Forecast performance is not only about choosing the right model. It is also about setting up the forecasting process correctly.
That is what makes the settings panel in Indicio so important. These controls shape which models are allowed into the comparison set, how stable your indicator mix remains over time, how strict the statistical screening should be, whether higher-frequency data can be used, and whether the final forecast can be explained clearly to stakeholders. Indicio’s public product positioning is built around exactly these ideas: variable selection, mixed-frequency modeling, explainable forecasting, and backtested model evaluation.
A well-configured forecast is usually easier to trust, easier to communicate, and more useful in decision-making. A poorly configured one can still produce a number, but that number may be harder to defend, less stable from run to run, or slower to react to new information. Indicio’s feature pages make clear that accurate forecasting is not just about output, but about using the right drivers and being able to see why the model moved.
In this article, we walk through the main settings shown in the Indicio panel and explain what they do, when to use them, and how they can improve forecast accuracy, forecast stability, and explainability.
Why forecast settings matter
Most forecasting teams focus first on the model. That makes sense. But in practice, configuration choices often shape performance just as much as the model class itself. A strong workflow filters out weak candidates, handles messy release calendars, adapts to mixed-frequency data, and makes the result interpretable for business users. Indicio’s platform is explicitly designed around that broader workflow, with variable selection, explainable forecasting, and mixed-frequency methods as core capabilities.
That is why settings deserve their own attention. They are not cosmetic controls. They are decision levers. Used well, they help you balance forecast quality, responsiveness, consistency, and stakeholder trust.
Filter models by MAPE
The first setting in the panel is model filtering by MAPE. MAPE, or mean absolute percentage error, is one of the most widely used forecast accuracy metrics because it expresses forecast error as a percentage. That makes it easy for both analysts and business teams to interpret. In Indicio, this setting lets you exclude models that do not meet a minimum performance standard. The logic fits neatly with Indicio’s broader positioning around backtested and verified forecasting workflows rather than intuition-led model selection. (indicio.com)
This is useful because not every model in a search run deserves equal attention. Once a platform tests multiple model families, transformations, and parameter combinations, the next question is not “Which one ran?” but “Which ones are accurate enough to take seriously?” MAPE filtering helps answer that question.
If your team is evaluating many candidates, this setting can save time and improve quality control by reducing the shortlist to models that meet your error tolerance. It is especially useful when forecasting is part of a recurring business workflow and you want a repeatable standard for acceptable model performance.
Lock indicator selection
Lock indicator selection keeps the same chosen indicator set across forecast updates.
This setting is valuable when consistency matters as much as adaptability. Indicio’s variable selection capability is built to identify the most relevant variables, also described as leading indicators or features, to improve model accuracy, efficiency, and reliability. But once a useful set has been identified, some teams may prefer to hold that set fixed for a period of time to preserve comparability across runs. (indicio.com)
That can be important in production forecasting. If the indicator mix changes every update, stakeholders may struggle to understand whether the forecast moved because the economy changed or because the model structure changed. Locking the indicator set makes the forecast narrative more stable and easier to communicate.
The tradeoff is that stability can reduce responsiveness. In fast-changing markets, the best indicators today may not be the best indicators next month. So this setting is strongest when consistency, governance, or internal alignment matter more than rapid re-optimization.
Passive main variable
The passive main variable setting gives you more control over how strongly the target series itself drives the model.
This matters because some forecasts lean heavily on the internal history of the target variable, while others benefit more from explanatory indicators. In some business contexts, especially around turning points, relying too heavily on the target’s own lag structure can delay recognition of change. A more passive role for the main variable can encourage the model to place greater weight on external drivers and leading indicators.
That makes this setting especially relevant when you are building forecasts meant to surface early signals rather than simply extend past patterns. It also aligns with Indicio’s emphasis on identifying market-specific leading indicators, rather than relying solely on univariate persistence. (indicio.com)
Ragged edge support
Ragged edge support becomes important the moment your indicators update on different schedules.
In real forecasting environments, that is normal. Some series are daily, some weekly, some monthly, and some arrive with publication lags. Near the end of a current period, you may have fresh values for a few indicators and stale values for others. Without a way to handle that mismatch, your forecast process can become unnecessarily rigid or throw away useful information.
This is exactly why mixed-frequency methods matter. Indicio’s MIDAS pages explain that when forecasting lower-frequency targets such as monthly or quarterly series, higher-frequency indicators can provide more up-to-date information on what is happening in the economy. Indicio also highlights mixed-frequency models as especially useful in nowcasting. (indicio.com)
Ragged edge support helps operationalize that idea. It makes the workflow more realistic by allowing the model to use what is available now, even when the latest data is uneven across series.
