Supervised Learning

Supervised learning is a machine-learning approach where a model learns from labeled examples, mapping known inputs to known outputs so it can predict outcomes for new data. It is closely related to Feature Engineering, Deep Learning, Neural Network and Overfitting, each defined in the hedgewing.ai glossary.

Supervised learning is the branch of machine learning in which a model is trained on examples that include both the inputs (features) and the correct answer (the label or target). The algorithm adjusts its internal parameters to minimize the gap between its predictions and the true labels, gradually learning the relationship between inputs and outputs. Once trained, it can produce predictions for new, previously unseen inputs. The two main flavors are regression, where the target is a continuous number (such as a future return), and classification, where the target is a category (such as up, flat, or down).

In a financial context, the inputs are engineered features describing the current state of a market, and the label is some future outcome the model is trying to anticipate, for example the direction or magnitude of the next period's return. The model learns by being shown many historical windows where both the features and the realized outcome are known. Because the future label is the thing being predicted, training must be structured so the model never sees information from after the prediction point; otherwise it learns to cheat and fails in live use.

Supervised learning matters to investors because it is the workhorse behind most predictive trading systems. Its strength is that it directly optimizes for a measurable objective, but that is also its risk: a model can fit historical noise rather than durable signal, so honest out-of-sample validation, regularization, and cross-validation are essential to know whether the learned relationship actually generalizes.

hedgewing.ai uses supervised learning across its four-model ensemble of LSTM, GRU, TCN, and Transformer networks, each trained to map roughly 45 engineered features to forward-looking targets. The models are validated with nightly walk-forward backtesting and produce calibrated confidence estimates, so a prediction is paired with an honest sense of how reliable it has historically been.

FAQ

What is Supervised Learning?

Supervised learning is a machine-learning approach where a model learns from labeled examples, mapping known inputs to known outputs so it can predict outcomes for new data.

Which terms are related to Supervised Learning?

Supervised Learning is closely related to Feature Engineering, Deep Learning, Neural Network and Overfitting. Each has its own entry in the hedgewing.ai glossary.

Is this definition of Supervised Learning investment advice?

No. This glossary entry explains what Supervised Learning 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.