SPY (the S&P 500 exchange-traded fund, or ETF) ended lower on 45.2% of trading days between 29 January 1993 and 26 October 2026, with dividends counted.
Loss aversion is the finding that a loss weighs more than a gain of the same size. Counting each loss at twice the weight of an equal gain makes the mean day in SPY a loss of 0.32%. The mean day actually returned a gain of 0.05%. The weight of 2 is our illustration. The post also tests 1.5 and 2.5.
What is loss aversion? Daniel Kahneman and Amos Tversky reported in 1979 that people weigh a loss more than an equal gain.
Their 1979 paper reports that even-odds bets to win or lose the same sum are distinctly unattractive. The dislike grows with the stake.
If we count each loss at twice the weight of an equal gain, the mean day, week and month in SPY since 1993 each come out as a loss. The mean year comes out as a gain of 9.41%.
The paper gives the direction of the effect and no multiplier, so the weight of 2 here is an illustration.
What is loss aversion?
A loss of $100 weighs more than a gain of $100.
Kahneman and Tversky set the idea out in "Prospect Theory: An Analysis of Decision under Risk", which Econometrica published in March 1979. Prospect theory is their model of how people choose between risky options. Our guide to behavioural finance covers the other named biases.
What did Kahneman and Tversky find in 1979?
The authors gave students and university faculty hypothetical choice problems. Respondents answered anonymously.
The sums were in Israeli pounds. Groups at the University of Stockholm and the University of Michigan answered some of the problems, with essentially the same pattern.
The paper reports that even-odds bets to win or lose the same sum are distinctly unattractive. The dislike generally grows with the size of the stake. The authors conclude that the value function, which turns a gain or loss into a felt value, is steeper for losses than for gains.
The paper also cites thirty decision makers from five earlier studies. All but one had a utility function that was considerably steeper for losses than for gains.
Why did the same outcome get different answers?
Problems 11 and 12 in the paper leave a person in the same position. In Problem 11, respondents received 1,000. They then chose between a 50% chance to win another 1,000 and 500 for sure. Of 70 respondents, 84% took the 500 for sure.
In Problem 12, respondents received 2,000. They then chose between a 50% chance to lose 1,000 and a sure loss of 500. Of 68 respondents, 69% took the 50% chance.
In final wealth the two problems match.
Each offers 1,500 for sure, or an even chance of 2,000 and 1,000. The sure sum won when the choice was framed as a gain. The gamble won when it was framed as a loss. The paper uses the pair to show that people judge a gain or loss from where they start, not from their final wealth.
How does a loss weight work?
A loss weight is the number we multiply a loss by before setting it against a gain. A weight of 1 treats the two alike. A weight of 2 counts a loss of $100 as $200.
Take an even-odds bet to win $150 or lose $100. The mean outcome is a gain of $25, from 0.5 × $150 minus 0.5 × $100. At a weight of 2, the loss counts as $200, so the bet works out at 0.5 × $150 minus 0.5 × $200, which is a loss of $25.
At even odds, the gain must reach the loss times the weight to break even. A $100 loss needs a gain of $100 at a weight of 1, $150 at 1.5, $200 at 2, $250 at 2.5 and $300 at 3.
Chart 1 plots how a gain or loss counts at each weight. The orange line is a weight of 2, while the dashed grey line is a weight of 1.
Source: YX Insights
The gap between the two lines on the loss side shows the extra weight a loss carries at a weight of 2.
How often does SPY end lower?
A window is a run of consecutive trading days. We started a window on every trading day from 29 January 1993 and measured SPY's return to the end of the window. The windows overlap.
We used six lengths: 1 day, 1 week (5 trading days), 1 month (21), 1 year (252), 5 years (1,260) and 10 years (2,520). The file holds 8,493 prices, which give 8,492 one-day windows. Of those, 3,840 ended lower, 4,590 ended higher and 62 ended unchanged.
The share ending lower falls as the window lengthens. It is 45.2% for a day, 41.1% for a week, 34.1% for a month, 18.2% for a year, 15.2% for five years and 8.0% for ten years. Chart 2 shows all six.
Source: YX Insights price data
Every ten-year window that ended lower began between 19 October 1998 and 7 November 2000. That is 476 windows. No ten-year window starting outside that span ended lower.
When do gains outweigh losses at a loss weight of 2?
We gave each window a felt return. A rise counts at its full size. A fall counts at its size times the loss weight, so a 5% fall counts as 10% at a weight of 2.
