The Nasdaq Didn't Lose Anything Until You Check the Denominator
Every number in this week's sell-off coverage is a fraction with the bottom half missing. The Nasdaq "wiped out its 2026 gain." The S&P 500 and Dow "entered a correction." Anthropic "slipped 10%." Those are three different measurements of three different things, and reading them as one event is how a normal repricing gets dressed up as a catastrophe.
None of the headlines are wrong. They're just answering questions nobody asked out loud: down from what, measured over what, and compared to how many.
Three headlines, three yardsticks
CNBC's framing stacks two different measures in one sentence. "Correction" is a distance from a high — by the standard definition, a 10% decline from a recent peak. "Wiped out the 2026 gain" is a distance from a date — the final close of 2025. A bear market, the next threshold up, is 20% off the high.
Those numbers can move in opposite directions and still be true at the same time. An index can sit 10% below its high and be exactly flat against last New Year's Eve. Which one matters depends entirely on when you bought, and no headline knows when you bought.
Fast Company's headline adds a third yardstick: markets "head for steep weekly losses." That's a forecast, not a result, and it's measured over five trading sessions. Keep the tense in mind. "Heading for" and "lost" are different claims.
What "wiped out the 2026 gain" actually divides
To wipe out a gain, there has to have been a gain. So the Nasdaq was up earlier this year, and this week's decline was at least as large as whatever that gain was. The reporting doesn't say how big it was — and that omission is the whole story.
If the index was up a few percent going into November, "wiped out" describes a round trip back to zero, which is annoying and not much else. If it was up a lot, the give-back is genuinely large and the word is earned. Same headline, wildly different reality, and the reader can't tell which one they're in without the number.
Here's the cleaner way to think about it: an investor who bought at the last close of 2025 is now roughly where they started the year. The person who bought at the 2026 high owns the entire decline. The index is identical for both of them. The denominator that decides who got hurt is cost basis, and it lives in your account, not in the tape.
Anthropic's 10% needs a sample size
WSJ reports Anthropic slipped 10%, and CNBC says it led the AI sell-off. Ten percent is a big number with a small question attached: ten percent of what, priced by whom?
If that figure is a traded price, it reflects real transactions by real sellers. If it's a valuation mark — a funding round, a secondary sale, a negotiated number between a handful of parties — then it's a sample of one, set by people with an interest in where it lands. A mark is not a market. Before you let one company's 10% explain a broad index move, find out which kind of number you're looking at.
"Leads" also deserves scrutiny. Leadership in a sell-off is a statement about ordering, not causation. Coverage tends to name the most famous company that moved, because famous is easier to write than weighted.
The index is not the average stock
This is the denominator problem that trips up almost everyone. The Nasdaq-100 is a hundred companies, weighted by size — the biggest names decide the headline number. The Nasdaq Composite holds thousands, but it's still cap-weighted, so the giants dominate there too.
Which means "the Nasdaq fell" can be an accurate sentence about a handful of very large companies while the typical stock did something else entirely. The index number is a weighted average, not a median. If you want to know whether this is a broad retreat or a megacap event, watch the equal-weight version of the index and the count of stocks falling versus rising. Those two figures answer the question the headline can't.
A week is five days out of about 250
"Steep weekly losses" covers roughly 2% of the year's trading sessions. Over a ten-year holding period, one week is a rounding error — about a fifth of one percent of your time in the market. That doesn't make a bad week painless. It makes the adjective do the work the number can't. Steep compared to what?
Corrections are not anomalies; they're the recurring cost of owning stocks. What's unusual is a year without one. The useful question isn't whether a correction happened but whether anything underneath it changed — earnings, credit, the willingness of buyers to show up.
What to watch, and what to do
Watch the percentage distance from the high, not the word "correction," and note if it crosses 20. Watch breadth — equal-weight performance against the cap-weighted index — to see whether this is a few stocks or the whole market. Watch whether Anthropic's 10% turns into an actual transaction price or stays a mark. And watch the December 31 close, because that's the denominator that will decide how 2026 gets remembered.
For your own account, do the boring arithmetic first. Write down your cost basis and your time horizon, then compare them to the decline. If you're contributing to a retirement account on a schedule, a down week is a discount, not a verdict. If you're holding individual AI names because they went up, ask whether you'd buy them at today's price with today's information. And don't rebalance on a headline written in the present tense about a week that isn't over yet.
Common questions
What does it mean that the Nasdaq wiped out its 2026 gain?
It means the index has fallen back to roughly where it closed at the end of 2025, per CNBC's reporting. That's a comparison to a date, not to a high. An investor who bought at the 2025 close is near flat for the year; anyone who bought at the 2026 peak owns the full decline.
Is a correction the same as a bear market?
No. A correction is conventionally a 10% decline from a recent high; a bear market is 20%. CNBC reports the S&P 500 and Dow entered correction territory. The threshold tells you how far prices have fallen from the peak, not how they compare to your purchase price or to earnings.
Why does one company's 10% drop move the whole index?
It usually doesn't, by itself. Major Nasdaq indexes are weighted by company size, so the largest names drive the headline number. A single stock's move matters through its weight, and through whether it signals something about the group. That's why breadth measures — equal-weight performance and the count of decliners — are worth checking alongside the index.
Should I sell after a week like this?
That's a personal financial decision, not something a headline can answer. The arithmetic worth doing first: your cost basis, your time horizon, and whether the reasons you bought still hold. A bad week is roughly 2% of a year's trading sessions, which is small relative to a long holding period but real if you need the money soon.