# Maximum Likelihood Estimation

*Math & Statistics — Finicade finance glossary*

Maximum likelihood picks the parameter values that make the observed data most probable under the assumed model. It's the general-purpose estimator behind GARCH, logistic regression and most of modern statistics, and under normal errors it coincides exactly with least squares. Its guarantees are asymptotic, so in small samples MLE can be meaningfully biased — variance estimates being the classic example.

**Also known as:** MLE, maximum likelihood

**Related terms:** [Ordinary Least Squares](https://finicade.com/glossary/ordinary-least-squares), [Bayesian Inference](https://finicade.com/glossary/bayesian-inference), [GARCH](https://finicade.com/glossary/garch), [Probability Distribution](https://finicade.com/glossary/probability-distribution), [Model Calibration](https://finicade.com/glossary/model-calibration)

Source: https://finicade.com/glossary/maximum-likelihood-estimation