Something interesting from Chatgpt/AI please post here

xiannan wrote: Fri May 08, 2026 8:50 pm Could you please tell me if the bubbles in the QuantMasterLiquityEvents indicator can be enlarged to match the size of the bubbles in the volume bubble indicator? Also, what do the red and green dashed lines in the indicator mean?

Instead of reading that link https://www.tradingview.com/script/dnbS ... me-Bubble/ you ask questions.

"Hi bro, I'd like to know what the differences are between version 1.01 and the previous version 1.0?"
I had once translated pinescript with Grok, now I proofread it with Gemini Pro

Pelle wrote: Fri May 08, 2026 9:22 pm Instead of reading that link https://www.tradingview.com/script/dnbS ... me-Bubble/ you ask questions.

"Hi bro, I'd like to know what the differences are between version 1.01 and the previous version 1.0?"
I had once translated pinescript with Grok, now I proofread it with Gemini Pro

Okay, I see. Thank you.

I converted that Pinescript indicator https://www.tradingview.com/script/irBB ... r-LuxAlgo/ to MT5,

Don't ask how to trade with it, just read the indicator's homepage.

Pelle wrote: Fri Dec 26, 2025 5:41 am I made a few more like this for Christmas.
namely Gemini_Scalping_Strategy, (https://www.tradingview.com/script/xFzLDeyd/) which I modified so that it does not alert on the arrow (I removed the arrows from my template), but alerts when the dashboarding CHOP (Trend) is in TREND mode (Not NEUT or SIDE).

In addition, QuantMasterLiquityEvents (https://www.tradingview.com/script/dnbS ... me-Bubble/), I added alerts to it. In addition, when the bubbles are small, I added a feature, when you put the cursor over the bubble you can see the bubble's properties in more detail (=hovering or something like that)

All made by AI (Gemini Pro).

If you already downloaded QuantMasterLiquidityEvents, I made (Gemini Pro did). a fix to the dashboard. Redownload

EDIT: ADDED MT4 version of Gemini_Scalping_Strategy,
EDIT: I don't have crypto on my MT4 account, please tell me if they also work with crypto on MT4

Brother, I've noticed a problem with this indicator. I opened it on the same broker's MT4 software, initially with two chart windows of the same timeframe (1 minute). Then I switched one to a 5-minute timeframe. When I switched the 5-minute chart back to 1 minute, the bubble data displayed was different from the other 1-minute chart. Initially, when both charts were 1-minute, the bubble data and numbers were identical, but after switching, they didn't match. What's going on? I've taken a screenshot; please take a look.

xiannan wrote: Tue May 12, 2026 10:06 am Brother, I've noticed a problem with this indicator. I opened it on the same broker's MT4 software, initially with two chart windows of the same timeframe (1 minute). Then I switched one to a 5-minute timeframe. When I switched the 5-minute chart back to 1 minute, the bubble data displayed was different from the other 1-minute chart. Initially, when both charts were 1-minute, the bubble data and numbers were identical, but after switching, they didn't match. What's going on? I've taken a screenshot; please take a look.

I have always tried to provide the code as I provided it.

That is why I always provide the source code as the original developer on the tradinview platform provided it.

If it doesn't work, you have the source code from which to develop it.

Pelle wrote: Thu May 14, 2026 4:23 am I have always tried to provide the code as I provided it.

That is why I always provide the source code as the original developer on the tradinview platform provided it.

If it doesn't work, you have the source code from which to develop it.

hdinview platform provided it.

If it doesn't work, you have the source code from which to develop it.
[/quote]

Okay, thank you very much. However, I'm not a programmer, and I don't know where to modify this code. So, I can only trouble you to help me see how to modify or optimize this metric. I'm sorry to bother you.

xiannan wrote: Fri May 15, 2026 1:33 am Okay, thank you very much. However, I'm not a programmer, and I don't know where to modify this code. So, I can only trouble you to help me see how to modify or optimise this metric. I'm sorry to bother you.

You don't have to be a programmer. Just literate enough to ask the right questions.

Try any of these...

As of mid-2026, the AI landscape is divided into **Frontier Agents** (ready-to-use autonomous systems) and **Reasoning Models** (the "brains," such as DeepSeek and Kimi, that power those agents).

Here is the comprehensive list of the top AI agents and models, categorised by their primary function.

### **1. Frontier Reasoning Models (The "Brains")**

These are the most powerful models that act as the engine for agents. They are similar to DeepSeek and Kimi in that they provide the raw intelligence and tool-calling abilities.

