Direct answer
Information about a “Trend Intensity Index” can be verified by separating stable concepts (definitions and general mechanics) from variable conditions (how a specific provider calculates it, parameter choices, and how it is applied). A practical approach is to: (1) confirm what is being measured and how it is defined, (2) reproduce the computation from the described inputs and assumptions, and (3) test that the same claimed behavior appears under the stated method, not under assumed or unstated settings.
If you see claims such as “it works” or “it predicts,” treat them as non-verifiable without a clear method definition, transparent data inputs, and reproducible evaluation. Verification should focus on whether the published description is internally consistent and repeatable.
Mechanics and definition
Begin with the concept itself. “Trend intensity” generally refers to measuring how strongly prices are moving in a directional way, rather than only whether a trend exists. “Trend Intensity Index” is a named indicator, but the key verification task is to confirm the exact formula and inputs for the specific definition you are reading.
When you verify mechanics, look for these items in the documentation:
- Inputs: what price series is used (for example, close vs. typical price), the sampling period, and any smoothing or lookback window.
- Transformation steps: intermediate calculations (for example, normalization, moving averages, slope, or volatility scaling) that explain why the index rises or falls.
- Output interpretation rules: how to map values to qualitative states (for example, “higher values mean stronger intensity”)—including whether thresholds are fixed or parameter-dependent.
A common mistake is to treat any “trend intensity” indicator as interchangeable. Verification should confirm that the “Trend Intensity Index” you are examining uses the same components and scaling as the one you want to compare.
Evidence and reproducible verification steps
Because no real-time market data is assumed here, reproducibility should rely on the method description and a controlled dataset.
Step 1: Create a method checklist
Write down exactly what the source claims about the index:
- formula or pseudocode,
- required input series,
- parameter values (e.g., lookback length, smoothing length),
- any normalization or scaling rules.
Verification passes when every computation step can be traced to a stated input.
Step 2: Recompute on an example dataset
Use a fixed historical dataset you already have (for example, a CSV of OHLC bars). Then:
- set the parameters exactly as stated,
- compute the index step-by-step,
- verify that your calculated outputs match the source’s example values (if provided) within a defined rounding rule.
State assumptions explicitly: if the source does not mention rounding, define your own rounding and note that verification is sensitive to formatting.
Step 3: Check stability under method-correct variation
To test whether behavior depends on unstated settings, repeat the calculation with only one change at a time:
- adjust parameter values within the ranges the source mentions,
- change only the rounding/precision handling,
- verify that the direction of change in the index aligns with the described interpretation.
This helps distinguish a real “mechanic” from a coincidental result.
Step 4: Evaluate claims without turning it into a signal
If a provider states that the index is useful, you can still verify the claim type:
- Does the claim specify an objective rule (entry/exit, decision threshold), or only describe correlation?
- Are performance results tied to transaction costs, execution assumptions, and data selection?
If those details are missing, you cannot reproduce the conclusion.
Limitations and risks
Several material limitations frequently prevent verification:
- Different “Trend Intensity Index” definitions: the same name may refer to different formulas across sources. Without the exact equation, verification is impossible.
- Parameter dependence: thresholds and lookback windows can change the index’s scale and behavior. A result that appears with one setting may not hold with another.
- Normalization and price choice: switching between close, typical price, or log returns can materially alter outcomes even if the indicator is “the same.”
- Costs and execution assumptions: even if an indicator correlates with past moves, net performance can differ once spreads, commissions, slippage, and latency are considered.
- Failure mode—overfitting: if thresholds are chosen after seeing historical outcomes, reported backtests may reflect selection rather than a general mechanic.