Nasdaq-100 Record High, Weak Breadth: What Eight Previous Episodes Show

Research note. Computed from QQQ daily closes, the closes of today’s Nasdaq-100 members and three Nasdaq breadth series we build every day from our own price database, frozen at the September 22, 2026 close. The methods are in How we checked it below; our editorial and AI standards are in How we use AI.

• ·Trend state. 37.3% of Nasdaq-listed stocks are on a point-and-figure buy signal, the 7th-lowest reading on any record day in this data. The weaker ones were in July 2021 and June 2024.
• ·Leadership. More Nasdaq stocks made a new 52-week low than a new high (-0.7% of issues), the 4th percentile among record days.
• ·Participation. 47 of 100 Nasdaq-100 members closed above their 50-day average, the lowest on any record day here; the prior low was 52.5% in 2025.
• ·The rankings belong to a reconstructed record. All three gauges are built from stocks listed today and, for the third, from today’s members, which lifts past readings and makes this one look rarer than it was. The size of that effect is unmeasured.

QQQ closed at 747.46 on Tuesday, a record by $1.30 over the June 2 close and its first new high in sixteen weeks. It got there in two sessions: up 1.8% on Monday and 1.7% on Tuesday, from 3.6% below the old record on Friday.

Underneath it, the Nasdaq looks like a market in a correction. Two of the three gauges below count every Nasdaq-listed stock in our database, about 2,500 names, most of them small: just over a third are on a point-and-figure buy signal, and more made a fresh 52-week low on Tuesday than a fresh high. The third counts only the hundred members of the Nasdaq-100, the index QQQ tracks, and fewer than half of those are above their own 50-day average. A record in the index can coexist with weakness in the thousands of stocks behind it. It is harder to explain when the index’s own members are not participating either.

Three gauges, one question

A record with weak breadth is a claim about participation, and participation can be measured several ways that do not have to agree. The Bullish Percent Index asks how many stocks are in an uptrend on their own chart. Net new highs asks how many are making fresh highs against fresh lows, which is leadership. The share of Nasdaq-100 members above their 50-day asks how many of the index’s own components are in a short-term uptrend. On Tuesday the answers were 37.3%, minus 0.7% of issues, and 47 of 100.

Against the 366 QQQ record closes since 2016 that have all three readings, that puts the BPI at the 2nd percentile, net new highs at the 4th, and Nasdaq-100 participation at the bottom. The typical record close had 58.7% of Nasdaq stocks on buy signals and 77.5% of the index’s members above their 50-day. Tuesday had about two thirds of each. One caveat sits beside every ranking on this page: the gauges are computed from stocks listed today, so companies that failed or were delisted are missing from the past and later additions to the Nasdaq-100 are counted before they joined. Both lift older readings, so these are the rarities in our reconstruction, and the true ones were somewhat smaller.

It is also, over this summer, a rally in the largest names. From the June 2, 2026 record to Tuesday, QQQ gained +0.2%. The average of today’s 99 Nasdaq-100 members returned -0.6%, the median member -2.9%, and only 42 of the 99 were higher. That equal-weight figure is a description of today’s membership and not an investable series; the fund that used to track all hundred members equally, QQEW, has tracked fifty quality-screened names since December 2025 and cannot stand in for it.

Read the baselines first. Every QQQ record close has been followed by a quarter averaging +2.4%, higher 70% of the time, with each case’s lowest close along the way averaging -5.8% below the entry. Counted as episodes, the same way the weak-breadth rows are, records averaged +4.3%, because the first record after a gap tends to come early in a run. Against the episode baseline, two of the three single gauges trailed: weak buy-signal breadth returned +2.2% and weak new highs +3.1%, each two points or more under the +4.3% for records counted the same way, on 18 and 22 cases. Weak Nasdaq-100 participation returned +4.6%, in line with it, helped by September 2025, when records printed with about half the members above their 50-day and QQQ rose 7% over the following quarter. All three sit above the day-level baseline of +2.4%. Which baseline you pick moves the answer by two points, which is about the size of the effects being measured.

The intersection is where the numbers turn on how you count. The 8 episodes with all three gauges weak averaged +2.0% over the quarter, 6 of 8 higher. Every qualifying day, counted separately, averaged -0.3% across 26 days, 38% higher, because the long stretches of qualifying days in mid-2024 and early 2025 each count many times. Spaced at least six months apart, 5 episodes remain and they averaged +2.8%, 4 of 5 higher. A result that ranges from -0.3% to +2.8% depending on the counting rule is not a signal. It is a small sample.

The 8 earlier weak-breadth records

These are the records on which all three gauges were in their weakest quarter: at or below 52% on buy signals, +2.3% on net new highs and 70% on participation. Tuesday was well inside all three cuts, at 37.3%, -0.7% and 47%, so these are broadly weak-breadth records rather than matches for Tuesday’s readings, which are lower than any of them on buy signals and on participation, though not on net new highs. 9 episodes, 8 with a completed quarter, 7 with a completed six months.