The break-even loss weight is the weight at which all the weighted gains and all the weighted losses cancel. We find it by dividing the sum of the rises by the sum of the falls, ignoring signs. A break-even below 2 means that at a weight of 2 the losses win.
Chart 3 gives the break-even weight for the four shortest windows. It is 1.13 for a day, 1.32 for a week, 1.77 for a month and 5.44 for a year.
Source: YX Insights price data
On one-day windows, the mean rise was 0.76% and the mean fall was 0.81%. Rises came on 4,590 days and falls on 3,840, so the rises add up to only 1.13 times the falls.
At a weight of 2, the felt return averages a loss of 0.32% for a day. It is a loss of 0.50% for a week and a loss of 0.29% for a month. For a year it is a gain of 9.41%. The table adds weights of 1.5 and 2.5.
Window | Weight 1.5 | Weight 2 | Weight 2.5 |
|---|---|---|---|
1 day | −0.13% | −0.32% | −0.50% |
1 week | −0.14% | −0.50% | −0.87% |
1 month | +0.34% | −0.29% | −0.91% |
1 year | +10.78% | +9.41% | +8.04% |
Mean felt return of SPY windows from 29 January 1993 to 26 October 2026, with dividends counted. A fall counts at its size times the loss weight. Source: YX Insights price data
Weight 2.5 gives the same signs as weight 2. At a weight of 1.5, the month turns positive at 0.34%, while the day and week stay negative. As a check, we restarted the windows in 2006. The break-even weights were 1.13, 1.33, 1.78 and 5.90, so the pattern holds.
What does the test leave out?
The windows overlap, so a year of 252 days shares most of its days with the next window. The test covers one fund in one market. It counts price changes with dividends. It leaves out tax and trading costs.
The weights are illustrations. The 1979 respondents answered hypothetical questions about sums in Israeli pounds, not about SPY. The test makes no forecast. Multi-model Signals covers SPY.
Our explainer on volatility covers how far SPY moves from day to day. Our guide to drawdowns covers falls from a high. Our explainer on the S&P 500 covers the index.
At a loss weight of 2, SPY's mean day, week and month come out as losses, while the mean year comes out as a gain. The shorter the window, the more often SPY ends lower. The 1979 paper gives the direction of the effect. The size of the weight is our illustration.
Get the next explainer
The Academy is free to read, here on the web. The Long Trend, our free weekly letter, comes by email.
Learn more with YX Insights
This explainer is part of the YX Insights Academy. Each one takes a single idea and checks it against real data.
The same approach runs through everything else we publish:
Systematic Portfolio: ready-made portfolios for Macro & Megacaps and for Commodities. We publish the holdings and every change.
Multi-model Signals: a daily long-or-flat call on every name we cover. Each call shows how strongly our four models agree.
Research: company deep dives, macro commentary and essays on how we test.
Good places to start on the website:
Academy: more explainers like this one.
Start here: get our free weekly letter by email.
Macro & Megacaps Systematic Portfolio and Commodities: the two portfolios in detail.
Track record: live results since 3 August 2026, plus the ten-year backtest.
Upgrade: get the Systematic Portfolio, Multi-model Signals and research in full.
Common questions about loss aversion
What is an example of loss aversion?
An example is declining an even-odds bet to win $150 or lose $100, even though the mean outcome is a gain of $25. If the $100 loss counts as $200, the bet works out at a loss of $25. The weight of 2 is an illustration, not a measured figure.
Is loss aversion the same as risk aversion?
No. Risk aversion is preferring a sure sum to a gamble with the same mean. Loss aversion is counting a loss for more than an equal gain. The two can point in opposite directions. In Problem 12 of the 1979 paper, 69% of 68 respondents chose a 50% chance of losing 1,000 over a sure loss of 500.
What is loss aversion in prospect theory?
In prospect theory, the model Kahneman and Tversky published in 1979, value attaches to gains and losses measured from a reference point. Loss aversion is the property that the value function is steeper for losses than for gains. The paper reports the direction and gives no multiplier.
How often does the S&P 500 fall?
SPY, the S&P 500 exchange-traded fund, ended lower on 45.2% of one-day windows from 29 January 1993 to 26 October 2026, with dividends counted. The share was 18.2% for one-year windows and 8.0% for ten-year windows. Longer windows ended lower less often.
DISCLAIMER: This article is intended for educational purposes only. Any information or analysis in this note does not constitute an offer to sell or a solicitation of an offer to buy any securities. Nothing in this note is intended to be investment advice, nor should it be relied upon to make investment decisions. Any opinions, analyses, or probabilities expressed in this note are those of the author as of the note's date of publication and are subject to change without notice.
