A Markov Transition Matrix and Neural Network for Market-State Forecasting
Summary
This article describes an Expert Advisor that classifies daily market conditions as flat, uptrend, or downtrend by comparing price changes with the Average True Range. It counts historical transitions among these states to estimate a 3-by-3 Markov transition matrix, assigning equal transition probabilities when a state has no observations. A multilayer perceptron then uses the matrix entries to forecast future price movement. The article also discusses hedging and risk controls as parts of the wider EA, though the supplied text does not fully specify those rules.
The author reports experimental performance figures, including annual return, maximum drawdown, Sharpe ratio, and profitable-trade share, and claims operation across quiet and volatile conditions. These are presented without enough information in the excerpt to assess data selection, out-of-sample testing, costs, or statistical reliability. The state definition, transition counts, and neural model are implementation choices that require validation and tuning; the article itself cautions that the system is not a guaranteed source of profit.
Key ideas
- Market states are defined by comparing daily price changes with ATR-scaled thresholds.
- Historical counts between flat, rising, and falling states form a transition probability matrix.
- Unobserved states receive equal transition probabilities rather than all-zero rows.
- A multilayer perceptron processes the matrix to predict future price direction.
- Reported performance lacks sufficient testing details in the excerpt to establish robustness.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.