- A 40% jump from October to November does not mean Northstar is growing 40% every month.
- That comparison mixes business growth with the holiday season.
| Growth rule | What it means | Northstar example |
|---|---|---|
| Average all months | Stable and smooths random variation, but adapts slowly to a changing rate. | Use Northstar's past comparable-month YoY rates when they fluctuate around a reasonably stable level. |
| Recent months | Adapts to acceleration or slowdown with less averaging; can be noisier. | Northstar considers the last three available YoY rates after a new market changes its growth pace. |
| Latest month | Reacts fastest, but one unusual observation can dominate. | A Northstar sale or weak comparison base can make the latest rate unrepresentative of next month. |
A little growth goes a long way
Northstar needs a growth assumption for next November. Compare observed year-over-year rates through October, select a rule using only those months, and test that rule on earlier months.
The question it answers: When does averaging growth smooth noise, and when does it lag a changing pattern?
Backtest April–October · predict each month using only earlier rates
| Method | MAE (pp) | Bias (actual − chosen, pp) |
|---|---|---|
| Average all | 3.07 | 3.07 |
| Last 3 months | 1.69 | 1.69 |
| Latest month | 1.00 | 0.83 |
Fine print & method
Two deliberately constructed YoY growth scenarios, not observed Northstar results. November's prior-year aligned sales index is supplied as 100. The November outcome is never used to select a rate or score the earlier backtests. The most recent method can be noisy; averaging is not universally right or wrong. Align weekdays/events and investigate base effects before applying growth to actual sales.