Glossary term
Ordinary Least Squares (OLS)
Ordinary least squares is a regression method that estimates the line or model coefficients that minimize the sum of squared residuals.
Updated
Read time
What Is Ordinary Least Squares (OLS)?
Ordinary least squares, or OLS, is a regression method that estimates the line or model coefficients that minimize the sum of squared residuals. A residual is the difference between the observed value and the value predicted by the model.
OLS is one of the most common tools in statistics, economics, finance, and business analytics. It is used to estimate relationships, test factor exposures, forecast values, and explain variation in data.
Key Takeaways
- OLS estimates regression coefficients by minimizing squared residuals.
- It is widely used in finance, economics, accounting, and business analytics.
- OLS can estimate simple or multiple regression models.
- Its usefulness depends on assumptions about the data and model structure.
- A good OLS fit does not prove causation or guarantee predictive power.
The Objective
OLS chooses coefficients to minimize the sum of squared model errors:
In this expression, yi is the observed value, ŷi is the predicted value, and the model chooses coefficients that make the squared residuals as small as possible across the sample.
For example, if a regression line repeatedly misses a fund's returns by large amounts, OLS will choose coefficients that reduce those squared misses across the sample. Large misses receive extra weight because the errors are squared.
Where OLS Shows Up
Use case | What OLS estimates | Financial meaning |
|---|---|---|
Factor analysis | Exposure to market or style factors. | Explains return drivers. |
Valuation | Relationship between fundamentals and price. | Supports comparative analysis. |
Forecasting | Relationship between predictors and outcomes. | Builds estimates with uncertainty. |
Risk modeling | Sensitivity to economic variables. | Tests how outcomes change with drivers. |
How to Interpret OLS Results
OLS produces coefficients, residuals, fitted values, and fit statistics such as R-squared. A coefficient estimates how the dependent variable changes when an explanatory variable changes, holding other variables constant in a multiple regression.
That interpretation depends on the model being appropriate. Omitted variables, outliers, nonlinearity, multicollinearity, changing variance, or reverse causality can weaken the conclusion. OLS can estimate relationships in the data, but the analyst must decide whether the relationship is economically meaningful.
OLS is often a starting point rather than the end of the analysis. It gives a transparent estimate that can be tested, challenged, and compared with other model forms.
What to Watch
Good OLS analysis looks beyond the headline coefficient. It reviews residuals, sample size, statistical significance, model stability, economic logic, out-of-sample performance, and whether the assumptions are reasonable for the problem.
In finance, OLS can be useful and dangerous for the same reason: it makes relationships look precise. Market relationships can shift, especially during stress.
The practical question is whether the fitted relationship remains useful when the environment changes. A regression that explains yesterday's market can still fail when liquidity, policy, or investor behavior shifts.
The Bottom Line
Ordinary least squares estimates a regression by minimizing squared prediction errors. It is a foundational tool for financial and economic analysis, but its results need diagnostics and judgment before they are treated as useful evidence.