Discussing the article: "Data Science and ML (Part 37): Using Candlestick patterns and AI to beat the market"

 

Check out the new article: Data Science and ML (Part 37): Using Candlestick patterns and AI to beat the market.

Candlestick patterns help traders understand market psychology and identify trends in financial markets, they enable more informed trading decisions that can lead to better outcomes. In this article, we will explore how to use candlestick patterns with AI models to achieve optimal trading performance.

The Basics of a Candlestick

The candlesticks' shadows or wicks show the day's high and low prices and how they compare to Open and Close prices. The shape of the candle varies based on the relationship between the candle's OHLC (Open, High, Low, and Close) prices. There are a plenty of candlestick patterns introduced by Munehisa back in the day and some emerged by traders recently.

Based on these candlestick patterns, we are going to identify, collect, and apply these candlestick patterns to machine learning models and observe how these candlestick patterns can add value to our AI-based trading models and see if they can aid us in beating the financial markets.

Author: Omega J Msigwa

 
Awesome article.
 
Love this attempt! Very interesting approach of labeling candle stick patterns for ML.
 
Using candlestick configuration instead of underlying price action is similar to using 256-color indexed image instead of true color image - you loose a lot of nuances which can be important. Moreother, depending from the time offset of a broker candlestick patterns can completely change on the same history of quotes. And even if you'd not use a large timeframe (such as D1 (in your case) or H4), simple artificial shifting of time (minutes) inside an hour will produce completely different formations of candlesticks. Unreliable, does not produce any value.
 
Stanislav Korotky #:
Using candlestick configuration instead of underlying price action is similar to using 256-color indexed image instead of true color image - you loose a lot of nuances which can be important. Moreother, depending from the time offset of a broker candlestick patterns can completely change on the same history of quotes. And even if you'd not use a large timeframe (such as D1 (in your case) or H4), simple artificial shifting of time (minutes) inside an hour will produce completely different formations of candlesticks. Unreliable, does not produce any value.

I explained this in the article already.

Right now, we are only considering detecting candlestick patterns and their signals based on their appearance, but the right way to extract the signals from a candlestick according to my sources, must include trend detection for example, for a hammer to be considered a bullish signal it has to appear on a downtrend.

Trend is a crucial part of the equation that you might want to consider if you want to take this project further.

Price action is important, no denying that.

 
Omega J Msigwa #:

I explained this in the article already.

Right now, we are only considering detecting candlestick patterns and their signals based on their appearance, but the right way to extract the signals from a candlestick according to my sources, must include trend detection for example, for a hammer to be considered a bullish signal it has to appear on a downtrend.

Trend is a crucial part of the equation that you might want to consider if you want to take this project further.

Price action is important, no denying that.

Your quote is not related to main idea of my point.

This website uses cookies. Learn more about our Cookies Policy.