Using Multi and Relative Domains to Forecast Price Range Changes
Summary
The article applies category theory concepts, especially multi-domains and relative domains, to forecasting changes in price bar ranges. A multi-domain allows repeated observations, so it can represent lagged price data even when multiple observations share the same value. The proposed method compares a new input with stored multidimensional observations, selects the nearest match, and uses its associated normalized range change as a projection. For multiple projections, the described trailing stop can use their minimum, mean, or maximum.
The article also introduces relative domains as a way to relate observations across datasets, and sketches possible extensions to volatility classification, index weighting, and portfolio risk analysis. It reports a EURGBP daily test setup using an Awesome Oscillator signal and compares it with a moving-average trailing stop, but the supplied text does not include the reports or enough detail to assess performance. The approach is presented as exploratory: results depend on the training observations, distance matching, lag selection, and projection aggregation. The article offers no demonstrated evidence that these choices generalize to other markets or periods.
Key ideas
- A multi-domain represents repeated observations while retaining their multiplicities.
- The proposed forecast matches current lagged price features to the closest stored multidimensional observation.
- Associated normalized range changes can inform trailing stop distances.
- When two projections are available, the method allows minimum, mean, or maximum aggregation.
- The article presents the approach as exploratory and does not provide enough reported results to establish its effectiveness.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.