Nasdaq-100 / QQQ: Official Evidence and Vendor Model Research
Contents11 sections
- Executive summary
- Data governance and source timing
- Official versus vendor basis
- Verifiable fundamentals and exposure
- Official performance and risk
- Vendor model and this study's audit
- Conditions under which a valuation judgment holds or fails
- Monitoring indicators and decision limits
- Source register and acceptance
- Further reading
- Frequently asked questions
Research cutoff and access date: 2026-10-30 (UTC+8). This document uses a traceable research format; it is not a formal rating by a licensed institution, a price target, or personal investment advice. Each official source carries its own date; all data have not been assumed to be same-day values as of 2026-10-29.
Site note: This is one of the 2 parts of the series "US equity index institutional research" (this part: Nasdaq-100 / QQQ; the other part covers S&P 500 / SPY). This article reproduces the site owner's research note in full: the content, numbers, tables, citations, links and the "gaps / limitations" passages are published as in the original, without deletion or rewriting. The only changes are removing the table-of-contents block at the top of the original draft and turning links to attachments in the author's research folder and to the shared model audit appendix (not published on this site) into plain text (file names and paths kept). The original file is "Nasdaq100 website data analysis-gpt.md" (translated name) in the author's research folder (US index research), with a research cutoff and access date of 2026-10-30 (taken from the original). The data / scripts attachments cited in the text (CSV, JSON, Python) are in the author's research folder and are not on this website. The original describes itself as using a traceable research format, not a formal rating by a licensed institution, a price target, or personal investment advice; this article is a research record, not investment advice.
Executive summary
- What can be verified is the exposure and the officially published values, not "currently undervalued". On the Invesco product page, QQQ's P/E is 29.48x and its forward P/E is 25.75x, both with a data date of 2026-09-30; the 27.10x on the WorldPeRatio website is a 2026-10-29 vendor value. Dates, earnings denominators and aggregation methods are not yet fully aligned, so the difference cannot be interpreted as an arbitrage or as a degree of undervaluation. O1, V1; both accessed 2026-10-30, unit: times (x).
- Concentration is a verifiable, substantive risk. Based on the official displayed holding weights as of 2026-10-28, this study's sum puts the top ten securities at 46.71% of QQQ's net assets and the top three at 21.89%; the top ten include two share classes of Alphabet, so they are not ten independent companies. O1, accessed 2026-10-30; calculated in this study, unit is percent of net assets, using already-rounded weights.
- High historical returns cannot substitute for future required returns. The official ten-year annualized performance of QQQ NAV is 20.91%, as of 2026-09-30; the total expense ratio listed on the page is 0.18% / year, accessed 2026-10-30, with no separately stated effective date for the fee. Fund-level expenses already reflected over the same period should not be deducted a second time from historical NAV returns; personal trading costs, taxes and FX are separate. O1, O2.
- The model now has a pseudo-OOS with strict maturity purging, but this is not a tradable, truly real-time out-of-sample validation. This study rebuilt it with the current-vintage vendor data; historically available vintages of earnings, constituents and prices were not obtained. The valuation model beat the then-matured sample mean at some horizons; the past-one-year-return model did not; long horizons remain limited by overlap and small samples. See the shared model audit (the author's model audit appendix, not published on this site).
- Conclusion: valuation rating withheld; no buy/sell or overweight rating is given. What can be confirmed is an exposure to large non-financial, growth and large-cap tech-related names; it cannot be confirmed that the vendor's "Undervalued" represents a fundamental discount. Official sector weights, the full definition of valuation aggregation, a reconstructable dividend-inclusive daily series and the maximum drawdown are still gaps.
Data governance and source timing
Evidence tiers
- A: Directly verified on official pages. The Invesco dynamic pages must be read with a normal browser; simplified HTML scraping misses data. The official Nasdaq Overview was also verified in a browser. Only the necessary facts were saved in the official source register (data/NASDAQ100/official_sources.json); no full documents were copied.
- B: Vendor public values. The WorldPeRatio page and its public arrays, saved in the vendor snapshot (data/NASDAQ100/vendor_snapshot.json). Verifying the page's arithmetic is not the same as verifying the original market data.
