Define what to assess: inputs vs outcomes
To assess “ADX and Moving Average,” you need two things: (1) the data used to compute each indicator, and (2) the way you interpret results against outcomes. This article focuses on the first part: what inputs are required, how they should be chosen, and what checks make the computed values trustworthy. Outcomes (returns, continuation, reversals) depend on market conditions, trading costs, and execution, so indicator values alone do not establish future performance.
Mechanism and definition: what data drives each indicator
ADX (Average Directional Index)
ADX is commonly computed from directional movement and then averaged over a chosen period. The required data inputs are:
- Price series used by the calculation (typically high, low, and close; some tools offer variants, so record exactly what your platform uses).
- Sampling frequency/timeframe (for example, 1-minute candles vs daily candles).
- ADX lookback/period length used by the indicator settings.
- How “smoothing” is applied inside the formula (many implementations use a form of smoothing; you must capture the implementation style from your platform’s indicator settings).
Moving Average (MA)
A moving average is computed from a selected price series using a specified window length and averaging method. The required data inputs are:
- Price series used (close, typical price, etc.).
- Moving average type and method (for example, simple vs exponential); record the exact option name.
- Window length (period) and timeframe.
Combined assessment
When you assess them together, you still need the same underlying inputs, plus alignment information:
- The timeframe used for both indicators must match (or you must explicitly document any intentional mismatch).
- The bar/candle boundaries must match the platform’s definition of “current” and “historical” values.
Evidence and example: a checklist for reproducible calculation
Use a control-style checklist to make the assessment independent of guesswork.
- Indicator configuration (the “settings” data)
- ADX period length.
- ADX calculation variant or smoothing method (as exposed in your platform).
- MA type (simple/exponential/etc.).
- MA window length.
- Exact price fields selected for each indicator.
- Market data provenance (where the numbers came from)
- Data vendor/provider used by your platform.
- Whether the dataset is spot, aggregated candles, or another form.
- Whether adjustments are applied (for example, how missing or abnormal candles are handled).
- Timeliness and alignment (when the data is “as-of”)
- Timezone and session boundaries used for candles.
- The cutoff time for “current bar” calculations (many platforms compute indicator values using the same bar definition; document that rule by checking platform behavior).
- Any resampling steps (for example, if you built 1-hour bars from 5-minute data).
- Quality checks (how to detect calculation issues)
- Missing bars or gaps: confirm the indicator output does not silently interpolate in a way you did not expect.
- Consistency: confirm that the same settings produce the same values when rerun on the same dataset.
- Sanity: check that indicator magnitudes fall into expected ranges for your implementation (this is a basic consistency check, not a prediction).
Limitations and risks: common failure modes
Mismatched inputs
If ADX uses one price set (e.g., high/low/close) and your MA uses only close, that is normal—but problems arise when the timeframe or candle construction differs. A mismatch can make the combined view misleading even if each indicator was computed correctly.
Hidden implementation differences
Different platforms may implement ADX smoothing or directional movement details differently, and may choose different default price fields. Even stable concepts can differ in exact output because the computational steps can vary.
Historical relationships are not guarantees
Even if ADX levels or MA behavior were linked to past outcomes, that relationship can change when market conditions shift. Historical association does not establish future predictability.
Data quality and availability
Missing candles, late updates, corporate action handling (where relevant), or changes in the provider’s adjustments can alter indicator values. These issues can create apparent “signals” that are actually artifacts.
Verification and next questions
To verify your assessment independently, recompute using the same inputs and confirm your platform matches. If you cannot recompute easily, at least document:
- The exact indicator settings names and parameters.
- The price fields used for ADX and MA.
- The timeframe and candle definition.
- The data source/provider and any resampling steps.