Direct answer
To assess ADX (Average Directional Index), you need (1) the price data used to compute it, (2) the method and parameter settings used in the calculation, and (3) checks on provenance and timeliness of that input data. You also need to know the limitations of ADX so you do not treat the value as a standalone guarantee of direction or future performance.
Mechanism and definition: what ADX calculation depends on
ADX is a technical indicator derived from directional movement components and then smoothed over a chosen lookback period. Because the indicator is computed, the “data needed to assess ADX” is mostly about the inputs that feed the computation and the configuration that determines how the computation is performed.
Data inputs to collect:
- The underlying price series (commonly high, low, and close) at a specific time interval (for example, each candle/bar).
- The exact lookback period used for ADX (often expressed as “period”).
- The smoothing approach used by the calculation (for example, the common idea of smoothing directional movement and then smoothing the ADX itself). Exact wording and formulas can vary by implementation.
Data provenance and timeliness checks:
- Source of the price series (where the highs/lows/closes come from) and whether it is based on the same instrument definition across your workflow.
- Timestamp alignment to ensure that the “high/low/close” belong to the same bar boundaries you assume when interpreting ADX.
Stable mechanics vs. variable conditions:
- Stable mechanics are that ADX is computed from directional movement and smoothed; this is a methodological property.
- Variable conditions include the market behavior during the sampled period, the chosen timeframe, and the quality of the input price data.
Evidence or example: what to document so another person can reproduce your ADX
A practical way to assess ADX is to treat it as a reproducibility problem. Even without real-time pricing, you can still verify facts by documenting the inputs and recomputing.
Example of what to record for an independent check:
- Timeframe of the source data (e.g., “one bar per X minutes”).
- Lookback period for ADX.
- Smoothing method as stated by the indicator implementation you used.
- Whether you used raw or adjusted prices (for example, any transformations applied by a platform).
- The date range of the dataset used (start/end timestamps).
Then quality checks on the input series:
- Missing bars: confirm there are no gaps or, if there are, confirm how the calculation handles them.
- Outliers and spikes: verify whether extreme high/low points are real or caused by data errors.
- Consistency: confirm that high/low/close values change coherently with the timeframe boundaries (no off-by-one bar issues).
AFV- or checklist-style confirmation (informational, not predictive):
- Evidence/document: record the exact indicator settings and dataset identity.
- Rode vlaggen (red flags): unexplained differences between two ADX readings using the “same” settings, caused by different price sources, timeframe definitions, or smoothing.
- Klaarcriterium (ready-to-trust criterion): another person can reproduce the indicator values from the same inputs and configuration within expected numerical precision.
Limitations and risks (material failure modes)
ADX assessment can fail in predictable ways. At least one material limitation is that ADX values reflect the behavior of the specific data and method used; changes in timeframe, parameter settings, or price source can change the resulting numbers.
Material limitation / failure modes to consider:
- Methodology mismatch: different implementations may use different smoothing conventions or parameter defaults, so two ADX series may not be comparable.
- Data quality problems: missing bars, misaligned timestamps, or incorrect high/low values can distort the directional movement components.
- Non-transferability: a historical relationship between ADX levels and later outcomes does not establish future results.
- Interpretational overreach: ADX is an indicator of characteristics of price movement over the sampled window; treating it as a standalone signal of future direction is an unsupported leap.
Uncertainty assumptions:
- This explanation assumes no real-time market data is required; it focuses on what information you must have to compute and verify ADX from existing data.
- Outcomes vary with market conditions, costs, execution, and jurisdiction; ADX alone does not remove these uncertainties.
Verification or next question
To verify ADX-related facts independently, you can confirm three things:
- The exact price data series (high/low/close) and timeframe used.