Nasdaq Record Close Comes Just as the Bulls Turned Cautious

Nasdaq - a record close in the worst month of a midterm year, right after the bulls gave up

The Composite closed at 27,122.09, its first record since 2 June, six days after Ed Yardeni cut his year-end target from 8,400 to 7,900 and told clients to proceed with caution. Cyber is 0.04% from its high and cloud 1.2%, with the chip funds 11% to 15% below theirs and room to run. The weakness is not in technology: the Nasdaq 100 ex-technology cut actually fell on the week.

In this edition

The Morning 10 Tue, Sep 22, 2026 ~90 seconds 08:00 CET

The ten points

The Nasdaq Composite closed on Monday at 27,122.09, up 2.26%, clearing the record it set on 2 June by 28.19 points. The June consolidation, three and a half months of it, has been resolved upwards.

The timing is the part worth sitting with. On 16 September Ed Yardeni - among the most consistently bullish strategists on the street - cut his year-end S&P 500 target from 8,400 to 7,900, lowered his forward multiple from 19.8 to 18.6, and told clients to proceed with caution as yields backed up. The S&P has risen 2.82% in the three sessions since and now sits 1.71% below that reduced target, with three months of the year still to run.

The leadership ladder is legible. Cybersecurity closed 0.04% below its 52-week high and cloud 1.2% below - both, in practice, at their highs. Software is 9.2% below, the semiconductor funds 11.3% and 14.7% below, and the memory basket 24.3% below. The chips are not leading; they are starting a leg with a double-digit distance still to travel.

And the weak part of this market is not technology. The Nasdaq 100 ex-technology cut fell 0.11% over the week the Nasdaq 100 rose more than five percent, and sits 7.07% below its own high. So does the equal-weight S&P, down 0.60% on the week. This is the seasonally worst stretch of the calendar, in a midterm year, after a rate increase - and the setup is pointing at new highs anyway.

  1. Nasdaq Composite 27,122.09, +2.26%, the first record close since 2 June (27,093.90), clearing it by 28.19 points; Monday's intraday high 27,183.93 against an all-time intraday high of 27,190.21
  2. The turn came six days after the bulls capitulated: Yardeni cut his year-end S&P 500 target to 7,900 from 8,400 on 16 September and said proceed with caution; the index has risen 2.82% since and is 1.71% from that reduced target
  3. The leadership ladder: cyber 0.04% below its 52-week high and cloud 1.2% below, against software 9.2%, the semiconductor funds 11.3% and 14.7%, and the memory basket 24.3% below theirs
  4. The weakness is not in technology: the Nasdaq 100 ex-technology cut fell 0.11% on the week and sits 7.07% below its high, the equal-weight S&P fell 0.60% and the Dow 0.28%, all while the Nasdaq 100 gained more than five percent
  5. Meta +11.34% to $741.25, its best day in thirteen months, after the Muse agent reached No. 1 on Apple's US App Store and Wells Fargo raised its target to $796 from $640; Meta Connect opens Wednesday
  6. The tilt to watch is the agentic end leading the complex: ARKK 2.08% below its 52-week high after an 8.66% week, against robotics 14.12% below and memory 24.3% below - unconfirmed, and the thing to test rather than assume
  7. The CPU complex led the single-stock tape: Arm +17.19% to $322.97 and about a third on the week, Intel +12.40% to $122.07, AMD +10.04% to $615.95 and within a fifth of a percent of its 52-week high, Qualcomm +9.22% to $194.12
  8. Our reading that the rate increase may have preserved the bull market by keeping the yen carry trade alive is so far holding: the Bank of Japan raised on Friday and the yen weakened again anyway, to 157.59 from 157.05
  9. Asia would not follow: Seoul opened 2.20% higher at 7,161.61 on a 350.7% memory export print, traded down to 6,986.27 and closed at 7,051.26 (+0.62%); the Taiex rose 0.17% with TSMC down 0.81%; Tokyo is shut for the second of three days
  10. No follow-through overnight: S&P futures 7,835.75 and Nasdaq futures 30,796.50 are within a rounding error of Monday's close; bitcoin 85,281 is 1.7% below its Monday close and gold down a second day; Meta Connect and Cintas Wednesday, BlackBerry, Costco and Trump-Xi Thursday, Micron on the 30th
  1. Nasdaq Composite 27,122.09, +2.26%, the first record close since 2 June (27,093.90), clearing it by 28.19 points; Monday's intraday high 27,183.93 against an all-time intraday high of 27,190.21

