Prospect Theory

The Core Model

Kahneman and Tversky published prospect theory in 1979 — Kahneman won the 2002 Nobel Memorial Prize in Economic Sciences for the work, later synthesized in Thinking, Fast and Slow (2011) — as a direct challenge to expected-utility theory's assumption that people evaluate outcomes by final wealth. Instead, people evaluate outcomes as gains or losses relative to a reference point — commonly the purchase price, a prior high, or whatever number feels like the baseline — and the same dollar amount carries different weight depending on which side of that reference point it falls. The theory's second departure from expected utility is probability weighting: people overweight small probabilities and underweight large ones, which is why long-shot bets and near-certain outcomes both get mispriced relative to their true odds.

Loss Aversion and Its Consequences

The theory's most cited finding is that losses loom roughly twice as large as equivalent gains — a loss of $100 hurts about as much as a gain of $200 feels good, a ratio that has held up across dozens of replications in both laboratory and field settings. This asymmetry, on its own, explains two of the most persistent patterns in trading behavior: the disposition effect, where investors sell winning positions too early to lock in a gain and hold losing positions too long to avoid realizing a loss, and panic selling, where a drawdown that crosses into loss territory relative to the reference point triggers action disproportionate to the position's actual risk. Both patterns cost real returns, and both trace back to the same asymmetric weighting rather than to any rational reassessment of the position's prospects.

How Closelook Uses It

Money Temperature reads crowd-level reference-point behavior directly — extremes in the gauge often coincide with reference-point-driven selling rather than fundamentals-driven selling, since a crowd sitting near its own cost basis behaves differently than one holding large unrealized gains. See Loss Aversion and Anchoring for the two mechanisms the theory predicts in practice, and Drawdown for the metric where reference-point effects show up most clearly on a chart. None of these tools tell a reader what to do with a position; they describe which bias is likely operating on the crowd at a given moment, which is the most a data layer can honestly claim to know. Holding both readings — the reference-point behavior at the crowd level and the reader's own reference point on a given position — side by side is the practical use of the theory, not a signal to act on either alone.