systems that simply scan the incomming headlines and excecute trades based on the content. Retail will never be as fast as they are...so always late to the party and a welcome victim of fake moves

Link to comment
Share on other sites

In my opinion it's not worth the effort to think about "ad hoc filters"

The big players spent tons of money for their systems that simply scan the incomming headlines and excecute trades based on the content. Retail will never be as fast as they are...so always late to the party and a welcome victim of fake moves

 

I agree, but this discussion is in relation to manual traders. A manual trader may be acting a certain way due to something he watched on the morning news several hour ago. This would be the type of informal "rules" that is hard to implement.

Link to comment
Share on other sites

Intuition cannot be programmed. Most of profitable trader mixes set of rules and intuition. Same like you play any games, you always fail at the first attempt. You practice, practice and practice. Then you start to conquer the game. That is intuition. That makes most of EA fail. If trading is simply about set of rules, then everyone will get rich.
Link to comment
Share on other sites

I see you are thinking logical as a programmer :)

 

Just a comment on your last statement. The ad-hoc rules based on "current market conditions" I was referring to could be the manual traders understanding of the current market based on any number of inputs and consequently very hard to formalize. If not formalized, it would be close to impossible to retrieve that information. However, as you said, if he/she evaluates current market depending on some external information that could be formalized then we would be able to query that.

 

And taking the subject further for the fun of it...if we had a large trading history of the manual trader and as much formal rules a possible we could have the ea adapt to the trading "style" of the manual trader. Ie, using artificial intelligence, the ea could learn to trade like the manual trader and consequently fill in the missing informal rules. But now its not exactly simple ea coding any more :)

 

Hello askalas,

 

I was thinking of this myself today. Sort of like a neural network, artificial intelligence, that has many properties. I plan on developing a system like this soon, I think it has great potential. Thank you again for your input.

Link to comment
Share on other sites

From my experience, a simple manual system is fairly easy to program into an EA. A really good comprehensive manual system, that seems straight forward to a human trader, is wickedly complex to put into a hard-coded set of rules. Add to that the fact that the logic must continually be "updated" to adapt to current market conditions.

 

Hello fughe,

 

I agree with you it can get very complex, but still doable. That is where artificial intelligence kicks in and takes care of that. Such systems are able to adapt, just like we humans do.

Link to comment
Share on other sites

Intuition cannot be programmed. Most of profitable trader mixes set of rules and intuition. Same like you play any games, you always fail at the first attempt. You practice, practice and practice. Then you start to conquer the game. That is intuition. That makes most of EA fail. If trading is simply about set of rules, then everyone will get rich.

 

Hello sgr,

 

I believe that intuition is mainly a game of emotions. Unless the trader is highly psychic to predict the future, it will be very difficult to be profitable relying in part on emotions. I am not saying it is impossible to be succesful doing it that way, but very difficult to achieve. If you are successful relying on intuition, I take my hat off to you.

Link to comment
Share on other sites

In my opinion it's not worth the effort to think about "ad hoc filters"

The big players spent tons of money for their systems that simply scan the incomming headlines and excecute trades based on the content. Retail will never be as fast as they are...so always late to the party and a welcome victim of fake moves

 

Hello iwjw,

 

Checking for news as a filter could be used simply as a delay to watch for reversals. As you know, I am a big fan of reversals. I've noticed these usually occur precisely at certain times during the hour. Taking this predictability into account, it could be very easy to trade by knowing when the reversals take place.

Link to comment
Share on other sites

I agree, but this discussion is in relation to manual traders. A manual trader may be acting a certain way due to something he watched on the morning news several hour ago. This would be the type of informal "rules" that is hard to implement.

 

Hello askalas,

 

That is precisely where emotions get inside the equation. We humans base our decisions mainly on fear or pleasure, emotions. And how delayed those emotions may be vary for each of us. But still it is a factor to consider.

 

This discussion is mainly philosophical, all input is welcomed. Thank you everyone for your thoughts.

Edited by prosper4all
Adding comment
Link to comment
Share on other sites

  • 5 months later...

An EA that trades like a professional and makes you rich IS possible.

 

It is, however, highly unlikely that anyone would release them, because to code such a thing requires an awful lot of coding time and patience.

 

The problem is basically one of pattern recognition. You know those captcha boxes that appear on the net to confirm that you are not a robot? Well, getting a robot to be able to read those blurry words is incredibly HARD to code. That's why the captchas exist, BTW... to weed out robots.

