Retail investors sold U.S. equities for a ninth consecutive week, according to Bank of America client flow data summarized by Investrade on September 30, and they did it in a week when the S&P 500 gained 1.2%. The index closed October 2 at 7,722.72, about 1.2% below its August high. Nine weeks of net selling into a market that stays near its peak is not what a fear-driven exit looks like. It is the signature of the disposition effect.
Hersh Shefrin and Meir Statman named the effect in 1985, arguing that investors hold losing positions too long and sell winning positions too early. Terrance Odean tested the claim in 1998 on the trading records of 10,000 brokerage accounts and found that investors realized 14.8% of their available gains but only 9.8% of their available losses. The mechanism is loss aversion, the asymmetry Daniel Kahneman and Amos Tversky formalized in 1979: a realized loss registers as a failure, while a realized gain registers as a win that can be banked before it disappears.
Mental Accounting Keeps the Bond Losses on the Books
The same investors hold a second account that looks very different. The U.S. bond market just finished its worst quarter this century, and the 10-year Treasury yield touched 5.304% on September 30, its highest since May 2002. In Canada, the 10-year Government of Canada yield reached 4.042% on October 1, a nearly three-year high, before closing the week at 3.945%. That is roughly 79 basis points above the 3.16% low of February 27.
Richard Thaler described the resulting behaviour in 1985 and again in 1999 as mental accounting: people sort money into separate mental accounts and judge each one on its own terms. A stock account with a gain and a bond account with a loss are not netted against each other. The gain is easy to realize and the loss is easy to ignore, which is the disposition effect operating across accounts instead of within one. Flow data cannot confirm that this is what retail clients are doing, but the pattern fits.
A Calm VIX Is Not a Calm Market
Equity volatility has not registered any of this. The VIX closed at 15.31 on October 2, down 6.59% on the day and below its 15.95 average across the 19 sessions since September 8. A reading that low next to the worst Treasury quarter of the century suggests equity investors are watching the stock account and not the bond account.
The VIX has closed between 14.21 and 17.84 every session since September 8, a narrow band that held through the week the 10-year Treasury yield reached its highest level since 2002.
The September 30 marker shows the day the 10-year Treasury yield touched 5.304%, its highest since May 2002. The average covers the 19 sessions from September 8 to October 2.
Canada Is Living a Different Tape
Canadian investors are not experiencing the American record-high headlines. The S&P/TSX Composite closed October 2 at 35,502.65, about 3.9% below its closing high of 36,957.60 on August 25. The loonie touched an 18-month low of C$1.4263 per U.S. dollar on October 1 and traded near C$1.4257 on October 2, its fourth straight weekly decline.
The gap between Canadian and U.S. two-year yields widened to about 157 basis points, the widest since February 2025, according to Reuters. The Bank of Canada next decides on October 28, with September CPI due October 19, and traders priced a 65% probability of at least a 25-basis-point hike in late September, according to LSEG data cited by Reuters.
A policy rate that has sat at 2.25% for seven straight decisions is a powerful anchor. Kahneman and Tversky showed in 1974 that people adjust too little from an anchor even when the evidence moves against it, and Canadian households with renewals ahead are the most exposed to that error.
What the Flow Data Cannot Show
The same Bank of America note said selling was led by institutional clients, who had bought the prior week, and that hedge fund clients were net buyers for a second week. Retail selling is one current in a market with several.
The note also flagged more tax-loss selling ahead. That would run against the disposition effect: Odean found that sales of losing positions rise in December, when the tax benefit becomes concrete. Whether the retail pattern persists into December is the cleanest test of the reading offered here.