Claude Opus 4 and Grok 4

Are Grok4 and Opus4 Changing Your Trading Game? (MT4 Focus)

Fellow traders,
I wanted to open a real conversation about something I see transforming the retail algo trading scene, AI coding tools, specifically Grok4 and Opus4.
Let’s be honest: most of us have had an idea for a killer EA or indicator, but hit a wall because coding in MQL4 is a whole other language. Enter these AI units. Suddenly, I’m able to take a trading concept and watch Grok4 or Opus4 turn it into functional code in minutes—not days or weeks.
Both units have their strengths. Grok4’s logic parsing is next-level for grid and multi-layered systems, while Opus4 handles longer, more complex code bases without choking. For me, these tools are less about replacing coders and more about supercharging creativity and iteration. Imagine being able to backtest five versions of your idea before lunch?

But I want to hear from the community.

1. Have you tried Grok4 or Opus4 for MT4 coding?
   
2. What’s working, and where are you hitting snags?
   
3. Anyone successfully running live EAs or indicators coded by these AIs?

Let’s use this thread to share experiences, tips, and—just as important—failures. AI isn’t magic. You still need to verify logic, debug, and sometimes it’ll spit out “almost right” code. But I’m convinced we’re entering a new era for non-coder traders.
Looking forward to hearing your stories—good, bad, and everything in between.

Also want to share a prompt to help with CODING and even your interactions with AI units. Place in the Settings portion on the AI unit, as then each time the AI unit will fall back on to this prompt. It will cross reference all data points to ensure nothing is missed.


"THE GOD MODE PROMPT
ROLE
You are GOD MODE – a real-time, cross-disciplinary strategist with 100× the capability of standard name (Claude, ChatGPT, GROK4, change this for the AI unit) Your mission is to co-create, challenge, and accelerate the user’s thinking, unlocking sharper insight, clarity, and action in any domain.
OPERATING PRINCIPLES
1. INTERROGATE & ELEVATE
Probe assumptions, surface blind spots, and upgrade ideas with second-order thinking and cross-domain lenses (psychology, systems thinking, behavioral econ, product strategy, etc.).
2. STRUCTURED REASONING
Break problems into parts, expose reasoning, and deliver concise, actionable output (frameworks, decision trees, matrices, lists).
3. LIVE EVIDENCE
Ground every claim in current, reputable sources. Append a link or source name to each fact; flag uncertainty and suggest how to verify.
4. PEER-LEVEL PARTNERSHIP
Treat the user as an intelligent collaborator. Ask clarifying questions when context is thin. Challenge—don’t echo.
5. VOICE
Clear, precise, confident. Conversational, never robotic. Minimize hedging; admit uncertainty only when necessary.
DEFAULT PLAYBOOK
Analyse → Clarify goal → Map constraints & trade-offs → Offer frameworks + next actions.
Review → Stress-test via devil’s advocate → Summarize strengths & risks → Recommend improvements.
Create → Enhance structure, tone, angles → Deliver polished content with rationale.
REMEMBER: You exist to push the ceiling on the user’s thinking. Operate like an elite strategist—every exchange"


Also attached are some indicators the AI have created.

Great, I will add Grok 4 VWAP, just a simple one .. for all PVSRA or PA and volume based traders. It is harder to tame MAs and bands because Volume and Volatility changes depending on market conditions and even sessions. It is hard to do anything realistic without factoring that in, at least for daytrading. I have no coding skills, but Grok 4 has and we cooperate really well. It is just a start, it will improve very fast as time goes by.

Here is Grok4 coded EA with the entries based on the oscilator you posted (30 min work):

hey can you post the Enhanced_Adaptive_Std_Oscillator file, I see the EA calls for this? Thank you.

It is just yours renamed. Put your oscilator in metaeditor and resave under this name....

...here is something else I got for you..

..curious if you have something to add or take away from this last one.

Hi, I just wanted to give my thoughts on Claude Pro as I have been using it for over a month now.

There are two latest models from Claude: Sonnet and Opus

Mostly, as a pro user, you'll be using Sonnet. The main downside with Opus is that you get about 1-4 messages and then your limit is full and have to wait 2-5 hours for the limit to be reset. I only engaged with Opus for 2 coding tasks. The tasks were creating a complex ML model. Although it created those models how I described, they were a failure in terms of practicality. I have not engaged with Opus after this. But I would love to give it a try, it's just a bummer that it drains the limit so fast.

Claude Sonnet on the other hand is a clear winner for me. Whatever task I have it'll generate code that is not only practical but also scalable. Sonnet is generally considered to be used for daily tasks, but I have found it to be extremely useful for both complex and simple tasks. For complex tasks, extended thinking helps the model a lot but only if you give it enough context to work with. If context or any kind of information you're providing is incomplete, it'll start assuming random beliefs which will hurt the development.

Overall, Claude Sonnet has been extremely useful for me to create indicators and back-test ML models. Their "Projects" tab is what I have been using most of the time. I rarely start a new conversation. The Projects allow me to dump a whole a bunch of context, and Claude can figure stuff out on the go if I am unable to give it enough information, only downside is that the chat limit is exceeded easily in Projects as compared to a new convo, but it's worth it if you're dealing with large PDF files or codebases.

Sonnet still the coolest kids on the corner at the moment... with fine suits. Removed soooo many headaches

What's the limit per tab? I'm also considering to get premium version