 

The market charts are JUST LIKE CAPTCHAS! You are looking for patterns, but sometimes the patterns are twisted sideways, or twisted out over long periods of time, to different degrees and in different shapes... but still recognizable to the human eye. Getting a robot to "read" the chart visually is very hard to code, especially in MQL4 which doesn't have any command specifically tuned to do this sort of analysis. I'm sure it would still be possible, but damn, even as a programmer I can't start to begin to think how to do this with MQL4's limited command structure, and in such a way to catch out every single nuance of the way that patterns form.

 

That said, I'm sure it's not impossible... but whoever has done it clearly had plenty of time to code it, and is probably already a successful manual trader who, despite having bags of money and time, decided to code it for the hell of it.

 

In the meantime, most EAs suck. The best I've seen that don't try to trade like a manual trader, use grid hedging.... mimicking actual dealer strategy (albeit flying blind without knowledge of any trader volume!), and even then you need a lot of equity to survive drawdown on open float.... something the average retail trader simply doesn't have!

Besides, it's actually the least effective method of trading.

 

Anyone else trying to write a robot should attempt to write something that at least is pretty good at picking off highs and lows of the day.

Edited by Rio
Link to comment
Share on other sites

It is much better to use artificial intelligence and machine learning. A machine can go through thousands of potential systems per second and find patterns that a human would never even think of. While the setups it finds are usually not apparent to the human eye, they surely must be better than anything any human can do.

 

No pro actually trades discretionarily by using charts. Sure, you can find people here who claim to do that. But if they are so profitable then why arent they sharing their signals? It is not like indo people have enough money to move the markets. I have never seen anyone on collective2 claim to trade discretionarily.

Link to comment
Share on other sites

What it comes down to is being able to quantify the rules into system of binary logic. There will be lots of calculations, but every part must come down to a true/false scenario at some point in the calculation. That is where it is sometimes nearly impossible to quantify a system. Some parts will require layers upon layers of binary logic to arrive at a simple conclusion that a human can do by glancing at the chart.
Link to comment
Share on other sites

To trade discretionarily, you have to be in the zone. I have done it before, and I know people who have. It is extremely unreliable because you have to be 100% focused and anything can break your focus and cause you to get on the wrong side of the market. The best thing to do when trading like that is to train yourself to stop trading the instant you make a loss. Not doing so is where people end up giving away all their earnings.
Link to comment
Share on other sites

and that is why traders and coders still exist with Hope of taking what they have experienced into a programmable methodology. mQL has been around since 06 and very popular by 07-08 . And the only good EA's are ones that focus on 1 particular event which doesn't happen everyday. A discretionary trader works the same way.
Link to comment
Share on other sites

What it comes down to is being able to quantify the rules into system of binary logic. There will be lots of calculations, but every part must come down to a true/false scenario at some point in the calculation. That is where it is sometimes nearly impossible to quantify a system. Some parts will require layers upon layers of binary logic to arrive at a simple conclusion that a human can do by glancing at the chart.

 

 

writing the code is the easy part... trying to translate thought into x=y+z is a bit more difficult....

 

thought contains all the bias of the past as it relates to the current situation... computers have "no memory"...

it's easier for the brain to recognise patterns than computers...

in thought you analyse every past scenario, and try do draw similarities as they relate to the current situation.... try telling a computer to do that...

 

have you ever tried saying directly what you are thinking? try it some time.... there's a filter between your brain and your lips...

you'll find it impossible to do, as thoughts occur all at once, while logic is a step by step thing...

you can't talk quick enough to get all the thoughts out... and by the time you come to saying something you were thinking about 2 minutes ago, your perception may have been changed by other thoughts....

 

only once we can map the brain directly to code, will we be able to do it...

 

my 0.02p

neural networks are just a more complex way of rolling a dice...

 

the markets are not random, just highly chaotic, with too many inputs to be able to reliably determine any specific outcome for any sustainable length of time (hence why <most> ea's fail within a year of being released... they worked at that time under those market conditions... but not now...)

Link to comment
Share on other sites

Oh, don't misunderstand me. I am of the belief that the markets are quite logical. They seem random, but are not actually random. I am convinced there is a way to decode what the market is doing at all times, I am just not sure there is a way to put it into code that a computer can handle.
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...