Power Trendline in Excel: How to Add and Interpret

Charts & pivot tables · Updated November 21, 2026

To add a power trendline in Excel for Microsoft 365 on Windows, select an XY (Scatter) chart, then choose Chart Design > Add Chart Element > Trendline > More Trendline Options and select Power. A power trendline fits a curve of the form y = a*x^b; both the x-values and y-values must be positive for Excel to create it.

What a power trendline shows

A power trendline describes a relationship in which one variable changes in proportion to a power of another. In y = a*x^b, a is a scale factor and b is the exponent that determines the curve’s shape. The chart draws the fitted values as a smooth curve through or near the observed data points.

The exponent gives a useful first clue about the relationship. If b is near 1, the model is close to a straight-line proportional relationship. If it is greater than 1, the fitted curve rises more sharply as x increases; if it is between 0 and 1, it rises more slowly. A negative exponent describes a decreasing relationship: for example, with b = -1, doubling x halves the fitted y-value, assuming the same coefficient and a positive x.

This model is not a generic curved line. It is appropriate when the relationship is plausibly multiplicative or scale-based, such as a quantity that changes by a consistent factor as another quantity changes. A curved appearance alone is not enough reason to choose it; a polynomial or another model may better represent different data.

How to add one to an Excel chart

Use an XY scatter chart when the horizontal values are numeric measurements. A standard Line chart treats horizontal-axis entries as categories spaced evenly apart, which can make the curve misleading when x-values have unequal intervals. To create and format the trendline in Excel for Microsoft 365 on Windows:

  1. Arrange the independent-variable values in one column and the corresponding dependent-variable values in another. Keep each x-y pair on the same row.
  2. Select both columns, including their headings if present. Choose Insert > Insert Scatter (X, Y) or Bubble Chart, then select a scatter chart style.
  3. Click a data point in the chart to select the data series. If the chart contains multiple series, confirm that you selected the series you intend to model.
  4. Choose Chart Design > Add Chart Element > Trendline > More Trendline Options. In the Format Trendline pane, choose Power.
  5. To show the fitted equation and its coefficient of determination, select Display Equation on chart and Display R-squared value on chart in the pane.

You can also right-click the data series and choose Add Trendline, then select Power in the Format Trendline pane. On Mac, the chart controls may be arranged differently, but select the series, use the chart’s Add Chart Element or trendline controls, and choose Power in the formatting pane. Excel for the web may not expose every chart-formatting option available in the desktop app; if the trendline type or equation-display controls are missing, open the workbook in desktop Excel.

How Excel fits the power equation

Excel can estimate a power curve by taking logarithms of both sides of the model. Starting with y = a*x^b, the transformation gives ln(y) = ln(a) + b*ln(x). That is a straight-line relationship between the logged values: the slope corresponds to b, while the intercept corresponds to ln(a). Excel estimates the coefficients and draws the resulting curve on the original chart axes.

This explains why zero and negative values are a problem: their real logarithms are not defined. Excel therefore cannot fit its power trendline when the selected data contain zero or negative x- or y-values. Do not replace such observations with small positive numbers just to make the option work; that changes the data and can produce a misleading result. Investigate why the values are nonpositive and decide whether another model is appropriate.

The logarithmic transformation also matters when evaluating fit. The displayed R-squared for a power trendline is associated with the transformed regression, not a guarantee that the curve explains the same share of variation in the original y-values as an ordinary linear regression would. Treat it as one diagnostic, not proof that the model is suitable. Compare predicted values with observed values in the units that matter for the analysis.

Choosing power instead of another trendline

Excel offers several trendline types, and choosing the right one depends on the shape and meaning of the data, not just on which option gives the largest displayed R-squared. Consider the common alternatives:

  • Linear: Use when the change in y is approximately constant for each equal increase in x. Its equation is a straight line.
  • Exponential: Use for a relationship where y changes by a roughly constant percentage or factor over equal increments of x. This differs from a power curve, where the exponent describes the scaling relationship between x and y.
  • Logarithmic: Use when y changes quickly at first and then levels off as x increases. This model requires positive x-values.
  • Polynomial: Use when the data show bends or turning points that a single power curve cannot capture. Higher-order polynomials can wiggle substantially and extrapolate poorly.
  • Moving average: Use to smooth fluctuations in a sequence, rather than to describe a power-law relationship between numeric x and y values.

A useful comparison is to plot the observations, add plausible candidate models, and examine residuals—the differences between observed and fitted y-values. A trendline that looks close overall may still systematically overpredict at one end and underpredict at the other. Scientific knowledge about the process should guide the choice alongside chart diagnostics.

