Baselines · Free
Drift
Random walk plus a constant average change.
What it assumes
Random walk with a constant average change.
When to use
Gentle linear trend without strong seasonality.
When to avoid
Curved trends, breaks, or seasonal cycles.
Knobs
drift = (y_n − y_1)/(n−1). ŷ_{t+h} = y_n + h·drift.
How it works
Drift is naive with a straight-line increment: the average step from the first point to the last is added once per forecast period. It is the simplest way to acknowledge a gentle trend without fitting Holt’s two smoothing weights. Curved trends, breaks, and seasonal waves will fool it.
Sample forecast question
Classroom enrollment rose from 10 to 21 over eight months. What is the drift forecast for the next three months?
Step-by-step on these numbers
| Period | Actual | Fitted |
|---|---|---|
| Jan | 10 | 10 |
| Feb | 12 | 11.57 |
| Mar | 13 | 13.57 |
| Apr | 15 | 14.57 |
| May | 16 | 16.57 |
| Jun | 18 | 17.57 |
| Jul | 19 | 19.57 |
| Aug | 21 | 20.57 |
| Period | Forecast | 95% interval |
|---|---|---|
| Sep | 22.57 | 21.60 – 23.54 |
| Oct | 24.14 | 22.77 – 25.51 |
| Nov | 25.71 | 24.03 – 27.39 |
Parameters the engine found
last= 21drift= 1.571
The question
Classroom enrollment rose from 10 to 21 over eight months. What is the drift forecast for the next three months? Sample months: Jan=10, Feb=12, Mar=13, Apr=15, May=16, Jun=18, Jul=19, Aug=21 (students).
Average step
drift = (y_n − y_1)/(n−1) = (21 − 10)/7 = 1.5714.
Forecast rule
ŷ_{n+h} = y_n + h · drift. With y_n = 21: Sep = 21 + 1×1.5714 = 22.57; Oct = 21 + 2×1.5714 = 24.14; Nov = 21 + 3×1.5714 = 25.71.
Interval
Residual σ = 0.495. For h = 1 the 95% band is [21.60, 23.54] around 22.57. The engine labels this interval native for the method.
Graph of this sample
The chart plots the canned table on this page (actuals, in-sample fitted, forecast, 95% interval). Not your Excel series. Free analysis never uploads raw data.
Educational only. Not investment, weather, or operational advice. In the add-in, rank is rolling-origin MASE — “best supported among candidates on this series.”