- C: Calculated in this study. Sums of officially displayed weights, OLS, maturity-purged backtests and error diagnostics; these are not officially published ratings or forecasts.
- D: Research design / unknown. The minimum number of training points is a design parameter; definitions and risk statistics that were not obtained are listed as unknown and are not filled in with arbitrary numbers.
Data timing table
All access dates below are 2026-10-30; each row's source link is the basis for that row's value or statement.
| Data | Actual as-of / document status | Subject, unit and basis | Source |
|---|---|---|---|
| Valuation, P/B, ROE | 2026-09-30 | QQQ portfolio characteristics; x / %; full aggregation specification not obtained | O1 product page |
| Standardized performance | 2026-09-30 | QQQ NAV, market price and the official reference index shown separately; % | O1 product page, O2 performance page |
| Top ten holding weights | 2026-10-28 | Securities as a share of total net assets (% TNA), not equal weights by company | O1 product page |
| Fees, periodic rebalancing and risk text | Current page disclosure; no separately stated fee effective date | QQQ annual expense ratio / description of rules; not forced to 9/30 or 10/29 | O1 product page, O3 fund overview |
| Index scope and number of constituents | Official page data header 2026-10-29 | Nasdaq-100; description of company scope / number of securities | O4 Nasdaq Overview |
| Vendor current valuation and model inputs | Page header 2026-10-29; arrays each carry their own month labels | QQQ proxy; x, annualized return %; not official same-day index data | V1 |
Version governance: The original session report is kept in place; this document does not reuse its arbitrary earnings-growth scenarios as forecasts. The SHA and access times of model inputs and calculation files are in the manifest (data/NASDAQ100/manifest.json); the rerunnable method is in the shared appendix. Official pages may be updated later; the register saves the necessary facts actually seen this time and does not guarantee the same values on a revisit.
Official versus vendor basis
| Question | Officially verified | Vendor / not yet aligned | Research treatment |
|---|---|---|---|
| What is the subject | Nasdaq officially describes large domestic and international non-financial companies listed on Nasdaq; QQQ tracks that index. [O1, O4] | The website uses QQQ as the valuation and return proxy. [V1] | An ETF is not the same as holding the index directly; the Nasdaq-100 is also not the Nasdaq Composite |
| Current P/E | 29.48x, 2026-09-30; a Forward P/E of 25.75x is listed separately. [O1] | 27.10x, 2026-10-29; insufficient information on the original EPS vintage, treatment of negative earnings and aggregation method. [V1] | No direct calculation of a discount, arbitrage or relative cheapness across indices |
| Earnings period and aggregation | The official page lists the ordinary P/E and the Forward P/E separately, but the HTML read this time did not provide enough of the formula. [O1] | The chart title Forward Return refers to future returns; it does not mean the horizontal axis uses Forward P/E | The ordinary P/E is not relabeled as "verified TTM" without basis; nor is a forecast period, averaging or harmonic method assumed for Forward |
| Sector classification | Invesco states that it uses ICB classification. [O3] | Other institutions may use GICS, and AI / semiconductors are not at the same level | Weights under different classifications are not subtracted from each other; no reliably readable official sector weight table was obtained this time |
| Returns | Official NAV, market price and index performance are shown separately. [O1, O2] | The vendor's original price adjustments and treatment of dividends and expenses are not fully verified | The vendor is not assumed to be the official dividend-inclusive NAV series just because the numbers are close |
| Historical consistency of the index | The official product page clearly states that weighting and maintenance methods changed from 2026-05-01. [O1] | The current-vintage arrays have no point-in-time historical versions | Long-run estimates may span methodology regimes; post-change performance cannot be assumed to follow exactly the same rules |
The units of market values in this table are times (x), accessed 2026-10-30; sources O1, O3, O4, V1.