Worked example and using the equation

Suppose a scatter chart uses x-values 1, 4, 9, and 16, paired with y-values 5, 40, 135, and 320. These points follow y = 5*x^1.5: for x = 4, for example, 4 raised to 1.5 is 8, and 5 multiplied by 8 is 40. Adding a Power trendline should therefore produce coefficients close to 5 and 1.5, with an essentially perfect fit, subject to the precision shown on the chart label.

A chart equation is a label, not a worksheet formula that automatically fills cells. If you want to calculate fitted values, first read the displayed coefficients and enter them in cells. For example, with x in A2, coefficient a in E2, and exponent b in E3, enter =$E$2*A2^$E$3 in another cell and fill it down. With E2 equal to 5, E3 equal to 1.5, and A2 equal to 4, the result is 40. This formula calculates a model prediction; it does not recalculate Excel’s trendline coefficients when the source data change.

For real data, the displayed equation is usually rounded for readability. If you copy rounded coefficients into a worksheet, calculated predictions can differ from the curve’s underlying estimates, especially when x is large or the exponent is sensitive. Increase the equation label’s displayed decimal places through its formatting options, or use a regression workflow that retains full-precision coefficients when accuracy is important.

Interpreting the curve and its limits

Interpret the model within the range of the observed x-values first. A trendline can be extended visually beyond the data, but that does not establish that the same relationship continues. For instance, a power curve that matches measurements over a narrow interval might predict implausibly large or small values far outside it. Extrapolation needs independent justification.

Check whether the data points show a consistent pattern around the curve. If errors are much larger at high y-values, the model may be missing structure, or the measurement error may be proportional rather than constant. A log-based fit gives relatively greater influence to proportional differences than a fit that minimizes squared errors directly in the original y scale. That distinction can matter when deciding which prediction errors are most costly.

Also remember that a fitted association does not establish causation. A high R-squared does not show that changing x will cause y to change according to the curve, and it does not account for omitted variables, sampling bias, or measurement problems. Use the chart as a descriptive model unless the data and study design support stronger conclusions.

Common problems and fixes

  • Power is unavailable or the chart does not show a curve: Confirm that the selected object is a data series in an XY scatter chart and that x and y contain positive numeric values. Check for zeros, negatives, text entries, or cells that look numeric but are stored as text.
  • The line appears to connect categories strangely: Rebuild the chart as an XY scatter chart rather than a Line chart. Scatter charts use actual x-values and respect unequal gaps between them.
  • The equation looks rounded or inconsistent with calculations: Increase the displayed decimal places and avoid using a short rounded label as a high-precision coefficient source.
  • The curve fits poorly despite a high R-squared: Inspect residuals and predictions on the original y scale, check for influential points, and compare a scientifically reasonable alternative model. R-squared alone cannot diagnose bias or validate extrapolation.
  • One point changes the fitted curve a lot: Verify that the point is a genuine observation and check its units and entry. Do not remove an outlier solely because it changes the result; document any exclusion and its reason.

Frequently Asked Questions

What is the formula for a power trendline in Excel?
A power trendline has the form y = a*x^b. The coefficient a sets the scale, and b is the exponent that controls how y changes with x.

Why can’t I select a Power trendline?
Excel requires positive x- and y-values for a power trendline. Also make sure you selected a chart data series; the option is available through the trendline formatting controls for a suitable chart.

Should I use a power or exponential trendline?
Use a power model when a relationship is better described by y changing as a power of x. Use an exponential model when y changes by a roughly constant factor for equal increases in x. Plot the data and consider the underlying process before choosing.

Can I use a power trendline with a Line chart?
Excel may allow trendline options in some chart situations, but an XY scatter chart is generally the right choice when the horizontal values are numeric measurements. A Line chart spaces categories evenly, even when the numeric intervals differ.

Does the trendline equation update when my data change?
The chart’s trendline and equation label update when the chart’s source data change. A separate worksheet formula using coefficients copied from the label does not update those coefficients automatically; you must refresh them or build a separate regression calculation.

Is a high R-squared enough to trust the power curve?
No. It is one measure of fit and, for a power trendline, relates to the logarithmically transformed regression. Check the plotted data, residual pattern, units, and predictions on the original scale, and avoid assuming the curve will hold outside the measured range.

Related

Try it: the Percentage Formula Generator: percent of, percent change, increase and decrease, with the formula.

More in Charts & pivot tables

Sources: Function behaviour and menu paths: Microsoft Support documentation for Excel. Method & sources

Spotted an error in this guide? Let us know.