Enable SHAP
This is one of the most valuable settings in the panel because it directly affects trust. Indicio’s explainable forecasting feature is built around SHapley Additive exPlanations. SHAP helps quantify each driver’s contribution to a forecast and visualizes positive drivers and negative barriers, while variable rankings show which inputs explain the most variance. (indicio.com)
That matters because a forecast that cannot be explained is much harder to use. Teams do not only want the number. They want to know what changed, which drivers pushed the forecast up, which pulled it down, and why the latest run differs from the previous one. SHAP gives that layer of explanation.
Mixed frequency support
Mixed frequency support is one of the clearest examples of Indicio’s forecasting specialization.
Indicio’s model library includes MIDAS and related mixed-frequency models, which are designed to use higher-frequency indicators to predict lower-frequency variables. These models are commonly used in nowcasting because they let analysts incorporate fresher information without simply aggregating it away. (indicio.com)
This is especially valuable when your target is monthly or quarterly but the signals moving it arrive weekly or daily. Sales teams may care about monthly demand, but the leading clues might be daily traffic, weekly claims, weekly freight data, or daily prices. Mixed frequency support helps bridge that gap.
In practical terms, this setting helps forecasts become more responsive to recent developments. It is particularly useful when conditions are shifting quickly and waiting for the next low-frequency release would make the forecast stale.
MAPE threshold
The MAPE threshold determines how strict your error filter should be.
If model filtering by MAPE is enabled, this threshold becomes the gatekeeper. A lower threshold means you are demanding stronger accuracy before a model is allowed through. A higher threshold is more permissive and may preserve a larger model pool.
This setting is powerful because it lets teams tune the balance between quality control and exploration. If the target is stable and forecast error is costly, a stricter threshold can make sense. If the target is noisy or the team is still exploring a new forecasting problem, a looser threshold may be more practical.
The value here is not only statistical. It is operational. The threshold helps define what “good enough” means for your forecasting process.
Significance level
The significance level controls how demanding the statistical screening should be.
A stricter significance level means weaker relationships are less likely to pass through into the model. A looser one allows more candidate relationships to remain under consideration. This is useful when you want to manage the tradeoff between robustness and flexibility.
For some teams, especially those presenting to highly analytical stakeholders, a stricter standard will be more defensible. For others, particularly in early exploration or noisier environments, a more flexible setting may help surface useful patterns that deserve further testing.
This setting works especially well alongside variable selection. Position variable selection as a way to identify the most relevant drivers and reduce the damage caused by noisy or redundant inputs. (indicio.com)
How these settings work together
The real strength of the settings panel is not in any one control by itself. It is in how the settings combine.
MAPE filtering and the threshold help define model quality. Lock indicator selection helps define consistency. Passive main variable changes how much the forecast leans on the target series itself. Ragged edge support and mixed frequency support help the workflow reflect real release timing. SHAP helps make the result explainable. Significance level helps keep the statistical screen disciplined.
Together, these settings map closely to Indicio’s broader product story: identify better leading indicators, use richer data structures, compare models systematically, and explain the output in a way that builds trust. (indicio.com)
How Indicio helps forecasting teams
This is where Indicio stands out from more generic analytics tools.
Indicio is built specifically for forecasting, and its public site emphasizes a workflow that combines econometric, AI, and machine-learning models with explainability, variable selection, and scenario analysis. Our tools are backtested and verified, and designed to help organizations detect trend shifts earlier while improving forecast accuracy. (indicio.com)
That matters because most forecasting teams do not just need a dashboard. They need a system that helps them answer practical questions: Which indicators matter most? Which models should we trust? Can we use higher-frequency inputs without breaking the process? Why did the forecast change? Indicio’s feature set is designed around those questions. (indicio.com)
Final takeaway
If you want better forecasting results, it is not enough to ask which model performs best. You also need to ask whether the forecasting process is configured to support quality, consistency, timeliness, and trust.
That is why the Indicio settings panel deserves more attention. Controls like MAPE filtering, indicator locking, SHAP, mixed frequency support, ragged edge support, and significance level are not minor options. They are the settings that shape whether a forecast becomes a useful business tool or just another model output.
A good forecast is not only accurate. It is stable enough to use, timely enough to matter, and explainable enough to trust. Indicio’s feature set is designed around exactly that idea.