* **DeepSeek V4-Pro:** The top open-weight model; known for extreme cost-efficiency and 1M token context.
* **Kimi K2.6:** Lead in agentic reasoning and sub-agent "swarms" for parallel task execution.
* **GPT-5.5:** OpenAI’s flagship; the gold standard for math and general reasoning.
* **Claude Opus 4.7:** Known for "Adaptive Thinking"—it adjusts its reasoning time based on task difficulty.
* **Gemini 3.1 Pro:** The leader in long-context memory (up to 2M tokens) and native multimodal (video/audio) reasoning.
* **Qwen 3.6 Plus:** Alibaba’s powerhouse; currently the top-performing open-weight model for coding.
* **GLM-5:** Zhipu AI’s flagship; highly popular for private enterprise deployment and specialised fine-tuning.
* **Mistral Large 3 / Medium 3.5:** The European champion of open-weight models, optimised for efficiency and low-latency.
* **Grok 4.3:** xAI’s model, featuring the most real-time data access and high-speed agentic execution.
* **Llama 4 Scout:** Meta’s latest open model, optimised for speed and local deployment on personal hardware.

### **2. Autonomous Productivity & Web Agents**

These agents use the models above to actually *do* work across your computer and the web.

* **OpenAI Operator:** A fully autonomous web-browsing agent that can book travel and buy products.
* **Manus AI:** A general-purpose "agent-of-all-trades" that handles research and stock analysis.
* **MultiOn:** An agentic layer that can navigate any website to perform complex workflows.
* **HyperWrite Personal Assistant:** A specialised agent for document-heavy administrative tasks.
* **Perplexity Computer:** An agentic evolution of the search engine that can execute research and report-writing autonomously.

### **3. Software Engineering Agents (The "Coders")**

These systems don't just write code snippets; they manage entire repositories.

* **Devin:** The world’s first autonomous AI software engineer.
* **Claude Code:** Anthropic’s specialised terminal-based agent for deep codebase refactoring.
* **Cursor / Windsurf:** "Agentic IDEs" where the editor itself can write and debug across dozens of files at once.
* **Replit Agent:** An agent that builds, deploys, and hosts applications from a single natural language prompt.
* **Aider:** A popular open-source command-line agent that works directly with your local git files.

### **4. Enterprise & Multi-Agent Platforms**

Systems that allow businesses to deploy "teams" of agents.

* **Salesforce Agentforce:** Autonomous agents built directly into the world's #1 CRM.
* **Microsoft Copilot Studio:** A platform to build agents that live in Teams, Outlook, and Excel.
* **Vybe:** A unique platform where agents build their own specialised web apps to solve your business problems.
* **CrewAI:** The most popular framework for orchestrating multiple agents to work together (e.g., a "Researcher" and a "Writer").
* **Zapier Central:** An agentic hub that connects AI to over 6,000 different business apps.
* **Glean:** An enterprise-search agent that acts as a "corporate brain" for internal knowledge.

### **5. Specialist & Growth Agents**

* **NoimosAI:** A "Marketing Squad" agent that manages social media and SEO autonomously.
* **Intercom Fin:** A high-autonomy customer support agent that resolves tickets without human help.
* **Albert.ai:** An autonomous agent for digital marketing and ad-spend optimisation.

I'm not a coder either, all my codes are made with AI.

And if you find any errors, ask some AI how to fix the error.

I myself can't reproduce the error you pointed out last. I went through all the timeframes and went back to M1, still the same

AI have analyzed (AI) the code and your description of the discrepancy. This is a classic issue in MQL5 development related to History Synchronization and State Management when switching timeframes.

The Root Causes
Context-Dependent Z-Scores (The "Start of History" Problem): Your Z-score calculation depends on a 60-bar lookback (InpLenStat). In MT5, index 0 is always the oldest bar currently loaded on the chart.

Chart A (Stayed on 1m): Might have 10,000 bars. A bar at 12:00 has plenty of history behind it to calculate a valid Z-score.
Chart B (Switched back): MT5 often loads history in chunks. If it only initially loads 500 bars, the Z-score for those same bars might be calculated differently (or not at all) because the "window" of 60 bars looks different or hasn't loaded yet.
Object Persistence Bug (ObjectFind): In your code, you check if(ObjectFind(0, name) < 0) before creating/setting properties. If a bubble was created during a "partial" history load (with an incorrect Z-score), it will never be updated when the full history arrives because the object already exists.

Binary vs. Fractional Delta (Method Jump): There is a logical discrepancy in how bar_delta is calculated:

On a 1-minute chart, the code uses the mfm (Money Flow Multiplier) estimation (Fractional Delta).
On a 5-minute chart (with InpTFLower = M1), it sums up 1-minute bars using a simple Close >= Open logic (Binary Delta).
The sum of Binary Deltas for a 5m bar will not match the Fractional Delta estimation for that same 5m bar.
prev_calculated Optimization Lag: When history is prepended (scrolling back or syncing), the indices in your buffers shift. The code currently only calculates new bars at the end. This means historical bubbles aren't recalculated even if the data context around them changes.

