Direct answer: what data you need to assess MACD
To assess MACD, you need (1) the indicator’s calculation settings, (2) the price data series it is computed from, and (3) checks that the data and settings match what others used. In practice, “assessing MACD” means you can explain what the lines represent, reproduce the calculation from the raw inputs, and judge whether your data is consistent and timely enough for the question you are asking.
Mechanism or definition: which inputs MACD depends on
MACD is built from moving averages of a price series. The core data inputs are:
- Price series
- The indicator typically uses a specific price field (commonly the closing price per bar).
- You must know which field is used (e.g., close vs. another price) because changing the input alters the MACD values.
- Timeframe / bar interval
- MACD values depend on how bars are formed (for example, 1-minute vs. 1-hour bars).
- Assessment requires the exact timeframe so the computation corresponds to the chart you are looking at.
- Moving-average parameters
- MACD is defined using at least two moving-average periods (commonly a “fast” and a “slow” period) and often an additional smoothing period for the histogram or signal line.
- You must capture the numeric periods used by the MACD variant you are assessing.
- Data completeness and consistency
- Your price series needs continuous bars for the lookback window relevant to the moving averages.
- If bars are missing or timestamps are misaligned, the computed averages and resulting MACD lines can be distorted.
- Computation method details
- If the indicator implementation allows choosing the moving-average type (for example, different definitions of moving averages), those choices are part of the required data.
Evidence or example: how to verify you assessed the same MACD
A practical way to assess MACD independently is to reproduce it:
-
Record the settings you are using (timeframe and moving-average periods, plus any signal/histogram smoothing setting).
-
Get the raw price series from your data source for the same timeframe and date range.
-
Recompute MACD from the raw series using the recorded parameters.
-
Compare results
- If your recomputed MACD differs from a platform’s published MACD, the mismatch usually comes from one of the inputs: a different price field, different moving-average type, different parameters, or different bar construction.
Even without any real-time assumptions, you can still do this verification using historical bars, as long as you clearly define the inputs and calculation settings you used.
Limitations and risks: where assessment can fail
Several limitations affect the reliability of MACD assessment:
-
Different implementations exist: MACD can be configured with different parameter values and potentially different moving-average definitions. Without those settings, two charts labeled “MACD” may not be comparable.
-
Timeframe sensitivity: MACD reacts to the bar interval. A conclusion drawn on one timeframe may not translate to another.
-
Data quality failure modes: missing bars, inconsistent candle building rules, or timezone/timestamp differences can cause calculation discrepancies.
-
No guarantee of future performance: historical relationships between MACD behavior and subsequent price changes do not establish future outcomes. Costs, execution details, and changing market conditions can also alter results.
-
Derived values can hide input issues: because MACD uses smoothing, small input inconsistencies can propagate through averages and produce larger visible differences over time.
Verification or next question: a checklist for dependable MACD data
Use this checklist to ensure your assessment is based on verifiable inputs:
- Which price field did you use for each bar (commonly close)?
- Which timeframe did you compute on (exact bar interval and bar construction rules)?
- What parameter set did you use (fast/slow periods and any smoothing for signal/histogram)?
- Which moving-average definition was applied (if selectable)?
- Is the data complete and aligned (no missing bars, consistent timestamps)?
- Can you reproduce the output by recomputing MACD from the recorded raw inputs?
If you can answer these, you can accurately explain MACD and independently verify the relevant facts behind any MACD plot you are analyzing.