Recency Bias and Historical Patterns
Use this file when: Checking your analytical framework for recency bias. The practical question: is your view driven by what has happened in your lifetime, or by what has happened throughout history?
The Primary Error Source
“My biggest mistakes in my career came from missing big market moves that hadn’t happened in my lifetime but had happened many times before in history.” — a-00127
Dalio’s explicit autobiography of error: every major miss in his career was a recency-bias failure. This is systematic — it affects all investors because humans anchor on personal experience. Events that haven’t occurred in ~40 years of an investment career are systematically underweighted.
Current Application: The “Unprecedented” Problem
“I believe that the times ahead will be radically different from the times we have experienced so far in our lifetimes, though similar to many other times in history.” — a-00126
The framework for 2025: current conditions feel unprecedented (global late-cycle, major power conflict, AI disruption) but they map closely to prior historical periods (1930s, 1970s, post-WWII). The investor who has only operated in 1990-2020 has no personal experience with sovereign debt crises, hyperinflation, or great-power conflict — these are systematically underpriced in their models.
The Consensus Extrapolation Error
“There is always a current most popular meme that just about everyone believes. It is reflected in the price and is bound to be wrong in some way. These memes are due to a mix of extrapolating what happened before and emotional considerations. Also, most investors don’t take into consideration market pricing.” — a-00272
The two failure modes work together:
- Extrapolation: “What worked before will keep working” (momentum)
- Ignoring pricing: “This is a great investment” without checking if it’s priced in
The consensus is often correct directionally but wrong about magnitude or timing. The opportunity is the second-order event: what happens when the consensus view breaks down?
Maximum Uncertainty = Maximum Opportunity
“What I don’t know is much greater than what I do know, and as I write this in early March 2025, I am at a maximum point of uncertainty.” — a-00125
Counter-intuitive: maximum uncertainty is when pricing of tail scenarios is most attractive. When consensus has high conviction (certainty), tail scenarios are priced out. When uncertainty is maximum, tail hedges are cheap and scenario dispersion is wide.
Practical Anti-Recency Checklist
Before forming a view, answer:
- What is the historical base rate for this condition (not just post-1990)?
- Have similar conditions occurred before in history? What happened?
- What is the consensus view? Is it reflected in prices?
- What scenario is the consensus most wrong about?
- What would need to happen for an event that “hasn’t happened in my lifetime” to occur?
Inference Build a personal "base rate library" — historical frequencies for the key macro events (debt crises, hyperinflations, reserve currency transitions, great-power wars). When pricing assets, use these base rates as a prior, updating with current conditions. The prior should be: "this has happened before; will it happen again in my investment horizon?"