Proposed Fixes
I will modify the indicator to:

Force Recalculation: Always refresh the last InpLenStat * 2 bars to ensure synchronization.
Update Existing Objects: Remove the ObjectFind check so that properties (color, tooltip, price) are updated if the calculation improves.
Synchronize Data Source: Add a "Data Quality" check to ensure lower timeframe data is actually ready before using it.
Stable Indexing: Use a more robust way to handle the start of the calculation.
I'll start by preparing an implementation plan.

Implementation Plan - Fixing Data Discrepancy in QuantMaster Liquidity
Proposed Changes
[Component Name]
[MODIFY]

QuantMasterLiquidityEvents.mq5
Logic Refresh: Modify OnCalculate to recalculate a larger buffer of bars (e.g., InpLenStat * 2) instead of just prev_calculated - 1. This ensures that as history syncs in the background, the Z-scores "heal" themselves.
Object Management: Change the object creation logic to always update properties (ObjectSet...) even if the object already exists. This ensures that if a Z-score changes after more history is loaded, the bubble size and color update accordingly.
Delta Consistency: Harmonize the delta calculation logic to reduce the gap between the mfm estimation and the sub-bar aggregation.
Data Sync Check: Add a check using SeriesInfoInteger to verify that the requested timeframe is synchronized before proceeding with CopyRates.

Pelle wrote: Fri May 15, 2026 2:10 am AI have analyzed (AI) the code and your description of the discrepancy. This is a classic issue in MQL5 development related to History Synchronization and State Management when switching timeframes.

The Root Causes
Context-Dependent Z-Scores (The "Start of History" Problem): Your Z-score calculation depends on a 60-bar lookback (InpLenStat). In MT5, index 0 is always the oldest bar currently loaded on the chart.

Chart A (Stayed on 1m): Might have 10,000 bars. A bar at 12:00 has plenty of history behind it to calculate a valid Z-score.
Chart B (Switched back): MT5 often loads history in chunks. If it only initially loads 500 bars, the Z-score for those same bars might be calculated differently (or not at all) because the "window" of 60 bars looks different or hasn't loaded yet.
Object Persistence Bug (ObjectFind): In your code, you check if(ObjectFind(0, name) < 0) before creating/setting properties. If a bubble was created during a "partial" history load (with an incorrect Z-score), it will never be updated when the full history arrives because the object already exists.

Binary vs. Fractional Delta (Method Jump): There is a logical discrepancy in how bar_delta is calculated:

On a 1-minute chart, the code uses the mfm (Money Flow Multiplier) estimation (Fractional Delta).
On a 5-minute chart (with InpTFLower = M1), it sums up 1-minute bars using a simple Close >= Open logic (Binary Delta).
The sum of Binary Deltas for a 5m bar will not match the Fractional Delta estimation for that same 5m bar.
prev_calculated Optimization Lag: When history is prepended (scrolling back or syncing), the indices in your buffers shift. The code currently only calculates new bars at the end. This means historical bubbles aren't recalculated even if the data context around them changes.

Proposed Fixes
I will modify the indicator to:

Force Recalculation: Always refresh the last InpLenStat * 2 bars to ensure synchronization.
Update Existing Objects: Remove the ObjectFind check so that properties (color, tooltip, price) are updated if the calculation improves.
Synchronize Data Source: Add a "Data Quality" check to ensure lower timeframe data is actually ready before using it.
Stable Indexing: Use a more robust way to handle the start of the calculation.
I'll start by preparing an implementation plan.

Implementation Plan - Fixing Data Discrepancy in QuantMaster Liquidity
Proposed Changes
[Component Name]
[MODIFY]

QuantMasterLiquidityEvents.mq5
Logic Refresh: Modify OnCalculate to recalculate a larger buffer of bars (e.g., InpLenStat * 2) instead of just prev_calculated - 1. This ensures that as history syncs in the background, the Z-scores "heal" themselves.
Object Management: Change the object creation logic to always update properties (ObjectSet...) even if the object already exists. This ensures that if a Z-score changes after more history is loaded, the bubble size and color update accordingly.
Delta Consistency: Harmonize the delta calculation logic to reduce the gap between the mfm estimation and the sub-bar aggregation.
Data Sync Check: Add a check using SeriesInfoInteger to verify that the requested timeframe is synchronized before proceeding with CopyRates.

Could you send me a fixed version? I tried using AI to fix it today, but it gave me an error and didn't show any results. Sorry, professional tasks should be handled by professionals; I'm too amateurish.