Filtering TD Sequential Signals with Neural Networks and Market Context
Summary
The article proposes using neural networks to filter signals from an adaptive version of Thomas DeMark's Sequential strategy. It focuses on Setup and Intersection signals near potential turning points, treating the network as a decision aid rather than a standalone trading strategy. Separate models classify buy and sell signals, while the signal window provides the points at which the market is evaluated.
The proposed training approach groups examples by daily context, derived from changes in price, volume, and open interest, and recommends matching training days to expected trading conditions. It also discusses output labels beyond simple profit or loss, such as whether a pullback occurs or a target is reached, and cautions that imperfect labels and limited model validity are unavoidable. The examples use GBP/USD and manually updated volume and open-interest data. The text explains a modeling framework, but does not provide enough reported performance evidence here to establish profitability; daily context, sample selection, and changing market behavior constrain generalization.
Key ideas
- The neural network is presented as a filter for TD Sequential signals, not as a complete trading strategy.
- Buy and sell signals are modeled separately to classify their validity.
- Training examples are grouped by daily context based on price, volume, and open-interest changes.
- Output labels can represent different market outcomes, not only whether a trade was profitable.
- Model confidence and usefulness may decay as market conditions shift, so broad generalization is not guaranteed.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.