Multi-Horizon Trend Analysis
Expand the time window to separate trend from noisy annual changes
- Difficulty
- Easy
- Time to result
- ~days to results
- Steps
- 5
- Confidence
- 96%
Multi-Horizon Trend Analysis tests an economic claim across progressively wider windows instead of relying on the customary year-over-year figure. Begin with the annual change, then calculate two-year and three-year changes before comparing the current level with a seven-to-eight-year trend. The mechanism distinguishes direction from accumulated magnitude: a measure may be falling annually while remaining far above its historical trajectory. That residual gap can explain why liquidity, demand, or inflation persists despite apparently restrictive recent data. Analysts then reconcile the conflicting signals rather than selecting the window that supports their prior view. The output is a contextualized assessment showing both what is changing now and how much of an earlier shock remains embedded in the system.
Origin
Extracted from Masters in Business during Jeffrey Sherman's discussion of pandemic-era M2 growth and misleading year-over-year recession signals.
Core principles
- 01A single year-over-year comparison can conceal the larger trend
- 02Pandemic-era distortions require longer historical context
- 03Evaluate both the current direction and the cumulative deviation
- 04Use several time horizons before accepting a market signal
How to run it
- 1
Identify the headline signal
Record the commonly cited year-over-year change and the conclusion the market is drawing from it.
Pro tip Write down the prevailing interpretation before expanding the window.
Watch out Do not treat the headline comparison as the final signal.
- 2
Expand the measurement windows
Calculate the same change over two-year and three-year periods to reduce base effects and reveal cumulative movement.
Pro tip Use consistent definitions and data vintages across every window.
Watch out Revisions or inconsistent series can create artificial differences.
- 3
Build the long trend
Plot the current level against a seven-to-eight-year trend or another representative pre-shock baseline.
Pro tip Test more than one sensible trend period to check robustness.
Watch out A trend chosen only because it confirms the thesis is not useful.
- 4
Measure the residual gap
Determine how far the current level remains above or below its historical trajectory, even after recent declines.
Pro tip Express the gap in both absolute and percentage terms.
Watch out A slowing growth rate does not mean an accumulated excess has disappeared.
- 5
Reconcile the signals
Explain how the short-term direction and long-term level jointly affect the economy or market under study.
Pro tip State what evidence would cause the interpretation to change.
Watch out Do not discard a conflicting horizon without explaining why.
In the wild
Year-over-year M2 contraction appeared recessionary after the pandemic. Extending the window showed that the monetary base remained substantially above its prior trend because of the earlier surge. The longer view supported a thesis of lingering liquidity rather than immediate monetary scarcity.
→ The analyst avoids mistaking partial normalization for a complete removal of liquidity.
A retailer reports a 10% annual sales decline after a one-time boom. Comparing sales with the pre-boom trend shows that revenue is still 18% above its expected level. Management treats the decline as normalization rather than evidence that the core business is shrinking.
→ Planning reflects both the recent slowdown and the still-elevated revenue base.
Common mistakes
Stopping at year over year
Annual changes can be dominated by an unusually high or low comparison base and may invert the apparent signal.
Confusing growth with level
A variable can contract while remaining far above its long-term trajectory, so the rate of change and accumulated level must be evaluated separately.
Choosing a convenient baseline
Selecting the trend period after seeing the result turns the method into confirmation bias rather than analysis.
Is it for you?
Best for
Investors and analysts evaluating data affected by shocks, stimulus, base effects, or regime changes.
Not ideal for
It is not sufficient by itself for short-term trading decisions driven by immediate catalysts.
From the transcript
“what you actually need to do is look at the two year number change, or look at the three year number change.”
“What you need to do is look at the trend line over the last seven or eight years, not just ye over year.”
“when you start to look at it and you take a different perspective versus year over year, and you go back a couple of years,…”
From the episode
Jeffrey Sherman on a Mathematician's Journey Into Finance
Jeffrey Sherman