Discussing the article: "Combining LLM, CatBoost, and Quantum Computing into a Unified Trading System"

New comment
 

Check out the new article: Combining LLM, CatBoost, and Quantum Computing into a Unified Trading System.

The article proposes a synthesis of new technologies to overcome the limitations of classical indicators in market data analytics. It shows how language models and quantum encoding can reveal hidden market patterns that traditional methods overlook. The experiment confirms the value of new technologies and proposes an updated analysis methodology aligned with the current state of computational innovation.

An analysis of a loss-making situation in the EURUSD market revealed critical shortcomings in classical technical analysis. RSI stood at 72.3, indicating overbought conditions; MACD showed a bearish divergence; and the stochastic oscillator was in the oversold zone. A neural network trained on these indicators predicted a downward move with 67% confidence. However, the short position that was opened resulted in a USD 340 loss, pointing to a fundamental problem not with specific indicators or the model, but with the very approach to market analysis.

The project is implemented in a single Python file (1,328 lines), which includes model training, feature generation, backtesting, forward testing, and integration with MetaTrader 5 for live trading. The results are reproducible; the parameters were not fitted to historical data.

The system already trades eight currency pairs and can easily be scaled to include cryptocurrencies, indices, and commodities. Quantum encoding is universal and suitable for any time series, and real IBM quantum processors could be used to generate critical signals in the future.

Author: Yevgeniy Koshtenko

 
He’s behaving rather interestingly on the demo account. Out of five trades, four were closed at a loss of $9, and one was profitable, yielding a gain of $11 over the course of a day. And he proudly declared that he was in profit.
 

Hi @Yevgeniy Koshtenko,

Good article, but why didn’t you create a LoRa model for the LLM instead of a reinforcement learning model?

 

Oh dear… History keeps repeating itself. Every coder sees themselves as an unrecognised god who’s bound to find and mess with the Grail on the market. And, most surprisingly, they’re already treading well-worn paths.
Although if I were to point out yet again that the mathematical (one might even say geometric) Holy Grail was described in sufficient detail 100 years ago, they wouldn’t believe me again. It’s not fashionable and it’s not on trend. Quantum computing and all sorts of other new-fangled nonsense are all the rage. As the saying goes – the mice cried and pricked themselves, but carried on munching on the cactus. ))
Let’s wish the next Einstein the best of luck! ;) I guarantee a zero result with a probability of 1,000 per cent.

 
Bogard_11 geometric) Holy Grail was described in considerable detail 100 years ago, they still wouldn’t believe me. It’s not fashionable and it’s not on trend. Quantum computing and all sorts of other new-fangled nonsense are all the rage. As the saying goes – the mice cried, pricked themselves, but carried on munching on the cactus. ))
Let’s wish the next Einstein the best of luck! ;) I guarantee a zero result with a probability of 1,000 per cent.

You’re on a coders’ website; here, a full stop acts as a separator.

So you’re guaranteeing a 1% chance of a zero result and a 99% chance of a positive one.
 
Ivan Butko #:

You’re on a coding website; here, the full stop acts as a separator.

That’s why you guarantee 1 per cent of results will be zero and 99 per cent will be profitable.
It actually uses a full stop as a separator (and a decimal comma). But you meant to say that the full stop is the decimal point.
 
@Yevgeniy Koshtenko, thank you for the article and the example – a working trading model in Python for MT5. I ran your example on a local Gemini instance, and the backtest showed 33 profitable trades out of 34. I ran the code through DeepSick, and it identified a few instances where the code looks into the future during backtesting. However, the example itself works well and is useful.
 
Bogard_11 #:
Although if I were to repeat once again that the mathematical (one might even say geometric) Holy Grail was described in considerable detail 100 years ago, people still wouldn’t believe me.
I’m prepared to believe it. Please could you tell me where to find a description of this Holy Grail?
New comment