Verifiable fundamentals and exposure
Portfolio valuation and quality proxies
Officially published, QQQ, as-of 2026-09-30, accessed 2026-10-30:
| Item | Value and unit | What it can / cannot say |
|---|---|---|
| Price/earnings ratio | 29.48x | Official portfolio valuation reading; details of the denominator and aggregation are insufficient, so it is not called a fully verified TTM |
| Forward price/earnings ratio | 25.75x | Official forward valuation reading; it does not guarantee earnings growth, and the ratio of the two P/Es cannot be taken directly as consensus EPS growth |
| Price/book ratio | 9.91x | Valuation relative to book value; intangible assets and sector composition affect comparability |
| ROE | 34.54% | Official portfolio characteristic; not each company's ROE, nor the fund's investment return |
Source: O1 Invesco product page. Same-basis time series of company earnings growth, cash flow and profit margins, and independent estimate-revision data, were not obtained, so no growth forecasts or price targets are filled in.
Sector and holding concentration
The official Nasdaq scope is 100 companies that are large non-financial companies; the Overview at that time showed 101 components (page data header 2026-10-29), indicating that the company count and the security count may differ. O4, accessed 2026-10-30. The QQQ product page showed 102 holdings on 2026-10-28, while the holdings table pagination showed 104 rows; this study did not obtain the full table to explain the difference, and does not force either number into an actual company count. O1, accessed 2026-10-30.
Official top ten holdings, as-of 2026-10-28, accessed 2026-10-30; unit is % TNA:
| Security | Name | Weight |
|---|---|---|
| NVDA | NVIDIA | 8.03% |
| AAPL | Apple | 7.76% |
| MSFT | Microsoft | 6.10% |
| MU | Micron Technology | 4.74% |
| AMZN | Amazon.com | 4.25% |
| AMD | Advanced Micro Devices | 4.03% |
| TSLA | Tesla | 3.28% |
| GOOGL | Alphabet | 2.95% |
| META | Meta Platforms | 2.82% |
| GOOG | Alphabet | 2.75% |
Source: O1 official holdings table. Summed in this study from the displayed figures: top ten 46.71%, top three 21.89%, the two Alphabet share classes combined 5.70%; all are percentages of net assets, at the same as-of, affected by rounding, and not a concentration index separately published by the issuer.
Research implication: Changes in the earnings and discount rates of large tech and related growth companies have a significant effect on the whole; different companies may also share factors such as AI capital spending, semiconductor demand, and the cloud and advertising cycles. This is an interpretation of the exposure mechanism, not an estimated factor beta. The official sector chart had no reliable weights in the content readable this time, so individual stock classifications are not summed into a "tech sector share", nor is any precise overlap with SPY assumed.
How far the index rules can be verified
- Official Invesco text: the fund and the index are rebalanced quarterly and reconstituted annually; accessed 2026-10-30, a description of the current rules on the page. O1
- Official product page: weighting and maintenance methods changed from 2026-05-01, and subsequent performance reflects the change; the effective date is not a return observation date. O1
- The Nasdaq Overview provides the official methodology PDF, the change log and the 2026 changes FAQ. The tools used this time did not parse the methodology PDF into readable provisions; there is no claim that current single-stock caps, tiered weight limits or special-adjustment trigger thresholds were checked. Old rules are not used to fill in the current methodology.
Official performance and risk
Official NAV performance, not a price forecast
As-of 2026-09-30, accessed 2026-10-30, nominal USD %, sources O1 and O2:
| Period | QQQ NAV | QQQ market price | Officially listed Nasdaq-100 Index |
|---|---|---|---|
| Year to date, cumulative | 20.85% | 20.88% | 21.01% |
| One year | 23.74% | 23.84% | 24.00% |
| Three years, annualized | 28.05% | 28.05% | 28.31% |
| Five years, annualized | 16.33% | 16.33% | 16.56% |
| Ten years, annualized | 20.91% | 20.90% | 21.16% |
| Since inception, annualized; start date 1999-03-10 | 10.88% | 10.87% | 11.11% |
Evidence boundary of the dividend-inclusive basis: The above is the officially published standardized NAV performance, not price returns calculated by this study from closing prices. The official product page lists the benchmark ticker XNDX, but the HTML readable this time did not spell out the full calculation terms for "on what date and at what price all dividends and capital gains are reinvested"; access to the factsheet was blocked. Therefore the official dividend reinvestment details remain to be verified, and it cannot be claimed that dividend-inclusive NAV performance was rebuilt item by item. Vendor returns are not passed off as this official series, and website data of unknown basis are not used to fill gaps. The official page states that performance reflects applicable fee waivers, periods under one year are cumulative, and market price uses the midpoint of the bid/ask at the US Eastern afternoon close. [O1, O2]
As of the access date 2026-10-30, the official product page listed the management fee, total expense ratio and net expense ratio all at 0.18% / year, with no separately noted effective date for the fee. O1 This current fee should not be assumed to apply to all historical years; do not deduct fees again from published NAV performance. The index does not bear the same management fee, trading costs and implementation frictions as the fund, so the difference between the two columns cannot be attributed only to the fee, nor treated as daily tracking error.
