Direct answer
Fractal Dimension Index can be combined with other non-duplicative analytical inputs—such as trend, volatility, or regime descriptors—so the overall analysis uses complementary information rather than repeating the same signal idea. The main practical risk is correlated-input dependence: if your added variables are derived from the same price path features that drive fractal dimension, they can fail together.
Mechanism or definition
Fractal Dimension Index (FDI) is a numerical measure intended to describe the apparent complexity of a time series, often with attention to how structure changes across scales. In plain terms, it turns a sequence of observations (for example, a price or returns series) into a complexity-like statistic.
“Combined with” can mean several things:
- Feature stacking in analysis: use FDI alongside other computed features in a decision framework.
- Conditioning: compare another metric differently depending on whether FDI is in certain ranges.
- Context labeling: treat FDI as one axis that characterizes the environment while other axes describe direction or variability.
To keep the mechanics clear, assume you have a time series X and compute FDI using a specific method (window length, sampling frequency, preprocessing such as returns vs. prices). Any combination should use the same time alignment and clearly stated preprocessing assumptions, because FDI values can change if you change inputs or calculation settings.
Evidence or example (with explicit assumptions)
Consider a simple analytical setup with no real-time data claims:
- Assume you compute FDI from a returns series over a rolling window of length W.
- Add a volatility feature, for example the rolling standard deviation of returns over the same window W.
- Add a trend feature, for example the slope of a fitted line over the same window W.
Why this can be non-duplicative: FDI is meant to characterize complexity across scales, while volatility is a measure of spread/amplitude and trend is a measure of directional persistence. In an informative case, you might observe:
- High FDI + high volatility: complexity rises and movement becomes more erratic.
- High FDI + low volatility: complexity rises without large amplitude, which can suggest choppy micro-structure.
- Low FDI + strong trend: the series appears more regular while direction persists.
Material limitation: these features are computed from the same underlying series, so independence is not guaranteed. If trend, volatility, and FDI all respond to the same underlying regime shift (for instance, a structural change in how prices evolve), the “combination” may mostly reflect one shared driver.
A second example is conditioning:
- Compute another metric (such as a momentum-like statistic) and interpret it differently when FDI indicates a more complex environment.
- This uses FDI as a context variable rather than a standalone predictor.
Limitations and risks
At least one material failure mode is correlated-input risk. If the extra variables are mathematically or practically linked to the same properties that FDI captures, your combined framework may overstate robustness: performance improvements can come from the shared dependency, not from complementary information.
Other limitations to expect:
- Method sensitivity: FDI typically depends on computation choices (window size, preprocessing, sampling). Changing these can change the numeric values.
- Regime dependence: relationships between complexity and other features may vary by market condition. Historical patterns do not ensure future behavior.
- Operational risk: if your time series is noisy, resampled inconsistently, or includes data-quality issues, FDI can react in ways that look meaningful but reflect preprocessing artifacts.
- Divergence/reversal: when FDI changes abruptly relative to prior behavior, it may indicate a regime shift, but it can also reflect calculation sensitivity or short-window instability.
Finally, outcomes depend on execution details, costs, and the broader environment. Even with careful verification, you cannot assume predictive accuracy or safety.
Verification or next question
To independently verify claims about combining FDI with other inputs, use a process that separates mechanics from results:
- Keep the FDI method fixed and document preprocessing assumptions.
- Test combinations on multiple, non-overlapping periods rather than one stretch of data.
- Check whether the added features truly add new information by measuring how much the combined approach changes when you remove one input.
Next question to resolve: which specific FDI method and parameters are you using (window length and whether you compute from prices or returns)?