I’m working on a tool to help convert trading strategy documents into structured MQL5 EA logic.

New comment
 

I’m working on a tool to help convert trading strategy documents into structured MQL5 EA logic.

Curious to know: what part of EA creation takes the most time for you?

 
David Chidiebere Chinweike:

I’m working on a tool to help convert trading strategy documents into structured MQL5 EA logic.

Curious to know: what part of EA creation takes the most time for you?

That depends on the specific "trading strategy documents" at hand. Do the documents designate a market to trade? An instrument to trade? How about a timeframe, or a defined custom chart structure (Renko, range bars, equal volume bars, etc. with settings)?

Assuming that the market and instrument are predetermined, settling on a chart structure that most reduces noise/chop can be quite time consuming. The Uniformity Factor script in the CodeBase can speed this up a bit.

If I didn't do this first, all indicators and logic would essentially be contaminated by the noisy/choppy underlying chart data─not exactly a good start to building an EA, nor any strategy for that matter.
 
David Chidiebere Chinweike:
what part of EA creation takes the most time for you?
To be fair, The most time I spent is on the idea how the EA should work, and (almost) all the coding does AI for me ( mad respect to anyone who still 100% code his own ea without ai tho)
 
Thomas Eduard Van Der Jagt #:
mad respect to anyone who still 100% code his own ea

Mad thanks right back at you.

At the end of the day, I can type code faster than I can review someone (or something) else's code.

 
Ryan L Johnson #:

That depends on the specific "trading strategy documents" at hand. Do the documents designate a market to trade? An instrument to trade? How about a timeframe, or a defined custom chart structure (Renko, range bars, equal volume bars, etc. with settings)?

Assuming that the market and instrument are predetermined, settling on a chart structure that most reduces noise/chop can be quite time consuming. The Uniformity Factor script in the CodeBase can speed this up a bit.

If I didn't do this first, all indicators and logic would essentially be contaminated by the noisy/choppy underlying chart data─not exactly a good start to building an EA, nor any strategy for that matter.
The tool is built in a way that you can add custom instruction to make the final result (EA) to meet your trading style and everythin you consider as a good strategy
 
Thomas Eduard Van Der Jagt #:
To be fair, The most time I spent is on the idea how the EA should work, and (almost) all the coding does AI for me ( mad respect to anyone who still 100% code his own ea without ai tho)
I don't worry about the EA no more because i have some prompt templates on how the EA should operate. most of my time goes into strategy structure, strategy monitoring and optimization.
 
David Chidiebere Chinweike #:
The tool is built in a way that you can add custom instruction...
David Chidiebere Chinweike #:
[I] have some prompt templates...

Ah, I see... AI.

So you need to have complete strategy rules in advance of using the tool.

If I were writing an EA template, I would include inputs for:

  • a tradable symbol (for executing orders), 
  • an analysis symbol (for custom charts),
  • an hour and minute entry filter,
  • a daily entry filter,
  • a monthly entry filter,
  • enable/disable buy entries,
  • enable/disable sell entries,
  • percentage of account balance lot sizing,
  • variable lot sizing based on N consecutive losses,
  • a fixed SL,
  • a trailing SL,
  • a fixed TP,
  • a Friday/market close-all exit time,
  • trade levels and labels thereof for custom charts,
  • data server disconnect alerts,
  • a custom enum for using various indicator calculations for entries, and
  • a custom enum for using various indicator calculations for exits.

(Obviously, the list above is not exhaustive and is skeletal at best, but it should give you some ideas).

 
Great initiative, David. For me, the most time-consuming part is translating vague strategy rules into precise, testable conditions — things that seem obvious on paper but require a lot of edge-case thinking when coding. A tool that helps structure that logic from the start would definitely speed things up. Looking forward to seeing how this develops.
 
Alvaro Albillos Gutierrez #:
Great initiative, David. For me, the most time-consuming part is translating vague strategy rules into precise, testable conditions — things that seem obvious on paper but require a lot of edge-case thinking when coding. A tool that helps structure that logic from the start would definitely speed things up. Looking forward to seeing how this develops.

I was thinking the same thing, especially for brand new traders. As trial lawyers say, "You can't get the answer that you want, if you can't frame the question." The same goes for AI, presumably.

I believe that writing a universal template would be a lot of work. A few more things to consider are:

  • the slope of an indicator over the course of N bars (not an angle that varies according to chart zoom, but a binary up/down indicator bar count),
  • the span between and price (OHLC, bid, ask, or last) and the indicator,
  • the span between multiple indicators,
  • crossovers between price and an indicator,
  • crossover between multiple indicators,
  • volume/tick volume analysis...
The OP's project strikes me as something that could potentially be an endless journey. It will definitely be interesting to see the finished tool.
 
Ryan L Johnson #:

Ah, I see... AI.

So you need to have complete strategy rules in advance of using the tool.

If I were writing an EA template, I would include inputs for:

  • a tradable symbol (for executing orders), 
  • an analysis symbol (for custom charts),
  • an hour and minute entry filter,
  • a daily entry filter,
  • a monthly entry filter,
  • enable/disable buy entries,
  • enable/disable sell entries,
  • percentage of account balance lot sizing,
  • variable lot sizing based on N consecutive losses,
  • a fixed SL,
  • a trailing SL,
  • a fixed TP,
  • a Friday/market close-all exit time,
  • trade levels and labels thereof for custom charts,
  • data server disconnect alerts,
  • a custom enum for using various indicator calculations for entries, and
  • a custom enum for using various indicator calculations for exits.

(Obviously, the list above is not exhaustive and is skeletal at best, but it should give you some ideas).

The inputs you mentioned is elaborative. Consider that AI has improved a lot in generating code, but that doesn't mean it isn't prone to error. Most of the coding is done by AI and I worry about my workflow and strategy building.
 
Alvaro Albillos Gutierrez #:
Great initiative, David. For me, the most time-consuming part is translating vague strategy rules into precise, testable conditions — things that seem obvious on paper but require a lot of edge-case thinking when coding. A tool that helps structure that logic from the start would definitely speed things up. Looking forward to seeing how this develops.
The same goes for me, that have been my major pain moving from ideal to EA to Validation. The aim is to build a tool that cut the amount of time I spent in validating a trading strategy. You can join the waitlist at zeff.codes
12
New comment