Forex Price Forecasting with Random Baselines and Trend Indicators
Summary
The article surveys several price-movement models for Forex, beginning with random trade direction as a baseline. It discusses how spread can undermine frequent position changes, then considers fixed holding periods and stop orders. It also describes searching random-number seeds for a favorable trading sequence with a secretary-problem stopping rule, while acknowledging that this search can be time-consuming. A further approach estimates future increments from an empirical distribution, trimming extreme observations to form a forecast channel rather than assuming a textbook distribution.
The later sections derive a trend-rate indicator from changes between consecutive prices and present a model relating traveled price distance to time. The latter estimates an exponent to distinguish flat-like behavior from linear trending and motivates weighted indicators using figurate numbers. The article reports that some optimized random strategies showed winning configurations, while results for the indicator tests were not impressive. It does not establish robust predictive power: optimization over random seeds risks selection effects, empirical increment channels rely on historical behavior, and the proposed indicators are explicitly described as needing further study.
Key ideas
- Random trade direction can serve as a baseline, but trading costs such as spread can make frequent turnover unfavorable.
- The article proposes selecting among random sequences with a stopping rule inspired by the secretary problem.
- An empirical distribution of price increments can be trimmed to construct a forecast channel.
- A trend-rate indicator is derived from changes across bars and can generate signals when its sign changes.
- A distance-versus-time exponent is used to characterize price movement, but the article reports weak test results for some indicators.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.