31. Forecasting in Excel

31.4 FORECAST.ETS.CONFINT, FORECAST.ETS.SEASONALITY and FORECAST.ETS.STAT

By Ravindra Bagale · Last updated: 7 October 2026

Example first. Salman's operations head asks two follow-up questions: "How sure are you about the 548 orders on Saturday?" and "Which pattern did Excel use?" FORECAST.ETS.CONFINT answers the first with a "±" range around the forecast. FORECAST.ETS.SEASONALITY answers the second with the pattern length. FORECAST.ETS.STAT reports the smoothing parameters and error measures behind the forecast. Together they turn a single number into a forecast you can defend.

First look at the data (before → after):

Expected result — With a 95% range (approximate)

D58: =FORECAST.ETS.CONFINT(A58,$B$2:$B$57,$A$2:$A$57,0.95,7)
E58: =C58-D58   F58: =C58+D58
H2: =FORECAST.ETS.SEASONALITY(B2:B57,A2:A57)   H3: =FORECAST.ETS.STAT(B2:B57,A2:A57,8)

FORECAST.ETS.CONFINT

FORECAST.ETS.CONFINT(target_date, values, timeline, [confidence_level], [seasonality], [data_completion], [aggregation])

• confidence_level is a number between 0 and 1; the default is 95%. A 95% interval means 95% of future points are expected to fall within this distance of the forecast (Microsoft's wording). A 90% interval is narrower, 99% wider.
• The result is the half-width (the "±" part), not the bound itself. Lower = forecast − CONFINT, upper = forecast + CONFINT.
• Use the same seasonality, data_completion and aggregation as in your FORECAST.ETS formula, or the interval belongs to a different model.

On this data the 95% half-width is about ±7 to 9 orders per day (LibreOffice gives values in this range; it uses a simulation, so its numbers move slightly between recalculations). Excel's value will be similar but not identical. The interval is narrow because the weekly pattern is very regular in this practice data; real data usually gives wider ranges.

FORECAST.ETS.SEASONALITY

FORECAST.ETS.SEASONALITY(values, timeline, [data_completion], [aggregation]) returns the pattern length that FORECAST.ETS uses when seasonality is automatic. Microsoft states the detected value is identical to the one FORECAST.ETS uses with the same inputs.

What to expect here: the true pattern is weekly, so the value you want is 7. Automatic detection does not always return the shortest cycle. LibreOffice's detector, for example, returned 21 (three weeks) on these 56 days and 14 on the first 49 days, both multiples of 7. Check what your Excel returns. If it is not 7, type 7 as the seasonality argument in every ETS formula. If it returns 0, Excel found no pattern and the forecast is a trend line.

FORECAST.ETS.STAT

FORECAST.ETS.STAT(values, timeline, statistic_type, [seasonality], [data_completion], [aggregation])

Use type 8 as a quick check that Excel read your timeline as intended (1 = one day; for monthly data on the 1st of each month Excel reports the step in its own units). Types 4 to 7 measure the error on the history. For a fair accuracy test, hold back the last week and measure it yourself (lesson 31.7). The exact alpha, beta and gamma depend on Excel's fitting method, so they are not printed here; they are always between 0 and 1.

Approximate: Excel's FORECAST.ETS fits its smoothing parameters with its own method, which Microsoft does not publish. These ETS numbers were computed with two other implementations of the same AAA exponential smoothing (LibreOffice Calc 25.2 and Python statsmodels, seasonality 7). They agree with each other within about 2 orders; expect Excel's results to be close to them, not identical.

Microsoft sources (checked 7 October 2026): FORECAST.ETS.CONFINT function; FORECAST.ETS.SEASONALITY function; FORECAST.ETS.STAT function.

Practice task (try it before you open the answer):

1. Write the formula for a 90% half-width for Saturday 3 October (A63).
2. If C63 is 548 and D63 is 8, what range do you plan for?
3. FORECAST.ETS.SEASONALITY returns 14 on your data. What do you do?