Searching Candlestick Patterns with Brute Force Optimization
Summary
The article describes a brute force method for searching price patterns and generating trading robots from the selected parameters. It represents a signal as a multivariable polynomial expansion whose inputs are candle price changes across several bars. Candidate coefficient arrays are searched in two stages: first for next-bar predictive measures such as expected payoff or profit factor, then through simulated orders and balance curves, with signal thresholds adjusted to examine the tradeoff between selectivity and signal count.
The author contrasts this compact parameter search with neural network configurations and discusses generating MT4 or MT5 EAs from a template. The reported process is a demonstration, not a validated trading result: the article says higher timeframes were not tested because of runtime demands and notes that forward demo testing had not yet been completed. Searching many candidates against historical quotes risks selecting patterns that do not persist, and the author characterizes the work as an introduction rather than a complete analysis.
Key ideas
- Candidate strategies are encoded as coefficient arrays in a multivariable polynomial of candle price changes.
- The search first scores next-bar predictions, then simulates trading behavior over the loaded data.
- Raising the signal threshold is intended to favor stronger signals while reducing their frequency.
- The selected coefficients can be inserted into a generated EA template for MetaTrader.
- The article does not establish out-of-sample durability, and higher timeframe and demo testing remain incomplete.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.