Risk evidence and missing statistics
Officially listed: the fund is not actively managed; concentration in specific sectors such as technology increases risk; the fund is non-diversified and may be more volatile than more diversified investments; foreign issuers carry additional currency, political-economic and tax risks; fund returns may not equal the index. Sources: O1, O2, current risk disclosures on the pages, accessed 2026-10-30; this is a risk classification, not a statistical value.
| Risk question | What can be answered this time | Explicit gap |
|---|---|---|
| Single-name / large-company concentration | Same-day holding weights listed above are verifiable | No full correlation or stress-loss estimates |
| Rates / discount rates and growth stocks | A sensitivity mechanism exists between valuation and long-run earnings expectations | No interest-rate beta, duration, or loss model for a specified rate hike |
| Past maximum drawdown, recovery period | No verified values this time | Lacking a complete official adjusted NAV / total return daily series; annual declines cannot substitute for peak-to-trough drawdown |
| Volatility, Sharpe, tail loss | No verified values this time | Lacking same-frequency returns, risk-free rate, windows and trading-day rules |
| Future return range | The vendor's formula and assumptions can be audited | OLS ranges cannot be treated as official risk limits or a principal guarantee |
Because no reconstructable official daily series was obtained, the report does not fill in a guessed maximum drawdown for the 2000 bubble or 2022. Official holding concentration and the stated risks are the risk evidence available this time; high annualized performance is not used to offset the data gaps.
Vendor model and this study's audit
This chapter cites the shared quantitative results of this study and does not treat the old report as a new backtest. The source is the public monthly points of V1, with a current page date of 2026-10-29 and accessed 2026-10-30. The target return unit is annualized percent and the error unit is percentage points; whether the original includes dividends and expenses is still not fully verified. The full formulas, data SHA and rerun entry point are in the shared model audit (the author's model audit appendix, not published on this site).
In-sample: reconstructable, not deployable
Data: model_metrics.csv (data/NASDAQ100/model_metrics.csv); rounded below, not an additional rating.
| Model | Monthly points | Start-date range | In-sample R² |
|---|---|---|---|
| P/E to one year | 318 | 1999-04 to 2025-09 | 0.112168 |
| P/E to five years | 270 | 1999-04 to 2021-09 | 0.480193 |
| P/E to ten years | 210 | 1999-04 to 2016-09 | 0.663219 |
| P/E to twenty years | 90 | 1999-04 to 2006-09 | 0.361093 |
| Past one year to next one year | 306 | 2000-04 to 2025-09 | 0.001403 |
| Past five years to next five years | 210 | 2004-04 to 2021-09 | 0.189808 |
| Past ten years to next ten years | 90 | 2009-04 to 2016-09 | 0.027251 |
The 44 displayed fields passed in this study's shared check, which proves that the public point data are compatible with the displayed precision, not that the original earnings are correct. At long horizons the same return path is reused by adjacent months, so monthly point counts cannot be treated as independent investment experiences; R² is not a probability of hitting a target.
Strict maturity purging: pseudo-OOS results
Data: purged_oos.csv (data/NASDAQ100/purged_oos.csv). Each test point uses only matured labels satisfying training start month + forecast horizon < test start month, with an expanding training window; the benchmark is the mean of the training Y matured at that time. The minimum number of training points is this study's research design, not market data.
