The Undistorted Baseline Rule
Replace anomalous year-over-year comparisons with a stable historical baseline
- Difficulty
- Easy
- Time to result
- ~days to results
- Steps
- 5
- Confidence
- 93%
The rule begins by testing whether the conventional year-over-year base period represents normal operating conditions. If the prior year contained a historic boom, shutdown, stimulus surge, or other extreme event, its comparison can produce a technically accurate but misleading percentage. The analyst then selects a more stable pre-shock period, calculates the change against that baseline, and presents it alongside the conventional figure. This separates the retreat from an extraordinary peak from the gap relative to normal conditions. The method does not erase recent history; it makes the distortion explicit and gives decision-makers a more useful estimate of current strength, weakness, and available supply.
Origin
Extracted from Masters in Business, where Jonathan Miller used pre-pandemic housing conditions as an alternative baseline after extraordinary pandemic-era activity distorted annual comparisons.
Core principles
- 01A comparison is only useful when its base period is representative
- 02Extraordinary booms can make ordinary slowdowns look catastrophic
- 03Historical normality matters more than calendar convenience
- 04Use more than one baseline when conditions remain unstable
How to run it
- 1
Audit the default comparison
Identify the event and conditions represented by the prior-year period. Determine whether activity, prices, or supply were unusually high or low.
Pro tip Graph several years rather than inspecting two isolated values.
Watch out Do not reject a baseline merely because it produces an inconvenient result.
- 2
Choose a representative anchor
Select a period before the shock that reflects ordinary market behavior. Match season, geography, and metric definition to the current period.
Pro tip A corresponding quarter often controls for seasonal effects better than an annual average.
Watch out An older period may require adjustment for population growth or genuine structural change.
- 3
Run parallel comparisons
Calculate the standard year-over-year change and the change from the representative anchor. Preserve both figures so readers can see the distortion directly.
Pro tip Label the comparisons as recent momentum and normalized position.
Watch out Never silently substitute the preferred baseline.
- 4
Explain the gap
Attribute differences between the two comparisons to identifiable conditions such as a stimulus boom, shutdown, or supply freeze. State what each comparison can and cannot establish.
Pro tip Use absolute levels alongside percentages to keep scale visible.
Watch out A normalized baseline does not automatically predict future reversion.
- 5
Update the anchor
Review whether the market has returned to stable conditions. Resume ordinary year-over-year analysis when recent periods become representative again.
Pro tip Document the rule for retiring the alternative baseline in advance.
Watch out Keeping a favorable historical anchor indefinitely creates a new bias.
In the wild
A housing report shows sales down 40% from a year earlier, but that earlier quarter was an exceptional pandemic-era peak. Comparing the current quarter with the corresponding 2019 quarter shows sales down 25% from normal instead.
→ Readers see both the sharp loss of momentum and the less dramatic gap from ordinary activity.
A software company compares renewals with a prior-year launch promotion that temporarily doubled sign-ups. It adds a comparison with the last stable pre-promotion cohort and clearly labels both measures.
→ Management avoids treating the end of a temporary surge as equivalent to structural collapse.
Common mistakes
Using the nearest period by default
Calendar proximity does not make a period representative when it contains an exceptional shock or boom.
Hiding the standard comparison
Replacing year-over-year data without disclosure can look like cherry-picking and deprives readers of recent momentum.
Ignoring structural change
A pre-shock period may also mislead if technology, population, regulation, or behavior permanently changed afterward.
Is it for you?
Best for
It is best for analysts interpreting markets, businesses, or metrics disrupted by a major temporary shock.
Not ideal for
It is not ideal when a structural break has made every earlier period fundamentally incomparable with current conditions.
From the transcript
“year year comparisons make it look like, you know, you're down forty percent, but you were up fifty eighty percent a year ago over the…”
“pre pandemic, which became my alternative metric to year over year because the distortion that has occurred in twenty twenty one two and early well…”
“compared to say, second quarter nineteen compared to second quarter of this year, we're down about, you know, in the twenty to thirty percent range…”
From the episode
Jonathan Miller on High Mortgage Rates