    Context

    QQQSPYMAGS

    What
    The Composite closed Monday at 27,122.09, 2.26% above Friday's 26,522.54, and that is the highest close in its history. The previous record, 27,093.90, was set on 2 June; Monday cleared it by 28.19 points, or 0.104%. In the twelve months to Monday the index closed above 26,900 on exactly five days. Four fall in the 28 May to 2 June cluster; the fifth is Monday. Between them sits a March low in the 21,000s and a summer spent rebuilding - the consolidation that began in June has now taken three and a half months to resolve. One qualifier belongs on the record. Monday's session high was 27,183.93 and the highest price the Composite has ever traded is 27,190.21, from June. The closing record was made without a record trade, finishing 6.28 points - two hundredths of a percent - beneath a level it never reached during the day.
    If
    The Composite closes above 27,190.21, which would make the closing and intraday records agree for the first time since June.
    Why
    A closing high and an intraday high are different claims. A closing high says the market was willing to carry the position overnight; an intraday high only says someone paid the price once. The closing record arriving first is the signature of patient accumulation rather than a chase.
    Then
    Today's close is the test, not today's open. A second close above 27,093.90 turns Monday into a level that can be defended; a close back underneath turns it into a single print.
  2. The turn came six days after the bulls capitulated: Yardeni cut his year-end S&P 500 target to 7,900 from 8,400 on 16 September and said proceed with caution; the index has risen 2.82% since and is 1.71% from that reduced target

    Outside view

    SPYQQQTLT

    What
    On Wednesday 16 September Ed Yardeni cut his year-end S&P 500 target to 7,900 from 8,400, a 500-point reduction. His stated reason was the backup in bond yields, with the ten-year above 5%: he lowered his year-end forward price-to-earnings assumption from 19.8 to 18.6 and warned that the risk of a downturn over the next three to six months had increased. He pushed 8,400 out to the middle of 2027 while keeping a 10,000 call for the end of the decade. The S&P closed at 7,551.81 that day. It closed Monday at 7,764.70 - up 2.82% in three sessions, and 1.71% below the target he had just cut to, with three months of the year left. Commentary elsewhere ran the same way, with the do-not-fight-the-Fed argument made prominently by Dan Niles among others. We have not verified those remarks directly and mark them as attributed. The Yardeni figures are from his published research note and the dated reporting on it.
    If
    The S&P takes out 7,900 before year-end, three months early, or the caution proves correctly timed and this is a rally inside a topping process.
    Why
    A well-followed bull cutting a target is a sentiment event, not an earnings event - Yardeni cut the multiple, not the earnings. Markets turning within days of that kind of capitulation is a pattern worth recording, though one instance is an anecdote and not evidence.
    Then
    Keep the 7,900 line on the board as a sentiment marker rather than a valuation one. The interesting question is not whether he was wrong but how quickly the index reached a number that was meant to take a quarter.
  3. Euro-AI 50 Europe’s AI exposure in fifty names — the only house index green on both the day and the week views this summer. Open Euro-AI 50 →
  4. The leadership ladder: cyber 0.04% below its 52-week high and cloud 1.2% below, against software 9.2%, the semiconductor funds 11.3% and 14.7%, and the memory basket 24.3% below theirs

    Structure

    CIBRCLOUIGVSMHSOXXDRAM

    What
    Measure Monday's closes against each fund's own 52-week high and the order is clean. CIBR, cybersecurity, closed $103.09 against $103.13 - four hundredths of a percent away, which is the same thing as being at it. CLOU, cloud, closed $28.93 against $29.29, 1.2% away. Both have effectively broken out. Behind them: IGV, software, 9.2% below its $117.99. SMH 11.3% below $671.83. SOXX 14.7% below $655.95. And DRAM, the memory basket, 24.3% below $81.34 - the deepest drawdown on the board. Read as a sequence rather than a snapshot, that is leadership already at new highs with the semiconductor complex beginning a leg and between eleven and fifteen percent of distance still available before it meets the same test. Memory has twenty-four.
    If
    The chip funds keep closing the gap toward their own highs, or cyber and cloud break out alone and the spread widens.
    Why
    A market where the leaders are at highs and the heavier complex behind them still has room is a different setup from one where everything tops together. It is not a guarantee the gap closes - but the distance is what a leg needs in order to exist at all.
    Then
    Run these six numbers as one dashboard each morning instead of the index level. The day SOXX and SMH join CIBR and CLOU at their highs is the day this stops being a rotation and becomes a broad-based advance.

C — free account

The free C account unlocks points 4 through 10 — the full morning read.

One tap with Google or one email — no password, no card. You are signed in until you sign out, on this browser, from then on.

Join the Look — free

Already joined on this browser? The full edition shows automatically — if it doesn't, sign in again here. Looking for the archive, portfolios and realtime? That is C+.

C · point 11 · members

Today in line 11: the desk's diary entry on a record that arrived at the worst possible moment in the calendar and the best possible moment in the sentiment cycle - why the June consolidation looks finished, why the fragile part of this market is everything that is not technology, and why a yen that keeps falling through a Bank of Japan rate increase may be the single most important number on the board.

The privileged, actionable read — what we do, and at which level — is in point 11, for members only.

Join the Look — it’s free

Free members account · one click · the ten points stay free, always.

Trade the Look AI Cycle 2030 — where AI profits move between 2026 and 2030, as an investable wikifolio certificate. See the wikifolio →

Further reading on Closelook

A daily overview, not advice — an investment diary. Published every trading morning at 08:00 CET. See the Daily Pulse and today’s check-in.