Minimum 60-point design; RMSE / MAE in annualized return percentage points:
| Model | Test n | Test start range | Model RMSE | Model MAE | Mean-benchmark RMSE | R²_OOS relative to benchmark |
|---|---|---|---|---|---|---|
| P/E to one year | 246 | 2005-04 to 2025-09 | 20.8975 | 16.7206 | 22.3862 | 0.12858 |
| P/E to five years | 150 | 2009-04 to 2021-09 | 9.8329 | 8.7983 | 14.8232 | 0.55998 |
| P/E to ten years | 30 | 2014-04 to 2016-09 | 7.7162 | 7.6086 | 14.2594 | 0.70718 |
| Past one year to next one year | 234 | 2006-04 to 2025-09 | 23.8422 | 20.1181 | 23.3704 | −0.04079 |
| Past five years to next five years | 90 | 2014-04 to 2021-09 | 6.0617 | 4.9440 | 8.1164 | 0.44222 |
Minimum 120-point sensitivity: P/E one-year / five-year are n=186/90, R²_OOS=0.19071/0.55743 respectively; past one-year / five-year are n=174/30, R²_OOS=−0.03840/0.35111 respectively. P/E ten-year has no testable points under this design; the P/E twenty-year and past-ten-year models have no testable points under either design, because there are not enough matured training labels. This means untestable, not a return of zero. Same CSV (data/NASDAQ100/purged_oos.csv)
Substantive reading:
- P/E at some horizons beat the matured-mean benchmark in this historical design; past one-year returns did not improve on the next year, so this signal should not be traded on its own.
- The ten-year model's 30 overlapping monthly results are not 30 independent ten-year market experiences; 0.70718 is not a 70.718% success rate, and it provides no guarantee of future returns.
- This is a current-vintage pseudo-OOS: it removes leakage from immature labels, but without historical point-in-time earnings, price revisions or constituent versions it cannot claim to have fully excluded all look-ahead / revision biases.
- No tradable signal, execution lag, turnover or transaction cost model was built, and no independent calibration of forecast-interval coverage was completed.
Extrapolation, intervals and combination diagnostics
- The past-ten-year-to-next-ten-year model's training x maximum is 13.3224% annualized, while the current input is 20.8925% annualized, so it is extrapolating; in-sample R² is only 0.027251. model_metrics.csv (data/NASDAQ100/model_metrics.csv), vendor current date 2026-10-29, accessed 2026-10-30, training start dates 2009-04 to 2016-09.
- The interval the website labels as "80%" uses a coefficient of 1.645. This is inconsistent with the usual central two-tailed 80% of the standard normal; it cannot be claimed as calibrated just by renaming it. Shared appendix (the author's model audit appendix, not published on this site) and V1; these are model coefficients / probability labels, not market data.
- This study's arithmetic reconstruction by inverse variance gives a combined ten-year output of about 17.7440% annualized, with the past-return model weighted at about 87.30%; this is reverse-engineered in this study, not a confirmation of the vendor's official method, and is not adopted as a financial-planning return.
- Same-start paired residual correlation coefficients at one / five / ten years are 0.94270/0.78808/0.55718 respectively, with paired n of 306/210/90; all are in-sample estimates on current data and cannot be taken as future error correlations. The common error is evident; combining while ignoring covariance may produce an overly narrow uncertainty. combination_diagnostics.csv (data/NASDAQ100/combination_diagnostics.csv)
All model numbers in this section are calculated from the vendor monthly data accessed in this study; data dates, bases and each model's training range are given in this chapter and the CSVs; the combined point estimate has not been repackaged as an institutional forecast.
Conditions under which a valuation judgment holds or fails
What can currently be supported: The relative position of the vendor's valuation within its self-defined historical distribution can serve as a descriptive reference; the official QQQ valuation and concentration can be used for an exposure inventory. Together they currently cannot establish a "definite margin of safety".
