Ramsey RESET Test Calculator
Enter aligned Y and X rows plus a maximum fitted-value power to calculate the Ramsey RESET joint F and chi-square references for a one-predictor OLS model. The calculator exposes the augmented regression but does not diagnose a specific omitted variable, prescribe a transformation or certify model quality.
Enter the base regression and RESET power
The engine fits Y = β0 + β1X + e, then adds fitted-value powers Ŷ² through Ŷᴾ to the original intercept and X.
Choose a whole maximum power from 2 through 5. All fitted powers from 2 through P are added.
Enter 20 to 500 finite response values without percent signs, one aligned value per row.
Enter one finite, nonconstant predictor value for every Y row. The page adds the intercept.
Fitted-power augmented regression
Entered Regression Specification Diagnostics 1.0.0.
On smaller screens, scroll horizontally to inspect the complete audit table.
| Row | Response Y | Predictor X | Base fitted | Base residual | Fitted-value powers | Augmented fitted | Augmented residual |
|---|
How this Ramsey RESET test is calculated
The base stage fits ordinary least squares with an intercept and the one entered predictor. Its fitted value for each row is retained. The augmented design then adds powers of that fitted value beginning at two and ending at the selected maximum power.
Selecting power two adds only the fitted square and creates one joint restriction. Selecting power three adds the fitted square and cube and creates two restrictions. Version 1.0.0 accepts maximum powers from two through five and uses the selected value exactly.
The primary F statistic compares the base residual sum of squares with the lower augmented residual sum of squares, scaled by the number of added powers and the augmented residual degrees of freedom. The displayed F reference uses those exact degrees of freedom.
For reconciliation with the governed nonrobust statsmodels convention, the page also shows the corresponding Wald chi-square statistic q multiplied by F and its chi-square reference. Robust, HC, HAC and clustered covariance variants are not silently substituted.
Worked example from the audited fixture
The governed example contains 48 aligned rows and a declared maximum fitted-value power of three, independently reproduced with statsmodels 0.14.6.
- The common base OLS fit has R-squared 0.78135869 and residual sum of squares 3.47334755. The augmented regression adds fitted squares and cubes.
- The augmented R-squared is 0.84317241 and residual sum of squares is 2.49137158, producing F 8.67131645 with reference 0.00066863 and chi-square 17.34263291 with reference 0.00017143. No causal diagnosis or adequacy verdict is attached.
Reproduce it: select “Load audited example” above. The governed engine retains full precision and rounds only the visible interface.
How to interpret the result
- Treat the output as a joint fitted-power reference for this specific base equation. It can flag that added nonlinear combinations of the fitted values explain residual variation, but it does not name the missing term.
- A small reference probability may be consistent with functional-form tension, omitted nonlinear structure or another specification problem. It is not proof that a polynomial in X is the correct repair.
- A larger reference probability does not establish correct specification. Power depends on sample size, the selected maximum power, the base regressors and the actual alternative.
- RESET is broad by design. Use residual plots, domain knowledge, pre-specified alternative models and out-of-sample evaluation to investigate any tension rather than mechanically adding fitted powers to a trading rule.
- The F and chi-square references answer the same joint fitted-power question under different finite/asymptotic reference systems. They are not two independent confirmations.
- Repeatedly trying powers two through five and reporting only the smallest probability creates selection bias. Declare the power first or disclose the full search and adjust the inference separately.
Assumptions and limits
- Enter 20 to 500 aligned finite rows and a whole maximum power from two through five. X must vary, the base residual sum must be positive, and the augmented design must remain full rank.
- Only one predictor plus an intercept is fitted. The page cannot reconstruct a multi-factor model, lagged dependent terms, weights, fixed effects, categorical encodings or researcher-selected transformations.
- The implementation uses fitted-value powers and nonrobust covariance. Exogenous-variable powers, principal components, HC covariance and other RESET variants can produce different results.
- Chronology, equal spacing, data provenance, stationarity, exogeneity, independence, normality, structural stability, influential observations and multiple-model selection are not verified.
- The test cannot distinguish omitted variables from incorrect transformations, regime changes, data errors or other misspecification sources.
- No functional-form diagnosis, model-adequacy decision, parameter recommendation, forecast, predictive-edge claim, backtest, strategy validation, grade, signal or trading recommendation is generated.
Which regression diagnostic answers which question?
These diagnostics are complementary rather than interchangeable. They use different auxiliary responses, added terms and maintained nulls. A reference probability only has meaning alongside its source model, entered rows, chosen lag or power, auxiliary design and data-selection process.
| Diagnostic | Primary question | Output reference | Declared specification | Requires original Y and X |
|---|---|---|---|---|
| White | Residual variance versus X and X² | LM χ²(2) plus auxiliary F | One predictor; generated square | No |
| Breusch–Pagan / Koenker | Residual variance versus declared predictor | LM χ²(1) plus auxiliary F | Residuals and variance predictor | No |
| Breusch–Godfrey | Residual levels versus own lags | LM χ²(q) plus joint F | Base Y, X and fixed lag q | Yes |
| Ramsey RESET | Added fitted-value powers | Joint F plus Wald χ² | Base Y, X and maximum power | Yes |
Frequently asked questions
- It fits a one-predictor OLS model, adds powers of its fitted values from two through the declared maximum, and tests those added coefficients jointly.
- Power three adds fitted-value squares and cubes, for example; Version 1.0.0 accepts one declared maximum power from two through five.
- The page uses fitted-value augmentation and nonrobust covariance, matching the disclosed default style rather than silently substituting exogenous powers or robust covariance.
- It compares the restricted base residual sum of squares with the augmented residual sum, scaled by the added-power count and augmented residual degrees of freedom.
- For the governed nonrobust convention, the Wald chi-square statistic equals the number of restrictions multiplied by F and provides an asymptotic companion reference.
- No. A fitted-power reference can indicate broad specification tension but cannot name the missing variable, transformation, regime or correct repair.
- No. It uses the entered maximum power exactly and does not search powers or correct for a hidden multiple-model selection process.
- No. It generates no functional-form verdict, model-adequacy decision, predictive-edge claim, forecast, backtest, signal or recommendation.
Sources and methodology
- statsmodels — Ramsey RESET — Official documentation for the fitted-power default, power range, covariance option and Wald reference.
- statsmodels — diagnostic source — Official implementation source used to reconcile augmentation and F-versus-chi-square output.
Version 1.0.0 was locked after the base regression, auxiliary statistics, degrees of freedom, reference probabilities, coefficients and worked examples were independently recomputed with statsmodels 0.14.6 and SciPy 1.13.1. The browser calculator performs local arithmetic and does not upload entered observations. Outputs remain conditional on the disclosed model and assumptions.
Continue the model-specification audit
Verify the source data before testing the regression
Reconcile the exact symbol, observation timestamps, timezone, sampling frequency, missing rows, spread, commission, financing, currency conversion, rollover adjustments and preprocessing before entering Y and X. Correct diagnostic arithmetic cannot repair a selected, misaligned, cost-inconsistent or otherwise misspecified source model.
Risk and affiliate disclosure: Leveraged forex and CFD trading can result in substantial losses. These are affiliate links, so ForexMT4Indicators.com may receive compensation if you register or trade through them, at no additional cost to you. Availability and terms vary by jurisdiction and broker entity.