Bayesian Adaptive Market Model
Continuously update market beliefs as distributions and data points move
- Difficulty
- Expert
- Time to result
- ~ongoing to results
- Steps
- 5
- Confidence
- 96%
This mental model treats market probability distributions as living systems. A bell curve is only a depiction; actual observations form a scattered set whose center, width, asymmetry, and tails shift as events unfold. The analyst begins with prior assumptions, incorporates new evidence, and updates both the estimated distribution and the expected rate of change. Attention is not restricted to a dramatic right-tail event: movement in either tail, overall flattening, or migration of ordinary observations may matter. The output is an adaptive forecast or portfolio stance that changes with the evidence rather than defending a static model. Applied consistently, Bayesian reasoning makes uncertainty explicit and turns updating into a normal operating discipline instead of an admission that the original forecast failed.
Origin
Extracted from Masters in Business, where David Kotok identified Bayesian theory as the investing concept he wished he had respected more earlier in his career.
Core principles
- 01Treat probability distributions as changing rather than fixed
- 02Update beliefs when new evidence arrives
- 03Monitor tails, shape, and rates of change
- 04Use estimates without pretending uncertainty has disappeared
How to run it
- 1
Define the prior
Record the current assumptions, probability distribution, and evidence supporting the investment view.
Pro tip State uncertainty ranges rather than only a point forecast.
Watch out An unstated prior cannot be updated transparently.
- 2
Observe new evidence
Collect data and events relevant to the distribution, including evidence that contradicts the prevailing view.
Pro tip Track the provenance and timeliness of each observation.
Watch out Selective evidence creates false confidence.
- 3
Map distribution shifts
Examine movement in the center, dispersion, right tail, left tail, and overall shape.
Pro tip Treat the curve as a visualization of changing observations, not a permanent law.
Watch out Do not focus only on spectacular black-swan outcomes.
- 4
Estimate the rate of change
Assess how quickly the observations and probabilities are moving rather than merely noting their current level.
Pro tip Compare multiple update periods to distinguish noise from acceleration.
Watch out A rapid change can invalidate a formerly reasonable position.
- 5
Update the decision
Revise forecasts, confidence, and portfolio exposure in proportion to the new evidence.
Pro tip Predefine what evidence would trigger a material change.
Watch out Updating too little is stubbornness; reacting to every data point is overfitting.
In the wild
An investor begins with a distribution centered on moderate inflation. New wage, supply-chain, productivity, and bond-market observations shift the center and alter the tails. Instead of declaring the first forecast right or wrong, the investor updates probabilities and adjusts duration exposure gradually.
→ Portfolio positioning evolves with evidence while preserving explicit uncertainty.
Common mistakes
Treating the curve as fixed
Historical observations do not guarantee that the center, width, or tails remain stable under new conditions.
Watching only black swans
Meaningful changes can occur throughout the distribution, not solely in extreme right-tail events.
Refusing to revise the prior
Bayesian reasoning requires adaptation; defending an obsolete assumption defeats the framework.
Is it for you?
Best for
It is best for investment and economic decisions where evidence arrives continuously and outcome probabilities evolve.
Not ideal for
It is not ideal when probabilities cannot be meaningfully estimated or decision-makers will not revise prior beliefs.
From the transcript
“It's Bayesian theory applied in finance and economics and probably a whole bunch of other things too.”
“the data points in those scatter plots are moving, and you have to examine those shifts and the rates of change in them.”
“it's a living thing, not a static thing.”
From the episode
David R. Kotok on Fed Studies of Pandemic Shocks (Podcast)
David R. Kotok