**Updated** update...with attached V209 ML_Trading_Framework_PRO_V209_Linux.py.txt Aside from all Float32 and tuple errors I finally managed to get the V209 to 'work'. The previous update I was trying to figure out why the gate score (F1) always ...

Experimental MT5 EA using Gemini Flash API - GeminiCommanderEA_v1

No offence taken. This whole thing started as a small project to integrate Gemini to help automate trade management. I wasn’t happy with the MT5 backtester, so I built my own — and naturally uncovered a ton of flaws in both the tester and my ...

Experimental MT5 EA using Gemini Flash API - GeminiCommanderEA_v1

The V208 testing has turned into somewhat of a behemoth of a task... It went from some simple fixes and debugging, to what I am testing now and most likely well into the weekend with V209 to replace V208. I have incorporated LSTM and have an ...

Experimental MT5 EA using Gemini Flash API - GeminiCommanderEA_v1

So if the model score (MAR ratio) is really good, why don't you just change the MIN Gate? I'm glad you asked the question. The current V208 I am testing with the changes I listed in the previous post does this. You'll see in the below output some ...

Experimental MT5 EA using Gemini Flash API - GeminiCommanderEA_v1

Methodology to diagnose the performance issues Due to a number of issue I have had to create controlled experiments to systematically isolate the source of the low F1 scores. At the bottom I have provided a bit more context. Attached are the two ...

Experimental MT5 EA using Gemini Flash API - GeminiCommanderEA_v1

ONNX Machine learning pipeline in python For training a neural network model to predict trade suitability scores for forex trading (specifically GBPUSD in this case). Modify this however you want. It should produce two files. In may case based on ...

Experimental MT5 EA using Gemini Flash API - GeminiCommanderEA_v1

What I have changed or modified in V208 so far: Still testing V208 as I found a few bugs and whilst fixing that I 'improved' part of the script. It is a fine line between just fixing a version and getting to a point where you have to just release a ...

Experimental MT5 EA using Gemini Flash API - GeminiCommanderEA_v1

This 'codes' all the strategies automatically using the AI and saves them based on their meta labels for different market regimes and saves them in the champion json, but this is only created after each run is completed and if a 'champion is found. ...

Experimental MT5 EA using Gemini Flash API - GeminiCommanderEA_v1

The fun fact no one asked for... I'm testing and will introduce the following into the framework: 1. Initial Strategy Selection is Capital-Aware (Tagging) The script ensures that the initial strategy chosen at the beginning of a run is appropriate ...

Experimental MT5 EA using Gemini Flash API - GeminiCommanderEA_v1

Still testing caveat... Here's V208 for anyone willing to try it. Feedback welcome.

Experimental MT5 EA using Gemini Flash API - GeminiCommanderEA_v1

As an FYI. It is good practice if you are testing the framework and make changes, always delete the Results folder before starting again. Why? If you have any previous results or strategies that have been developed from the AI, it is most likely ...

Experimental MT5 EA using Gemini Flash API - GeminiCommanderEA_v1

Still testing. This is a hobby in between my normal work so I don't have a lot of time to react to issues during working hours. I hope this run will be the last before I upload it. Screen of the new MAR ratio score (replaced Sharpe) of 6.3864. ...

Experimental MT5 EA using Gemini Flash API - GeminiCommanderEA_v1

Sorry, Just saw this. And I hope the below answers your question, most likely in a round about way. As much as simple indicators and buying randomly seems like a good idea that could work for a bit this project (script) actually tests those ...

Experimental MT5 EA using Gemini Flash API - GeminiCommanderEA_v1

Sorry for the late reply. I don't normally read through the thread after I post. 1. prefix/suffix - Yes this can be added easily through prompting to Gemini with the first API call that already checks the assets (csv files) in the directory during ...

Experimental MT5 EA using Gemini Flash API - GeminiCommanderEA_v1

If my ideas don't work then certainly switch back. Sometimes keeping it simple is the best path to take. It's a matter of trial and error. I have had more errors in my time.

XAUUSD and Machine Learning Trading

Slightly off topic but does tie in with the Gemini EA and any EA being developed. Why use external News API's and Calendar events when you can use Google to do all the heavy lifting in the EA just through prompting? Through the course of this thread ...

Experimental MT5 EA using Gemini Flash API - GeminiCommanderEA_v1

Great start, I have some ideas below. Good luck with the testing. Feedback and Summary of Changes This enhanced script improves upon the original logic in several key areas, focusing on robustness, clarity, and better machine-learning practices. 1. ...

XAUUSD and Machine Learning Trading

If you can provide the full backtest report (export it) then I could assess the data a bit more. My strongest recommendation is to AVOID changing the core logic or trying to "improve" the win rate. You have a proven, 8-year model. Tinkering with it ...

Request for Feedback on EA Performance (8 Years of Backtesting)

I had to ask.. 1. User Base Estimates for Competing Frameworks We can use public metrics from GitHub (stars, forks, contributors) and community indicators to create a strong proxy for their popularity and adoption. Here are the approximate metrics ...

Experimental MT5 EA using Gemini Flash API - GeminiCommanderEA_v1

Whatever csv assets you have in the directory the framework will automatically load. It is best to have at least two different assets with three different timeframes in your directory as it is designed for multi-timeframe training. In my case I have ...

Experimental MT5 EA using Gemini Flash API - GeminiCommanderEA_v1