5 of the 8 were lower a week later, a mild short-term lean. By a quarter 6 of 8 were higher, median +3.1%, in line with an ordinary record. The two that were not ordinary are December 13, 2024 and February 19, 2025: the first was down 9.4% a quarter later and the second lost 11.2% in a month, and each went on to close more than 20% below its record. They were two qualifying episode starts before the same selloff (QQQ set further records between them, on December 16 and February 18), and both reached their lowest close on the same day, April 8, 2025. They are two entry dates into one decline, which is why the spaced count above keeps only one of them.

One exploratory observation, offered as that. Two earlier cases had more new lows than new highs on the record day: December 2024, which went on to the 21.6% decline, and May 2026, whose six months are not up yet. Tuesday shares that reading. February 2025, which fell just as far as December, had positive net new highs, so the reading is neither necessary nor, on this evidence, sufficient. I would not build a rule on it. It is the reason the first marker below asks for new highs to outnumber new lows as well as for the buy-signal share to recover.

What I would watch

The observation is strong and the forecasting evidence is thin. I would not sell the Nasdaq because its record is unconfirmed: in this sample, that would have been wrong more often than right at a quarter. I would not treat the record as an all-clear either, because the two worst cases in the table were also unconfirmed records, and one of them was falling within a week. The three markers below are prospective observations. Each says how I would reinterpret this record if it fires; none is a tested trading rule, and this study tests none.

Marker A is confirmation, and it uses the two Nasdaq-listed gauges only: the buy-signal share back at 52% or higher, the weakest-quarter line, with new highs outnumbering new lows on the same day, within 42 sessions. It leaves out Nasdaq-100 participation because that gauge is the one most exposed to the membership problem, and because a hundred large stocks recovering their 50-day averages while the other two thousand do not would be the same narrow rally in a different costume. If A is met, I would read the record as early rather than hollow and say so here. Marker B is QQQ closing 5% below the record. If B fires, the divergence resolved the way the two 2025 cases did, and the question this piece asks is closed. Sequence matters when both fire: A and then B means this confirmation failed, which counts against treating a breadth recovery as a tell, though one failure does not settle that either; B and then A is a decline followed by a recovery, and I would score it as B. Marker C is a higher record with the buy-signal share still under 52, the weakest-quarter line. If it fires, the Nasdaq will have set a second record without breadth reaching even the bottom of the normal range, which extends the observation rather than resolving it, and I would add the new case to the table. Participation was the weakest on any record day in our reconstruction, and the other two gauges were unusually weak. These precedents do not establish a dependable sell signal, and 8 cases that turn on the counting rule cannot.

How we checked it

Every number here comes from QQQ daily closes, the closes of today’s Nasdaq-100 members and three breadth series we compute from our own stock database. In plain terms:

• ·A record close is a QQQ close above every earlier close, since the fund began in 1999. There have been 433; the 366 since August 2016 have readings on all three gauges and form the study.
• ·The three gauges. The Nasdaq-listed Bullish Percent Index is the share of Nasdaq stocks whose point-and-figure chart (1% box, three-box reversal) last gave a buy signal. Net new highs is Nasdaq-listed 52-week highs minus 52-week lows, as a percent of Nasdaq issues that day. Nasdaq-100 participation is the share of the index’s current members closing above their own 50-day average.
• ·Weak means at or below the bottom-quartile cut of that gauge across all record days: 52% for the BPI, +2.3% for net new highs, 70% for participation. The cuts use the whole record, so they were not known in 2017. An episode starts on the first qualifying record day after 21 sessions without one.
• ·Returns are QQQ price changes without dividends, graded only where the window is complete. “Worst loss from entry” in the group table is the AVERAGE across cases of each case’s lowest close against its entry close; it is not the single worst outcome and not a peak-to-trough drawdown. “Low vs record” is the lowest close of 126 sessions against the record in force on day one. The spaced row keeps only episodes more than 126 sessions after the previous kept one, so two entries into one decline count once.
• ·Survivorship. All three gauges are computed from stocks in our database today. The two Nasdaq-listed series count every listed stock we hold, so delisted names are missing from past readings; the Nasdaq-100 series uses today’s members all the way back, so past readings include later additions and exclude removed members. Both effects bias old readings up, and they make Tuesday look rarer than it truly was by an amount we have not measured. Every ranking, cut and episode on this page belongs to that reconstructed record, and nothing here is validated predictive evidence. The equal-weight comparison uses today’s members for the same reason and describes this summer only; it is not an investable series.
• ·What is frozen and what updates. The breadth counts are rebuilt daily from the current stock list, so a past reading can change. The tables and the chart markers are rendered from a snapshot saved at publication (Sep 22, 2026) with a SHA-256 fingerprint of each of its 104 input files (QQQ, the three breadth and membership files, and one price file per Nasdaq-100 member), available for download below. The latest-reading box and the marker panel read current data.