Feature Importance
Feature importance measures how much each input variable contributes to a model's predictions, helping explain which signals actually drive a forecast. It is closely related to Feature Engineering, Overfitting, Data Leakage and Ensemble Disagreement, each defined in the hedgewing.ai glossary.
A machine-learning model that uses dozens of inputs is only interpretable if you can rank which inputs matter. Feature importance does that. For tree-based models it can be read from how often and how usefully a feature is used in splits; for neural networks it is usually estimated by permutation (shuffling one feature and measuring how much accuracy drops), by gradient-based attribution, or by SHAP values that assign each feature a share of each individual prediction.
Feature importance is as much a debugging tool as an explanation tool. If a feature that should carry no information ranks near the top, it often means the pipeline has leaked future data into it. If importance is spread almost uniformly across many weak features, the model may be fitting noise. Stable importance rankings across walk-forward windows are a reassuring sign; rankings that reshuffle every retrain suggest overfitting.
hedgewing.ai's models are trained on engineered features spanning price momentum, volatility, volume, and technical indicators. Importance analysis is one way to sanity-check any architecture of this kind, and the per-model breakdown shown on a forecast is a complementary form of transparency: it tells you which models agree, while feature importance tells you which inputs they are responding to.
FAQ
What is Feature Importance?
Feature importance measures how much each input variable contributes to a model's predictions, helping explain which signals actually drive a forecast.
Which terms are related to Feature Importance?
Feature Importance is closely related to Feature Engineering, Overfitting, Data Leakage and Ensemble Disagreement. Each has its own entry in the hedgewing.ai glossary.
Is this definition of Feature Importance investment advice?
No. This glossary entry explains what Feature Importance means. It is educational reference material, not personalised investment advice and not a recommendation to buy or sell any security.
Create a free Hedgewing account to try hedgewing.ai yourself.