Experts: Larry Williams XGBoost Onnx

New comment
 

Larry Williams XGBoost Onnx:

adoption of Larry William's method using AI Time-Series XGBoost

Larry Williams XGBoost Onnx

Author: Taufiqurrachman Assauqi

 
I saw your code, it's very simple and useful, but I would like to suggest it without ONNX model, so it can be run standalone in MT5 and MT5 VPS also, use normal CSV file for data, don't use ONNX model, if you are using ONNX model people forced to know about Python language also. I am professional coder since 1992 and I know Python also but always I like my software to be run standalone without any external support and it's best method and minimum requirement, I suggest you to develop it with auto self learning feature, so when any user implement it first time EA find data for that symbol and if doesn't find create data file automatically, no Python requirement, no self learning requirement. it's quiet simple and super also. Thanks for this code.
 
allvpsmt5 #:
I saw your code, it's very simple and useful, but I would like to suggest it without ONNX model, so it can be run standalone in MT5 and MT5 VPS also, use normal CSV file for data, don't use ONNX model, if you are using ONNX model people forced to know about Python language also. I am professional coder since 1992 and I know Python also but always I like my software to be run standalone without any external support and it's best method and minimum requirement, I suggest you to develop it with auto self learning feature, so when any user implement it first time EA find data for that symbol and if doesn't find create data file automatically, no Python requirement, no self learning requirement. it's quiet simple and super also. Thanks for this code.
Hello mate, may i ask if you have a article regarding this method? i'd like to know to workflow regarding the way to implement machine learning without onnx
 
Nice project. Am I right in thinking you had it train over the whole period you subsequently ran your MT5 backtest? This obviously makes it a perfect graph as it knows the best settings to use, over the period you ran it on. 
I would strongly suggest doing the following:
  • Proper Time-Based Split: Train (2020-2023), Validate (2024), Test (2025-2026) - no overlap​ (<--Out-of-Sample Validation)
  • Multiple Instruments: Test across multiple markets, not just 1
Good luck
 
Larry williams speak about waves .
s
you need to define a fractal pattern where  4 difiiferent zigzag meet then 2 fastest move to create a 123 pattern wave for trade .
 
I wonder what the equity curve and all the other trading statics would look like without the XGBoost and the AI part?İt looks like XGBoost did give an extra edge.
 

  1. unzip larry_william.zip (This is the process of unzipping larry_william.zip.)

  2. run command pip install -r requirements.txt
    Run the command `pip install -r requirements.txt`

  3. First, open MetaTrader 5.

  4. run python download_csv_metatrader5.py
    Run Python download_csv_metatrader5.py

  5. run python train_larry_williams.py
    Run Python train_larry_williams.py

  6. run python convert_onnx_larry.py
    Run Python convert_onnx_larry.py

Could you explain in detail how this compressed file works after being extracted? It doesn't do anything when I open it after extraction.
New comment