Skip to content
All library documents

Using DSU and Deep Belief Networks to Filter Volatility Breakouts

Article MQL5 articles
Read the original

Summary

The article describes a custom MQL5 breakout signal that combines Disjoint Set Union (DSU) clustering with a Deep Belief Network (DBN). DSU groups adjacent high-volatility bars into time-bounded clusters using ATR expansion and Bollinger Band width as candidate shock conditions. A stacked restricted Boltzmann machine then scores each cluster, with the signal passing only when its bounded output exceeds a chosen threshold. The proposed aim is to distinguish lasting regime changes from temporary spikes such as stop hunts.

The article contrasts this localized, time-series approach with a prior B-Tree and Bayesian model intended for cross-asset relationships. Its reported tests are mixed: one algorithm-only configuration produced 68 trades, about 5% net profit, a drawdown close to 20%, and a profit factor of 1.09; the combined filter configuration made four trades, only one profitable, with lower relative drawdown. These limited results do not establish an edge. The method is presented for volatile, liquid breakout settings and may struggle in choppy ranges; the article also cautions that longer tests across symbols are needed.

Key ideas

  • DSU can join adjacent high-volatility bars into discrete time-series clusters.
  • ATR expansion and Bollinger Band width provide candidate boundaries for volatility shocks.
  • A Deep Belief Network is proposed as a filter for separating persistent breakouts from transient spikes.
  • The reported tests show a trade-off between modest gains with substantial drawdown and sparse trades under combined filtering.
  • The article frames the approach as experimental and calls for longer testing across symbols.

Tags

BreakoutVolatilityMachine learningTechnical indicatorsBacktesting

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.