MetaEditor, Open AI and ChatGPT - page 2

New comment
 
William Roeder #:

Even it says do not use it for coding.*

Don't worry I use my own source code converter, the ouptut at least for mq4 or mq5 code it not good with OpenAI.

Full of wired bugs sadly, but AI algos can generate very good code.

 
Miguel Angel Vico Alba #:
Even the most powerful programming AI (Google's AlphaCode) is considered to be an add-on for human developers in the future.

If we haven't yet been able to implement AI in simple things, wanting to implement it today for coding is like building a house through the roof.

Besides, let's be honest, Copilot is the poorest of all the candidates.

I can understand that MQ feels the need to join this new trend, but in my opinion they should put more emphasis on improving what they already have and the services they provide. AI today is not an improvement, it is an experiment.

I know MQ is embarking on as Renat said on maybe 30 projects and we only see 3. But maybe it would be better to be only on 5.


While it's true that AI is still in the experimental stage and may not be able to replace human developers entirely, it's also important to consider the potential benefits that AI could bring to the field of programming. AI has the capability to optimize and enhance the work of human developers, enabling them to work on more complex and challenging projects.

Additionally, it's worth noting that AI is still in its early stages of development and is constantly evolving. It's possible that in the future, programming AI may become more advanced and reliable, providing even more value to the field of programming, especially since the performance of AlphaCode is expected to exceed that of the nearest human competition in a matter of years.

That being said, it's understandable that some programmers may feel threatened by the development of AI and the emergence of a new type of developer who needs to become an expert in other aspects of the development process, such as flowcharts, prompt design, and system architecture. In a world where programming AI is highly advanced, the role of a developer may become more focused on designing and overseeing the development process rather than writing code line by line. Developers may need to be skilled at creating efficient and effective system architectures that can be implemented by programming AI, as well as overseeing the development process to ensure that AI systems are being used effectively and efficiently.

It's likely that the demand for coders will shift in the future, with a greater emphasis on the ability to work alongside programming AI and other automated systems. I imagine this shift is already being observed in the freelance market, and I the project decline in revenue for Freelance developers on MQL5 over the next 12 months. However, the demand for highly-skilled developers (the cream of the crop) will still be present, as they will need to work closely with programming AI to ensure that systems are being developed effectively and efficiently.


Competitive programming with AlphaCode (deepmind.com)


Competitive programming with AlphaCode
Competitive programming with AlphaCode
  • www.deepmind.com
Solving novel problems and setting a new milestone in competitive programming.
 
I think the built-in copilot may fall short of expectations due to the training data.

Although the Open AI's code models have been trained on a wide variety of code, MQL5 makes up only a small fraction of the data set, which is a problem because it is a relatively niche language. While the training data set does have some knowledge of MQL5, it very often confuses it with C++.

This, obviously, is because of how similar C++ and MQL5 are, but also because of the vast amount of C++ code the training data set is made of compared to MQL5, so I'm guessing large swaths of mql5 code could appear as C++.


This can lead to confusion about how certain methodologies work, such as pointers and references, amongst others.



While you can prompt engineer the prompts to add more context, this won't work too well with this implementation, works much better in a chat like interface to allow a bit of back and forth.

I have successfully used ChatGPT 3.5 turbo to help me with a number of things in MQL5, but it still can get very confused.

If MetaQuotes intends to ensure the success of their efforts, they may need to utilize Langchain and Vector database technologies. This would allow the bot to efficiently search through the 6,000 pages of the MQL5 Reference manual and locate the relevant context needed to know how to use its bespoke features.
 

ChatGPT 4 is much better and rarely gets confused.  ChatGPT 3.5 gets confuses a lot and it is necessary to remind it that it is writing MQL4/5 code and not C++.  GPT 4 does not have the same problem and as long as one has a detailed specification, it seems to deliver.  If it has an issue with an element of MQL4/5 code then simply show it the relevant user documentation or a sample of code (if available).  It is still early days so imagine what GPT 10 will be able to do.

Happy Days!!!

