16 — Decision Traps

Sunk Cost Fallacy#

Statistical1 min read
◆ The PatternContinuing an action because of previously invested resources (time, money, effort) rather than future value.

The sunk cost fallacy — continuing a course of action because of past investment, rather than future expected value.

Past cost = irrelevant → Only future costs and benefits should drive decisions
In trading:
• "I've already lost $5K — I can't sell now" → the $5K is gone either way
• "I've spent 6 months researching this company" → past time doesn't change future outlook
• Doubling down on a loser to "make it back" → escalation of commitment

Rational rule: ignore what you can't change (past), decide based on what you can (future).

Worked example

100 shares bought at 50, now at 35: 1,500 is gone whether you sell or not. The only live question is where the 3,500 should sit. If your honest expectation for the stock is 4% a year and an index fund’s is 7%, holding costs about 105 a year in expected return. “Waiting to get back to even” needs a 43% rise — a target set by the past, not by the stock.

Where it misleads
  • Past costs can still carry information: why you bought, and whether that reason still holds, is relevant; what you paid is not.
  • Taxes can make the price paid matter: a realized loss may be deductible, which favours selling.
  • Sticking with a plan through a drawdown is not the fallacy if the plan was right; the test is whether you would buy it today.
In code
shares, paid, now = 100, 50.0, 35.0
sunk = (paid - now) * shares                     # lost whether you sell or not
hold_view = 0.04                                 # your honest expected return from here
best_other = 0.07                                # e.g. an index fund
cost_of_holding = (best_other - hold_view) * now * shares   # a year's expected shortfall
back_to_even = paid / now - 1                    # the rise "getting out at even" needs
Python, in your browser — edit the code and run it again

Assumes import numpy as np and import pandas as pd. Each snippet carries its own example numbers; the comments say which are made up.

Sources
  • H. R. Arkes & C. Blumer, “The Psychology of Sunk Cost”, Organizational Behavior and Human Decision Processes 35(1), 1985
  • B. M. Staw, “Knee-Deep in the Big Muddy: A Study of Escalating Commitment to a Chosen Course of Action”, Organizational Behavior and Human Performance 16(1), 1976
  • R. Thaler, “Toward a Positive Theory of Consumer Choice”, Journal of Economic Behavior & Organization 1(1), 1980

Last reviewed October 2026 · How this site checks its content · Report a mistake

Check yourself

1. What is the sunk cost fallacy?

Show answer

A. Past costs are the same whichever way you decide now, so they should not affect the choice.

2. Which question cuts through it for an investment?

Show answer

A. If the answer is no, holding it is the same decision as buying it.

3. Buying more of a losing position to “win back” what was lost is…

Show answer

B. Averaging down can be reasonable when the outlook justifies it; the size of the past loss is not a reason.

Pattern bridge: Weighting past investment that can’t be recovered. In ML, early stopping fights sunk-cost thinking: stop training even though you’ve invested epochs.
Linked from 1
← Previous
Information Cascades
Open in the full reader, with the topic sidebar