| Candidate judgment | Must be satisfied before it holds | Failure / withdrawal conditions |
|---|---|---|
| Valuation low relative to history | Same subject, same EPS period and treatment of negative earnings, same aggregation formula; reconstructable history and constituent changes | Change in the mean window or trimming method, EPS downgrades / vintage switches, index methodology changes making it incomparable |
| Valuation can improve long-run return estimates | Support from more truly point-in-time data and independent market periods; forecast errors and coverage can be calibrated | No longer beats the benchmark after the sample is expanded, extrapolation, unstable residual structure |
| Growth premium supported by fundamentals | Verifiable cash flow, margins, return on investment and EPS revisions support the price requirement | Earnings downgrades, capital spending failing to turn into cash flow, rising discount rates, or deteriorating fundamentals at concentrated companies |
| ETF suitable as an exposure tool | A clear holding period, tolerable drawdown, portfolio overlap and cost constraints | Single-name / common-factor concentration exceeding one's own risk budget, liquidity and trading frictions not meeting needs |
No arbitrary growth rate, valuation reversion endpoint, price target, scenario probability or promised return is specified. A falling P/E may come from a falling price, rising earnings or a change in data methodology; it must be decomposed and cannot always be called cheaper.
Monitoring indicators and decision limits
| Monitoring area | What to track | Data priority / limits |
|---|---|---|
| Valuation vintage | Dates and denominator definitions of the official P/E and Forward P/E, changes to vendor sources / algorithms | Align the basis before comparing; no unsupported buy points |
| Company fundamentals | Same-period reported EPS, cash flow, margins, capital spending and estimate revisions | Official filings / traceable consensus; historical series missing this time |
| Concentration | Same-day largest holdings, top ten securities and company exposure after merging share classes | Invesco daily holdings; do not just count the number of names |
| Sector and methodology | ICB sector weights, index reconstitution / special-adjustment notices, methodology version | Nasdaq and Invesco; PDF parsing and sector weights still to be filled in |
| Implementation costs | Current fee, effective date, NAV / market price differences and trading costs | Official fund disclosures; do not deduct fees again from historical NAV |
| Historical risk | Dividend-inclusive NAV daily series, peak-to-trough drawdown, recovery time, volatility and tail loss | Recalculate only after obtaining the full adjustment and reinvestment specification |
| Model stability | Expanding-window errors, mean benchmark, extrapolation, residual correlation, interval coverage | Keep maturity purging; upgrade the assessment only after adding truly point-in-time vintages |
Monitoring frequency can be set according to the official publication cycles and actual use; this report has no automated ordering, personal position advice, or authorization to deploy the model.
Source register and acceptance
Official and vendor sources
All access dates are 2026-10-30; detailed as-of, units and evidence status are saved in official_sources.json (data/NASDAQ100/official_sources.json).
- O1: Invesco QQQ product page: valuation, holdings, fund fees, performance and methodology-change notice actually read in a browser; the main data source.
- O2: Invesco QQQ performance page: standardized performance table, data date, fees and risk notes actually read in a browser; simplified web scraping is not enough to see the table.
- O3: Invesco QQQ overview / holdings page: verified the index scope, rebalancing frequency and ICB classification; the holdings chart on this page had no reliable readings, so the top ten were taken from O1's explicit table instead.
- O4: Official Nasdaq NDX Overview: verified the index scope, number of constituent securities and links to the official methodology documents; its price index readings are not mixed into NAV total returns.
- O5: Official QQQ factsheet: the scrape this time returned a country-splash image page, and no verifiable PDF content was read; numbers from search snippets are not used as evidence.
- O6: Current Nasdaq methodology PDF: the tool returned raw PDF content that was not parsed into readable provisions; cap thresholds and details are listed as gaps, and no packages were installed nor was the check falsely claimed as complete.
- V1: WorldPeRatio Nasdaq-100: vendor current values as of 2026-10-29 and public monthly data; details of the original market sources, earnings and returns remain limited.
Delivery and acceptance
| Acceptance item | Status | Evidence / limits |
|---|---|---|
| New version in the designated reports folder, Traditional Chinese, clickable table of contents | Done | This file; the original session report was not changed |
| Official numbers have URL, as-of, access date and unit | Done | Each table and the source JSON; the fee's missing effective date is explicitly kept as such |
| Official and vendor valuations directly comparable | Does not hold | Dates and denominators / aggregation not aligned; discount and valuation ranking dropped |
| Fundamentals and exposure | Partly done | Official portfolio characteristics, top ten and concentration done; sector weights and company fundamental series insufficient |
| Official NAV performance | Partly done | Official table verified; dividend reinvestment details not obtained, no false claim of item-by-item reconstruction |
| Historical drawdown / volatility / full methodology rules | Data gap | No official daily series and no readable full methodology provisions; no guessed values filled in |
| OLS / maturity-purged pseudo-OOS | Done to the disclosed extent | CSVs and shared appendix; point-in-time vintage, overlap and untestable-horizon limits kept |
| Truly real-time out-of-sample / tradable predictive ability | Not proven | Forecast validation is not declared complete just because the arithmetic was reproduced or the pseudo-OOS did better |
| Arbitrary price targets, growth rates, formal ratings | Not used | The old report's teaching scenarios are not copied as research forecasts |
Reading conclusion: What can be trusted is this study's data with timing and sources, and the clearly limited model diagnostics; what cannot be trusted is an "undervalued" claim without same-basis support, a narrow return range without calibration, or treating high historical returns as a promise for the next investment cycle.
