Discussing the article: "Predicting Renko Bars with CatBoost AI"

 

Check out the new article: Predicting Renko Bars with CatBoost AI.

How to use Renko bars with AI? Let's look at Renko trading on Forex with forecast accuracy of up to 59.27%. We will explore the benefits of Renko bars for filtering market noise, learn why volume is more important than price patterns, and how to set the optimal Renko block size for EURUSD. This is a step-by-step guide on integrating CatBoost, Python, and MetaTrader 5 to create your own Renko Forex forecasting system. It is ideal for traders looking to go beyond traditional technical analysis.

In the silence of my study, only the monitor screen glows. The MetaTrader 5 terminal produces the latest tick data, and the algorithm I developed methodically transforms the market chaos into an orderly system of Renko bars. Three a.m. is the time when real trading strategies are born. I run my hand over my unshaven chin and take a sip of my now cold coffee, my favorite Carte Noir. The model's accuracy of 59.27% is a victory. A real victory for those who understand the unpredictable nature of financial markets.

I first encountered Renko charts at a conference in Singapore. A Japanese trader, whose name I promised never to reveal, showed me his terminal with unusual rectangular blocks instead of the usual candlesticks. "These are Renko charts," he said, "charts that show only what matters - price movement".

Predicting Renko Bars with CatBoost AI

In a world oversaturated with information, Renko charts offer digital Zen — stripping away noise, time factor and anything else that distracts from the true movement of the market. Each new block is formed only when the price moves a certain distance up or down. There is incredible power hidden in this simplicity.


Author: Yevgeniy Koshtenko

 
You can't get more than 63 per cent. Whatever signs are added. 63 for forex is a fiasco
 
sportoman #:
You can't get more than 63 per cent. Whatever signs are added. 63 for forex is a fiasco

I got more than 75% on Midas (multimodular, where many modules combine their signals into a single signal). But then another problem arises - signals once a month and a half))))The more reliable the signal, the less often it occurs...That's why I started to forget about high winrate in recent robots, I try to predict the depth of movements to take the risk to profit in 1:3, 1:4, etc. I set up marks in such a way that they give bonuses to models through DQN for correct marking of deep movements....This eventually resulted in the robot LSTM Europe.

 
Yevgeniy Koshtenko #:

I got above 75% on Midas (multi-modular, where many modules combine their signals into a single signal). But then another problem arises - signals once a month and a half))))The more reliable the signal, the less often it occurs...That's why I started to forget about high profitability in recent robots, I try to predict the depth of movements to take the risk to profit in 1:3, 1:4, etc. I set up marks in such a way that they give bonuses to models through DQN for correct marking of deep movements....This eventually resulted in the robot LSTM Europe.

75 on LTSM at lookback = 5 bars_ahead = 1 but this is for binary trading, not for forex.

 

So where's the AI? Or have you elevated catbust to that rank?

You should do a test or something. 58% is okay? This Accuracy is just a benchmark. The main indicator is the balance on the test based on the signals received.

It's not interesting.