What data is needed to assess Rmi?

Direct answer

To assess Rmi, you need more than a value. You need (1) the exact definition and calculation inputs, (2) the provenance of each input (data source and how it was produced), (3) timeliness and consistency (timeframe, sampling, and whether inputs match the period you are judging), and (4) quality checks (data gaps, outliers, and scaling/normalization choices). Because you may not have real-time data, you should focus on whether the underlying mechanics are well-specified and verifiable, rather than expecting predictive certainty.

Mechanism or definition

First define the concept you are assessing. “Rmi” should be treated as a named indicator/metric whose meaning depends on its formula. The minimum “inputs” for assessing Rmi are therefore:

1. The formula and naming scope: the specific expression used to compute Rmi, including each component term.
2. The raw data feeding those components: commonly this means price series or derived measures (for example, values and differences computed from a price series). Identify what price type is used (such as close, open, or another field) and whether the components are directly observed or calculated.
3. Window and parameter settings: lookback length, smoothing method, and any scaling rules.
4. Time alignment rules: whether the calculation uses the current bar close, prior bars only, or a shifted value.

How the data “works” is straightforward: once you know the formula, you can reproduce Rmi from the stated inputs. The resulting sensitivity is then driven by how the inputs are built—especially the sampling timeframe and any transformation steps.

A practical way to structure your assessment is to separate stable mechanics from variable conditions:

• Stable mechanics: the stated formula, parameter definitions, and alignment rules.
• Variable conditions: the market period, volatility regime, trading costs/quality of fills (if you later study outcomes), and the data provider’s implementation choices.

Evidence or example

Use a reproducibility check as “evidence” of correct assessment.

• Reproduce from a known dataset: take a historical price series for the exact timeframe, apply the stated parameters, and confirm that the intermediate component values match the method’s expectations (as far as published documentation allows).
• Document your assumptions: if you use an example, write down the assumed timeframe (e.g., the bars represent 1-hour candles), the parameter values, and the alignment rule.
• Compare across implementations carefully: if two sources claim to compute “Rmi,” compare whether they use the same input fields, lookback windows, and normalization/smoothing. Differences in any of these break comparability.

If you cannot reproduce it, treat that as a failure mode: the assessment is incomplete because the definition and inputs are not fully specified.

Limitations and risks

At least one material limitation should be part of your assessment:

• Implementation ambiguity: the biggest risk is assuming that “Rmi” has one universal definition. If the formula, parameterization, or time alignment differs, the same label can produce different numbers.
• Quality problems in inputs: missing bars, outliers, corporate action adjustments, or inconsistent timezone handling can change derived values and therefore Rmi.
• Historical-to-future uncertainty: even if Rmi correlates with outcomes in one period, historical relationships do not establish future results. Changes in volatility, liquidity, execution quality, or costs can alter any relationship.
• No real-time assumption: if you only use historical data, you should not infer real-time performance. Timeliness and execution details cannot be validated from past snapshots alone.

Verification or next question

To verify that your assessment is complete, check these “ready-to-audit” criteria:

• Do you have the exact Rmi formula and the meaning of every parameter?
• Can you identify the input data type(s) and the method that generated derived components?
• Is the timeframe and bar alignment explicitly stated?
• Are data-quality checks defined (how missing values are handled, how outliers are treated, and how normalization is done)?
• If you compare providers, have you confirmed that their implementation details match?

If you want to go one step further, the next question to answer is: Which precise definition of Rmi are you using (formula, parameters, and alignment), and what raw data fields feed it?