New paper out
🎉Our latest research emerged from a simple yet profound question: How can we truly understand the heartbeat of cryptocurrency markets?
Using a Directional Change (DC) methodology, we've developed a meta-learning framework that implementing adaptive parameter tuning and dynamic feature selection for crypto algorithmic trading.
💡 Key scientific innovations:
-
🔹 Event-driven price sampling (DC) replacing constant frequency methods. It means DC method samples prices based on significant market movements rather than fixed time intervals, enhancing its flexibility in responding to market volatility
-
🔸Meta-learning model trained on multi-dimensional feature sets in different regime
-
🔻 Return-weighted training approach enhancing model sensitivity
Key findings
⚡️:
🔻 Four group of features ( almost 300 features) show that there is no single group work best in all situation!
🔍 Analysis of the meta-learning models shows a low correlation between outputs from models trained on distinct feature categories, suggesting that each group captures a unique aspect of parameter selection in different conditions.
🔸 Feature behavior analysis reveals that different categories were most informative at various points in the history, with strategy meta-information and DC indicators standing out as the most impactful features
Interested in diving deeper? read the paper here:
linkedin 💠Group :
@machinelearningnetChannel:
@machinelearningnet2
