Rebias the Investment Process
Design the decision environment so predictable biases support better actions
- Difficulty
- Advanced
- Time to result
- ~weeks to results
- Steps
- 5
- Confidence
- 96%
Rebiasing starts from the premise that investors cannot remove their evolved emotional responses through willpower alone. Instead, they should identify predictable failure modes—such as becoming reluctant to invest immediately after a loss—and redesign the process around them. Before stress arrives, the investor establishes objective valuation signals, decision thresholds, and rules for challenging those signals. When falling markets activate fear, the process deliberately presents evidence that may point toward buying rather than retreating. Human judgment remains available, but an override must explain what the model has missed. This converts a recurring vulnerability into a structured decision prompt. The output is not emotionless investing; it is an environment in which emotion has less power to dictate action without evidence.
Origin
Extracted from Masters in Business, where James Montier applied the psychological idea of rebiasing to GMO's valuation process during bear markets.
Core principles
- 01Human reactions cannot simply be switched off
- 02Fear evolved for survival rather than investing
- 03Recent losses distort subsequent risk decisions
- 04Processes should anticipate predictable behavioral failures
- 05Objective signals can redirect emotion-driven behavior
How to run it
- 1
Map the behavioral failure
Identify the emotional response that repeatedly damages decisions, such as avoiding investment after a loss. Describe its trigger and typical consequence.
Pro tip Use past decisions rather than hypothetical self-assessments.
Watch out Generic labels like irrationality are too vague to redesign a process around.
- 2
Choose a counter-signal
Select an objective measure that becomes informative when the bias is strongest. For investing, this may be prospective return or another valuation measure.
Pro tip Keep the counter-signal visible and understandable during stressful periods.
Watch out A complicated signal may be ignored precisely when it is most needed.
- 3
Precommit to action rules
Define what the signal will prompt before the emotional trigger occurs. Include thresholds, position limits, and who may approve an exception.
Pro tip Create the rules during calm conditions.
Watch out Rules invented during a crisis are easily tailored to justify avoidance.
- 4
Raise the override burden
Require anyone rejecting the signal to state what is wrong with its assumptions and provide supporting evidence. Fear alone does not qualify.
Pro tip Ask what specific fact the model does not know.
Watch out Do not make the burden so absolute that genuine structural change cannot be recognized.
- 5
Audit the behavior
After the event, compare the process-driven decision with the instinctive alternative and subsequent evidence. Refine the counter-signal or rules where necessary.
Pro tip Evaluate process quality separately from one lucky or unlucky outcome.
Watch out Outcome bias can cause a good process to be discarded after a single loss.
In the wild
As markets fall, fear encourages investors to stop buying after recent losses. GMO's prospective-return models instead display improving expected returns, forcing a discussion about whether the models are missing something rather than allowing fear to settle the decision automatically.
→ The team receives a structured prompt to consider buying when instinct favors withdrawal.
A product team repeatedly delays launches after one failed release. It defines evidence-based readiness thresholds in advance and requires specific unmet criteria, rather than anxiety about the previous failure, to postpone the next launch.
→ Recent failure informs safeguards without creating indefinite paralysis.
Common mistakes
Trying to eliminate emotion
The framework assumes people remain human. Its purpose is to redesign choices around emotion, not to pretend emotion can be removed.
Using an unvalidated anchor
A process can consistently produce bad decisions if its counter-signal is poorly designed. Test the model before increasing its authority.
Forbidding all judgment
Rebiasing should increase the evidence required for an override, not eliminate thoughtful exceptions.
Is it for you?
Best for
It is best for individuals and teams whose sound plans repeatedly collapse during volatility or poor performance.
Not ideal for
It is not ideal when the underlying model is untested, structurally obsolete, or disconnected from the risks being taken.
From the transcript
“if you cannot d buy us, then re buy used.”
“the best thing to do is to try and, knowing that they're going to fall into these behavioral pitforces, to design a process that will…”
“Our value models are all else being equal, going to be saying, hey, look things are getting cheap, you should be buying UM.”
From the episode
James Montier on Fear and Investment (Podcast)
James Montier