Further reading
- The other part of this series: S&P 500 / SPY: official exposure, valuation basis and model evidence research
- QQQ returns by period and technicals (Chinese), CNDX returns by period and technicals (Chinese)
- VOO returns by period and technicals (Chinese), CSPX returns by period and technicals (Chinese)
- Custom comparison: QQQ vs VOO returns over the last 5 years (Chinese)
- Out-of-sample evidence on valuation-based timing signals
- Quant investing in US stocks for beginners: quantify the "market state" before quant trading
Frequently asked questions
What is QQQ's P/E ratio right now?
According to Invesco's official product page, QQQ's P/E is 29.48x and its forward P/E is 25.75x, both with a data date of 2026-09-30; the 27.10x from vendor WorldPeRatio is a vendor value as of 2026-10-29. The original stresses that the dates, earnings denominators and aggregation methods of the two are not yet fully aligned, so the difference cannot be interpreted as an arbitrage or as a degree of undervaluation.
Is the Nasdaq-100 undervalued right now?
This research does not conclude that it is "currently undervalued". The original's conclusion is to withhold a valuation rating and give no buy/sell or overweight rating: what can be confirmed is an exposure to large non-financial, growth and large-cap tech-related names, but it cannot be confirmed that the vendor's "Undervalued" label represents a fundamental discount; official sector weights, the full definition of valuation aggregation, a reconstructable dividend-inclusive daily series and the maximum drawdown are still gaps.
How much of QQQ do the top ten holdings make up?
Based on the official displayed holding weights as of 2026-10-28, this study's sum puts the top ten securities at 46.71% of QQQ's net assets and the top three at 21.89%, with the two Alphabet share classes combined at 5.70%; the largest are NVIDIA 8.03%, Apple 7.76% and Microsoft 6.10%. This is a sum of already-rounded displayed weights, not a concentration index separately published by the issuer, and the top ten include two share classes of Alphabet, so they are not ten independent companies.
What is QQQ's ten-year annualized return, and what is its expense ratio?
The official ten-year annualized performance of QQQ NAV is 20.91%, as of 2026-09-30; over the same period the officially listed Nasdaq-100 Index returned 21.16%. As of the access date 2026-10-30, the total expense ratio listed on the official page was 0.18% / year, with no separately stated effective date for the fee. The original cautions that high historical returns cannot substitute for future required returns, and that fees should not be deducted a second time from published NAV performance.
Can the P/E ratio predict future Nasdaq-100 returns?
This study ran a strictly maturity-purged pseudo-OOS on the vendor's monthly data: the R²_OOS relative to the benchmark is 0.55998 for P/E to five years and 0.70718 for P/E to ten years, but the ten-year case has only 30 overlapping monthly results, which are not 30 independent ten-year market experiences, and 0.70718 is not a 70.718% success rate; past one year to next one year is −0.04079, no improvement. The original states clearly that this is a current-vintage pseudo-OOS, not a tradable, truly real-time out-of-sample validation.
Can the "80%" return range on the website be trusted?
The original points out that the interval the website labels as "80%" uses a coefficient of 1.645, which is inconsistent with the usual central two-tailed 80% of the standard normal, and it cannot be claimed as calibrated just by renaming it; same-start paired residual correlation coefficients at one / five / ten years are 0.94270/0.78808/0.55718 respectively, the common error is evident, and combining while ignoring covariance may produce an overly narrow uncertainty.