Why some traders that know how to code EAs still prefer to trader manually?

To add comments, please log in or register
Mrluck07
425
Mrluck07  

I personally know a lot of traders that can create their own EA's very well, but somehow they still choose to trade manually, why this phenomenon happens?

There are big disavantages, for example 1 person cannot monitor multiple markets at the same time, or work around the clock

Icham Aidibe
11287
Icham Aidibe  
Mrluck07:

I personally know a lot of traders that can create their own EA's very well, but somehow they still choose to trade manually, why this phenomenon happens?

There are big disavantages, for example 1 person cannot monitor multiple markets at the same time, or work around the clock

Well, manually you remain connected with the financial world, you have to follow charts, news, believe it or no, there are restaurant's tables where you can forget to sit if you're not aware of the latest oil prices for example :-)
Jox90
686
Jox90  
Mrluck07:

I personally know a lot of traders that can create their own EA's very well, but somehow they still choose to trade manually, why this phenomenon happens?

There are big disavantages, for example 1 person cannot monitor multiple markets at the same time, or work around the clock

I've never heard about an EA that can do what a human can do. It doesn't think, it cannot reason. But I would like to say: currently it cannot think... Maybe in the future if we find a way to make a "strong" artificial intelligence, that differs from "weak" A.I. (which is current artificial intelligence), we'll be able to make an EA which would do the work in our place, and which would make trading world see its end, together with our world... XD
Icham Aidibe
11287
Icham Aidibe  
jox90:
I've never heard about an EA that can do what a human can do. It doesn't think, it cannot reason. But I would like to say: currently it cannot think... Maybe in the future if we find a way to make a "strong" artificial intelligence, that differs from "weak" A.I. (which is current artificial intelligence), we'll be able to make an EA which would do the work in our place, and which would make trading world see its end, together with our world... XD

Well, an EA would do whatever you teach it to. 

Icham Aidibe
11287
Icham Aidibe  
Zee Zhou Ma:

EA can only be created to only trade a certain market structure. If you have to create an EA for every type of market structure, it may take one's life time effort.

Yes it's based on condition and patterns. The more patterns you'll teach it, the more pattern it'll be able to handle, very coldly, where there's no place for "let's imagine", "let's suppose" etc ... 


I would add that by trading manually comes a point where you yourself feel like an automate, it's actually the good moment to start to code your own EA.

soubra2003
31844
soubra2003  
Maybe they have not created a robot that They Trade Manually Yet.
Or Testing a Strategy Manually To Create Or Modify On It Then Create The EA In The Future.
Amir Yacoby
1221
Amir Yacoby  
Icham Aidibe:

Well, an EA would do whatever you teach it to. 

But you can't teach it to recognize a face (unless you use AI) - so their level in recognizing complex patterns is far lower than humans.
Tomas Rodriguez Doherty
9464
Tomas Rodriguez Doherty  
Amir Yacoby:
But you can't teach it to recognize a face (unless you use AI) - so their level in recognizing complex patterns is far lower than humans.

Most of EA programmers have limitations. It is easy to say that if you can do it manually then you can do it by code. But there are two limitations here, SOme imaginable code is so complex to create, I mean, human mind works a level of complexity, as human perception, and code works on basic levels of logic. The other limitation is in mql language which has very limited possibilities and runs in every tick, so there cannot be too much complexity in the algorithms . They can be quite complex but at a certain amount of variables and conditions it would start lagging the computer or running out of memory. At the moment working with indicators and checking basic parameters is all we can imagine. Some things are easy to imagine but difficult to translate into code. ALso there is a limitation in the knowledge about maths and logic that the coders have like me. I would love to create a robot that learn from mistakes and constantly evolves on certain conditions and that interacts with trades with the market and could be aware of what is happening in the market. But this is far away from my reach at the moment. 

Yohana Parmi
4990
Yohana Parmi  
Mrluck07:

I personally know a lot of traders that can create their own EA's very well, but somehow they still choose to trade manually, why this phenomenon happens?

Because -- nothing can defeat what God has created :) 
Eyes, Brain, Mind and Analysis, Feelings, and Hearts, resulting in a unique experience.
-- and the experience of every human being can always be different,
-- as the difference in the retina of the eye in every human.

Amir Yacoby
1221
Amir Yacoby  

To create, test, and maintain a real good adjustable robot that trades, is a task of many human years, maybe hundreds or thousands. and thus it is hard for 1 person with limited free time and a day job. It is a kind of task that a company should do, or a team of programmers. 

A team of 10 or 20 programmers working in collaboration for a year is equal to 20 human years of work. 

*- which is a good idea I think to open here team working requests on projects. It will be a lot faster to collaborate between programmers to achive a good result. 

 

Icham Aidibe
11287
Icham Aidibe  
Amir Yacoby:
But you can't teach it to recognize a face (unless you use AI) - so their level in recognizing complex patterns is far lower than humans.

Face recognition, that's how they do it : 

https://medium.com/@ageitgey/machine-learning-is-fun-part-4-modern-face-recognition-with-deep-learning-c3cffc121d78

Fortunately, charts are less complex than faces, it's not difficult to register for each positions opened numberous differents parameters (vol, price, ma, volatility, whatever ...), analyze order history, reference & classify orders with such characteristics, patternize it, and self adapt based on previous results being more often profitable or unprofitable.

Actually, the more complex task there seems to me to be extracting a rule based on logs to apply for next positions ( key value limits for volatility for example, relative position to a ma etc ...)

Machine Learning is Fun! Part 4: Modern Face Recognition with Deep Learning
Machine Learning is Fun! Part 4: Modern Face Recognition with Deep Learning
  • 2016.07.24
  • Adam Geitgey
  • medium.com
Update: This article is part of a series. Check out the full series: Part 1, Part 2, Part 3, Part 4, Part 5, Part 6, Part 7 and Part 8!
12
To add comments, please log in or register