Conditional Mean Reversion
Estimate a moving fair value before betting on a return to average
- Difficulty
- Advanced
- Time to result
- ~weeks to results
- Steps
- 5
- Confidence
- 96%
Conditional Mean Reversion asks not merely whether a variable will return to average, but which average is economically appropriate now. The analyst identifies the forces that determine equilibrium and estimates a fair-value range conditional on those forces. For interest rates, inflation is central; for equity valuation, real interest rates help determine how high price-to-earnings multiples can reasonably be. Current values are then compared with this moving range rather than a fixed full-history mean. A gap can improve forecasts over five or ten years, but it offers little precision about the next year. The framework preserves the useful intuition of mean reversion while avoiding the dangerous assumption that changing economies must return to one timeless numerical average.
Origin
Extracted from Masters in Business. Davis explained Vanguard research that adjusted market valuation and interest-rate expectations for changing economic conditions rather than fixed historical means.
Core principles
- 01Mean reversion requires the correct mean.
- 02Fair value changes with economic conditions.
- 03Low rates can justify higher valuation multiples.
- 04Inflation anchors the sustainable level of nominal yields.
- 05Valuation informs long-horizon returns better than near-term timing.
How to run it
- 1
Specify the reverting variable
Choose the yield, valuation ratio, currency measure, or other series under review. Define the horizon over which reversion would matter.
Pro tip Use a horizon aligned with the variable's demonstrated predictive power.
Watch out Do not assume every financial series is meaningfully mean-reverting.
- 2
Identify equilibrium drivers
Determine which structural conditions set the variable's sustainable level. Examples include inflation for nominal yields and real rates for equity multiples.
Pro tip Favor drivers supported by economic mechanism and historical evidence.
Watch out A purely statistical average can conceal major regime changes.
- 3
Estimate a fair-value range
Model the level justified by current drivers and include uncertainty around it. Allow this range to change as underlying conditions change.
Pro tip Use a range because fair value cannot be observed exactly.
Watch out Do not replace one false point estimate with another.
- 4
Measure the conditional gap
Compare the current value with the estimated fair-value range. Determine whether it is below, within, or above the range.
Pro tip Recalculate the gap after material changes in inflation or real rates.
Watch out A large gap does not reveal when it will close.
- 5
Set long-term expectations
Use the gap to adjust expected returns or risks over an appropriate multi-year horizon. Preserve diversification and account for the cost of waiting.
Pro tip Combine the signal with rebalancing rather than all-or-nothing market timing.
Watch out The framework has limited predictability for next year's return.
In the wild
An investor fears a return to the much higher bond yields of the 1970s and 1980s simply because today's yields are low. Conditional analysis asks whether trend inflation will also return to those levels; without that change, a lower long-term yield can remain fair.
→ The investor avoids positioning for a historical average unsupported by current inflation conditions.
Instead of comparing CAPE with one average drawn across all history, an analyst estimates the valuation range consistent with prevailing real interest rates. Lower real rates support a higher multiple, while a move above the conditional range implies more guarded five- to ten-year returns.
→ Valuation becomes more useful for long-horizon forecasting without becoming a short-term market-timing device.
Common mistakes
Reverting to the full-history average
A fixed average ignores changes in inflation, real rates, and other forces that alter equilibrium through time.
Treating fair value as exact
The equilibrium level is uncertain and should be represented as a range, not a precise target.
Using valuation for one-year timing
Conditional valuation can inform five- or ten-year expectations while offering little reliable information about the next year's return.
Is it for you?
Best for
It is best for long-term investors assessing bond yields, valuation multiples, currencies, or other variables whose equilibrium changes with conditions.
Not ideal for
It is not ideal for predicting next year's return or timing an exact market reversal.
From the transcript
“mean reversion is clearly the most powerful force in finance. It's also the most dangerous.”
“My question is what's the danger and mean reversion is what mean are we reverting to?”
“That fair value can change through time with economic conditions and trend conditions”
From the episode
Vanguard's Joe Davis Discusses Global Economics (Podcast)