Discussing the article: "Triangular arbitrage with predictions"

New comment
 

Check out the new article: Triangular arbitrage with predictions.

This article simplifies triangular arbitrage, showing you how to use predictions and specialized software to trade currencies smarter, even if you're new to the market. Ready to trade with expertise?

Arbitrage is very curious, it's been prohibited from the bookies of sports betting. Imagine you have some winning odds of 1.25 for Real Madrid to win the champions 2024, and Borussia Dortmund has 3.60 odds, that means Madrid has 100/1.25 =  80 % of probabilities to win and Borussia 27.7  % to win. If you add those two, you have 107.7%, that is because bookies want to win money and that over 100 % is their commission. But, imagine you find Bookie number 2 and hey offer odds for Borussia of 19% probabilities to win, odds of 5.26. Then you could bet in Bookie number 1 to Real Madrid and Bookie number 2 for Borussia, and if you bet the appropriate quantity to each team, you will win money in the game, because both add less than 100%. This is a simple way to explain why its prohibited in sports betting and what is arbitrage.

Imagine you are a "legal" person and you don't want to have your sports account closed by doing arbitrage, you know that even if you bet for Madrid, you could do "legal" arbitrage if you waited for minute 70' of the game if draw or wait to Real Madrid to score to have those odds for Borussia and have a win win... this seams a bit risky, but here is where we can take advantage of Deep Learning, we know Real Madrid is gonna score, so you are gonna have those odds with a 98 % of probabilities (we know this with cointegration between the predictions and the real values). This is what's new with Deep Learning and Arbitrage.

Author: Javier Santiago Gaston De Iriarte Cabrera

 
MetaQuotes:

Check out the new article: Triangular arbitrage with predictions.

Author: Javier Santiago Gaston De Iriarte Cabrera

Thanks Javier, for the article. Very educative. 
 
MetaQuotes:

Check out the new article: Triangular arbitrage with predictions.

Author: Javier Santiago Gaston De Iriarte Cabrera

Thank you, this is enlightening. 

 
Isaac Amo #:
Thanks Javier, for the article. Very educative. 

You're welcome! Thanks'!

 
Clemence Benjamin #:

Thank you, this is enlightening. 

Thanks!

 

Thanks! Very nice article.
I want to ask you a question: do you know how to sell EAs that incorporate ONNX models?

I was thinking about using OnnxCreateFromBuffer with an array of bytes representing the ONNX model, but I'm not sure if that's the best solution.

Thank you!

 
Emanuele Mastronardi #:

Thanks! Very nice article.
I want to ask you a question: do you know how to sell EAs that incorporate ONNX models?

I was thinking about using OnnxCreateFromBuffer with an array of bytes representing the ONNX model, but I'm not sure if that's the best solution.

Thank you!

Yes, I've tried it some time ... but don't know why some EA's get lots of sells and other don't.

The problem with a EA with models, is that you have to keep it updated, so you can't sell a bot, you have to sell a subscription.

Another problem is that you have to have a model for each time period and each symbol.

I hope this answers some questions.

ONNX models are used to transport to other languages. You can also create a NN with mql5.

But yes, I would also like to sell bots.

 
Javier Santiago Gaston De Iriarte Cabrera #:

Yes, I've tried it some time ... but don't know why some EA's get lots of sells and other don't.

The problem with a EA with models, is that you have to keep it updated, so you can't sell a bot, you have to sell a subscription.

Another problem is that you have to have a model for each time period and each symbol.

I hope this answers some questions.

ONNX models are used to transport to other languages. You can also create a NN with mql5.

But yes, I would also like to sell bots.

hello! thank you for your reply.
i finally managed to import the onnx model by converting it to bytes in python and including it directly as a uchar array in mql5.
Yes, certainly including an onnx model in ea involves extra care and limitations, although it is very convenient.
As you said it is limited to the timeframe and symbol used during training, which is avoidable if you include a library of a neural network directly in mql5.
We use lstm as a type of neural network and I found a free library on mql5.
Obviously training the network in python brings with it many advantages, but I will try to use that library to make the network "fit" any timeframe and any symbol.
Another advantage is that you could set up automatic re-training every few timeframes.
It would be very nice if you would consider this aspect and do an article!
Thanks again

https://www.mql5.com/en/code/24200

LSTM Neural Network
LSTM Neural Network
  • www.mql5.com
Long Short-Term Memory Neural Network - for time series analysis.
 
Gianvito Fiume #:
hello! thank you for your reply.
i finally managed to import the onnx model by converting it to bytes in python and including it directly as a uchar array in mql5.
Yes, certainly including an onnx model in ea involves extra care and limitations, although it is very convenient.
As you said it is limited to the timeframe and symbol used during training, which is avoidable if you include a library of a neural network directly in mql5.
We use lstm as a type of neural network and I found a free library on mql5.
Obviously training the network in python brings with it many advantages, but I will try to use that library to make the network "fit" any timeframe and any symbol.
Another advantage is that you could set up automatic re-training every few timeframes.
It would be very nice if you would consider this aspect and do an article!
Thanks again

https://www.mql5.com/en/code/24200

Hi, thanks!

In this article,  the EA has uses dll to run a py script, and one of the py scripts is one to create an onnx model. It can help you.

Sentiment Analysis and Deep Learning for Trading with EA and Backtesting with Python - MQL5 Articles

Sentiment Analysis and Deep Learning for Trading with EA and Backtesting with Python
Sentiment Analysis and Deep Learning for Trading with EA and Backtesting with Python
  • www.mql5.com
In this article, we will introduce Sentiment Analysis and ONNX Models with Python to be used in an EA. One script runs a trained ONNX model from TensorFlow for deep learning predictions, while another fetches news headlines and quantifies sentiment using AI.
New comment