Behavioral finance
The measured gap between what funds return and what their investors actually capture is not an information problem, it is buying euphoria and selling panic with conviction, and this doc is about the process that keeps that machine from reaching the portfolio.
Everything earlier in this school assumes the person executing it functions under pressure, and that assumption fails on schedule. The measured gap between what funds return and what their investors capture, several points a year in some studies, is not an information problem. It is buying euphoria, selling panic, and doing both with conviction. This doc is about the machine that runs all the other docs.
The five that actually cost money
The bias catalog runs to hundreds; portfolios are wrecked by a recurring handful:
| Bias | What it does | Its signature in a portfolio |
|---|---|---|
| Loss aversion | Losses hurt ~2x gains; you hold losers to avoid making the loss real | A museum of down positions with dead theses |
| Confirmation bias | You seek what agrees and rate it higher | A research feed that never disagrees with the book |
| Recency | The last regime becomes the forecast | Max equities in 2021, max cash in March 2009 |
| Overconfidence | Certainty rises faster than accuracy | Sizes creep past the rules on the ideas that feel best |
| Herding | Comfort in the crowd; agony outside it | Buying what compounded for others AFTER it compounded |
The uncomfortable finding, replicated repeatedly: education about biases barely reduces them in the moment, because the moment is the problem: stress narrows thinking exactly when the money is on the line. Professionals do not defeat their biases; they build processes that route decisions AROUND the compromised moments. That reframe, from self-improvement to process design, is this entire doc.
The process designs, mapped to the failure they block
- The written thesis with a dated falsifier blocks revisionism: the position cannot drift into being held for a new reason nobody underwrote. (The thesis doc's five lines are anti-bias machinery wearing research clothes.)
- Sizing rules and caps set in advance block overconfidence at the exact moment it peaks, because the cap does not care how sure you are.
- The drawdown ladder blocks panic improvisation: the de-risking decision was made years earlier by a calmer person with the same name.
- Rebalancing on bands forces buying fear and selling greed mechanically, converting the crowd's cycle from a threat into a small income.
- The decision journal (what you believed, when, at what price, expecting what) makes hindsight bias auditable: the record disagrees with the flattering memory, in writing.
- The pre-mortem and a designated dissenter institutionalize disagreement so confirmation bias must argue with someone whose job is arguing back.
The same forces, at market scale
Aggregated, these biases are not noise; they are the mechanism behind the anomalies the factor doc described. Momentum exists because investors underreact and then herd. Value exists partly because extrapolated despair oversells. Bubbles and crashes are recency and herding compounding in public. This is the deep link of the school: the behavioral failures in THIS doc are the funding source of the systematic returns in the others. The market pays disciplined investors out of the pockets of undisciplined ones, and every process above is how you choose which side of that transfer you are on.
The close: what the whole school compresses to
- Understand the business before the model, and the model before the position (the valuation track).
- Know the object you are trading and what actually drives it (the asset-class track).
- Measure honestly, optimize humbly, distrust backtests (the quant track).
- Know the weather (macro), write the thesis, match the instrument, size to survive (the process track).
- And then: protect all of it from the person holding the controls. That is this page, and it is the one that compounds.
- Behavior gap
- The measured difference between fund returns and what investors in those funds actually capture, several points a year in some studies: the cost of buying euphoria and selling panic, on a schedule.
- Loss aversion
- Losses hurting roughly twice as much as equal gains please. The engine of holding losers (selling makes the loss real) and of most drawdown-era mistakes.
- Disposition effect
- Selling winners quickly and clutching losers, the portfolio-level fingerprint of loss aversion: realize the pleasure, defer the pain, hold a museum of dead theses.
- Confirmation bias
- Seeking and over-weighting whatever agrees with the position. Its antidote is structural, not moral: pre-mortems, designated dissent, and deciding where disconfirming data would appear before looking.
- Recency bias
- The last regime becoming the forecast: maximum equities after the rally, maximum cash after the crash. The bias factor investors harvest from everyone else.
- Overconfidence
- Certainty rising faster than accuracy, peaking exactly when sizing discipline matters most. The reason caps exist that do not care how sure you are.
- Herding
- The comfort of the crowd and the agony of standing apart. Aggregated, it builds momentum and bubbles; individually, it buys what compounded for others after it compounded.
- Hindsight bias
- The past reorganizing itself into having been obvious. Makes every crash 'predictable' afterwards and every lesson unlearnable; a dated decision journal is the only audit that survives it.
- Decision journal
- A dated record of what you believed, at what price, expecting what, decided when. The instrument that lets you grade your PROCESS rather than your luck, and the only witness hindsight bias cannot bribe.
- Pre-mortem
- Assume the position failed and explain why, before entering. The three most plausible failure stories are the real risk list, and better than any generic one.
- Falsifier
- The observable fact, chosen in advance, that would prove the thesis wrong and trigger exit. Deciding it while calm is cheap; deciding it mid-drawdown is expensive, which is why professionals write it down first.