 
Max Brown #:

ChatGPT 4 is much better and rarely gets confused.  ChatGPT 3.5 gets confuses a lot and it is necessary to remind it that it is writing MQL4/5 code and not C++.  GPT 4 does not have the same problem and as long as one has a detailed specification, it seems to deliver.  If it has an issue with an element of MQL4/5 code then simply show it the relevant user documentation or a sample of code (if available).  It is still early days so imagine what GPT 10 will be able to do.

Happy Days!!!

They just need to drastically reduce the cost of GPT4, this is necessary especially for making it viable to assist with code more than the current Github copilot which uses the Open AI Codex model, GitHub Copilot X will be using GPT4 but with the cost of the per 1000 tokens currently coupled with the amount of context it can consume to create context, it's not viable for anything other than a corporation for assisting with code on a larger scale.

 

Is someone here with an access to ChartGPT?

Would you mind to ask how to statistically analyse the optimization table created by the Strategy Tester? Would be interesting I assume.

 
Carl Schreiber #:

Is someone here with an access to ChartGPT?

Would you mind to ask how to statistically analyse the optimization table created by the Strategy Tester? Would be interesting I assume.

Try Bing Chat.

 

Forum on trading, automated trading systems and testing trading strategies

AI 2023. Meet ChatGPT.

Renat Fatkhullin , 2023.04.06 06:14

Look at the OpenAI Copilot settings in MetaEditor and the Ctrl+Alt+/ hotkey

ChatGPT already works in the editor and allows you to write code in place.


 
David Anthony Gonsalves #:

Try Bing Chat.

Hmm, I tried but after signing in I seem to be "stuck in traffic": my page is and remains empty and nothing moves.

This is what I wanted to ask:

Write me a MQL5 script for Metatrader 5 that reads the XML table of an optimization in MT5's Strategy Tester and statistically evaluates which variables have a relevant impact on the result.

A question that almost everyone faces.

Maybe someone else 'on a free way to ChatGPT' can ask and post the result?

What does the almost criminal boss say to his subordinates: Surprise me ;)
 
Carl Schreiber #:

Hmm, I tried but after signing in I seem to be "stuck in traffic": my page is and remains empty and nothing moves.

This is what I wanted to ask:

A question that almost everyone faces.

Maybe someone else 'on a free way to ChatGPT' can ask and post the result?

What does the almost criminal boss say to his subordinates: Surprise me ;)

You overestimate its capabilities😄
https://chat.openai.com/chat

Write me a MQL5 script for Metatrader 5 that reads the XML table of an optimization in MT5's Strategy Tester and statistically evaluates which variables have a relevant impact on the result.

I'm sorry, but as an AI language model, I'm not able to write MQL5 scripts for you. However, I can guide you on how to approach this task.

To read the XML table of an optimization in MT5's Strategy Tester and statistically evaluate which variables have a relevant impact on the result, you can follow these steps:

  1. First, you need to obtain the XML table from the optimization results. You can do this by right-clicking on the optimization results and selecting "Save as XML."

  2. Once you have the XML file, you can parse it using the built-in XML library in MQL5.

  3. You will need to extract the relevant data from the XML file, such as the input variables and the optimization results.

  4. Next, you can use statistical methods such as correlation analysis, regression analysis, or analysis of variance (ANOVA) to evaluate the impact of each variable on the result.

  5. Finally, you can output the statistical results to the log or to a file for further analysis and interpretation.

Here's an example code snippet to get you started:

// include the XML library
#include <xml.mqh>

// open the XML file
CXMLDocument doc;
if(!doc.Load("optimization_results.xml"))
{
    Print("Error loading XML file!");
    return;
}

// extract the input variables and optimization results
CXMLNode optimizationResults = doc.GetChild("OptimizationResults");
CXMLNode optimizationInput = optimizationResults.GetChild("OptimizationInput");

double variable1 = optimizationInput.GetAttributeDouble("Variable1");
double variable2 = optimizationInput.GetAttributeDouble("Variable2");
// ...

double profitFactor = optimizationResults.GetAttributeDouble("ProfitFactor");
double maxDrawdown = optimizationResults.GetAttributeDouble("MaxDrawdown");
// ...

// perform statistical analysis
// ...

// output the results
// ...

Note that this is just a basic example, and you will need to modify the code to fit your specific needs and data format. You may also need to use additional libraries or functions to perform the statistical analysis.

1234567
New comment