Discussing the article: "Using association rules in Forex data analysis"
Apparently, it is assumed that the reader must already have some knowledge of such a method, and if not?
I don't understand the metrics that are mentioned, in particular:
Lift has become my favourite indicator. After hundreds of hours of testing, I noticed a pattern - rules with lift above 1.5 really work in the real market. This discovery seriously influenced my approach to signal filtering.
If I understood the method correctly, correlating signals are searched for in quantum segments. But I didn't understand the next step. What is the target one? I assume that the resulting rules are checked against the target and evaluated against the metrics.
If so, it echoes my method, and it's interesting to evaluate performance and efficiency.
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
Check out the new article: Using association rules in Forex data analysis.
I have been working with data for a long time and have noticed that many successful ideas come from related areas. Today I want to share my experience of using association rules in trading. This method has proven itself in retail analytics, allowing us to find connections between purchases, transactions, price movements and future supply and demand. What if we apply it to the foreign exchange market?
The basic idea is simple - we are looking for stable patterns of price behavior, indicators and their combinations. For example, how often does a rise in EURUSD follow a fall in USDJPY? Or what conditions most often precede strong moves?
In this article, I will show the complete process of creating a trading system based on this idea. We will:
Why this particular stack? MQL5 is great for working with stock exchange data and trading automation. In turn, Python provides powerful tools for analysis. From my experience, I can say that such a combination is very effective for developing trading systems.
The first step in the analysis is to understand the distribution of the main metrics of the rules found. The distribution graph of 'support', 'confidence', 'lift' and 'leverage' helps to evaluate the quality of the found rules and, if necessary, adjust the algorithm parameters.
Author: Yevgeniy Koshtenko