Retail Investors Have Bought US Stocks 22 Days Straight. Why Wall Street Eyes a Year-End Bounce
Stocks are falling and retail investors keep buying. Corporate buybacks are about to reopen. If institutions that have been selling turn buyers too, what happens?
The 20-second read
1. Retail investors did not sell the dip
Retail investors were net buyers of US cash equities for 22 straight sessions, according to Citadel Securities. They bought an average of $1.5 billion a day, about $33 billion in total. It is the sixth-longest buying streak since 2017.
Net retail purchases this year have reached $251 billion, topping the full-year 2021 record of $241 billion. Notably, buying picked up on down days.
One caveat: the figures are based on trades observed on Citadel Securities' platform. They are not a complete count of all US retail trading.
In 12 past episodes of 15 or more consecutive days of net buying, the S&P 500 rose in 10. The median gain was 2.2%. That does not mean there is an 83% chance the market rises from here. It is a historical observation that retail buying may cushion declines.
2. A second and third buyer could appear into year-end
Retail investors are not the only possible buyers.
US companies had authorized $1.5 trillion of new buyback programs through Nov. 2 this year. Citadel estimates that around Nov. 9, about 85% of S&P 500 market capitalization will be out of the blackout windows around earnings that restrict repurchases.
But authorized amounts are not always fully executed. Companies adjust the pace based on cash flow, share price and their financial position.
Another variable is CTAs, institutional strategies that trade automatically on price trends. In Citadel's model, their US equity positioning has fallen to -3.3 standard deviations.
If prices rebound and the trend turns, CTAs could cover shorts or rebuild equity exposure. If the downtrend continues, selling pressure remains.
Potential sources of buying
- Retail investors: steady net buying during declines
- Public companies: buyback windows widen
- Trend-following institutions: could turn to buying once a rebound is confirmed
- Result: extra demand that could support a year-end rise. The three do not necessarily happen at the same time.
3. AI earnings rose, but the stocks got cheaper
The more interesting shift is in semiconductors and tech.
From June 30 to early November, 12-month forward EPS estimates for the semiconductor sector rose 30%, while the forward P/E fell 35%. The Nasdaq 100's forward P/E dropped to 19.1 times.
This suggests that expectations for AI growth have not vanished. Rather, the price investors are willing to pay for that growth has come down.
Still, a lower P/E alone does not make a stock undervalued. If returns on AI capital spending fall short of expectations, future earnings estimates could be cut.
The more important question than whether shares got cheaper is whether current earnings forecasts hold.
Earnings forecasts and valuations moved in opposite directions
| Sector / index | Forward EPS estimate | Forward P/E |
|---|---|---|
| Memory | +32% | -50% |
| AI infrastructure | +31% | -36% |
| Semiconductors | +30% | -35% |
| Nasdaq 100 | +14% | -19% |
Change from June 30 to early November 2026. Earnings estimates rose while P/E multiples fell by more, meaning share prices rose less than earnings, or fell. Source: Citadel Securities / Bloomberg.
4. A market collapse is not the same as a leadership rotation
US stocks swung sharply in October, led by tech. But over the 20 sessions through Nov. 2, the equal-weighted S&P 500 gained more than 4%, while the Nasdaq 100 fell almost 10%.
That allows a reading in which investors did not dump US equities across the board. Instead, they moved money out of large-cap tech and into health care, utilities, consumer staples and similar sectors.
Even so, Big Tech's weight in the index remains overwhelming. The top 10 S&P 500 companies make up 37% of the index. Even if the rest of the market rises, a large drop in the megacap tech names could pull the whole index lower.
- No. Historical data shows only that this buying can cushion declines. It does not rule out a sharp drop ahead.
- No. That figure is the size of newly approved programs. It is different from the amount actually carried out and when it is carried out.
- No. It is an aggregate for a specific S&P 500 semiconductor industry group. It should not be confused with the P/E of an individual company such as Nvidia or of semiconductor ETFs as a whole.
- Citadel analyzed that the S&P 500 was higher both 6 months and 12 months after every midterm election since 1950. But the sample is limited and economic conditions can differ, so there is no guarantee it will rise this time.
- Rather than rushing to chase a year-end rally for fear of missing out, a reasonable approach is to review the earnings outlook of the companies you own. It also makes sense to keep your cash level and staged-buying plan in line with your own risk tolerance.
What to watch
Insight Times Editorial Desk
