Home › Articles

Research articles

Research notes and methodology essays on US stocks and ETFs: how market signals are read, when they fail, and what a backtest can and cannot prove. For information and education only; not investment advice.

US index institutional research

Two traceable research notes on Nasdaq-100/QQQ and S&P 500/SPY: official valuation and concentration, official versus vendor definitions, a pseudo-out-of-sample check of vendor models, and data gaps.

  1. 2026-10-30
    Nasdaq-100 / QQQ: Official Evidence and Vendor Model Research
    Invesco lists QQQ at 29.48x P/E and 25.75x forward P/E (2026-09-30) vs a 27.10x vendor value; bases differ, so no undervaluation claim. Top ten holdings 46.71%; ten-year NAV 20.91%. Research note, not investment advice.
  2. 2026-10-30
    S&P 500 / SPY: Official Exposure, Valuation Definitions and Empirical Model Research
    SPY FY1 P/E 20.48x, index P/E 24.48x and WorldPeRatio 24.06x use different definitions; top ten 38.03%; ten-year NAV 15.18%. The site's -1.31% ten-year output is not validated. Research note, not investment advice.

VOO/QQQ market-timing evidence

Valuation, trend, sentiment, forecast combination and machine learning, and turning points: how much of each timing method survives out-of-sample tests and real-world costs.

  1. 2026-10-26
    Out-of-sample (OOS) evidence on valuation-based timing signals (S&P 500 / US large caps; Nasdaq covered where possible)
    Valuation timing for VOO/QQQ: OOS R² is almost all negative in Goyal-Welch 2008; a CAPE rule's 1900–2015 Sharpe was 0.37 vs 0.38 for buy-and-hold, before costs. Research note, not investment advice.
  2. 2026-10-26
    Trend-Following and Time-Series Momentum (TSMOM) Timing: Out-of-Sample (OOS) Evidence for the S&P 500 and Nasdaq
    Can a 10-month SMA or TSMOM time VOO/QQQ? Real-time OOS: Sharpe improves only 7–12% after costs, returns trail buy-and-hold, about 80% of sell periods lose to T-bills. Research note, not investment advice.
  3. 2026-10-26
    Positioning/Sentiment and Option-Implied Risk-Pricing Timing Signals: Out-of-Sample Evidence
    In GWZ 2024, vrp, rsvix and sntm are dismissed; short interest alone survives (monthly OOSCT R² 1.26%) but adds only 0.10% CEV over 100% equities. Research note, not investment advice.
  4. 2026-10-26
    Forecast Combination and Machine-Learning Market Timing: Out-of-Sample and Replication Evidence on the US Equity Premium
    GKX 2020 neural nets lift S&P 500 timing Sharpe from 0.51 to 0.77 before costs; RSZ combination gains fall to −0.14% and 0.11% vs buy-and-hold. Research note, not investment advice.
  5. 2026-10-26
    Turning-Point Prediction, What Is Special About the Nasdaq-100/QQQ, and the Realities of Retail Implementation
    No method calls tops in real time: after a 100% industry run-up crash odds rise, yet prices keep rising about 6 months; Nasdaq investors lost 5.3 percentage points a year to timing. Research note, not investment advice.

Why Taiwan, US and UK ETFs show different returns

Why ETFs tracking the same index look different in Taipei, New York and London: the main article explains the causes, and three data notes carry the exchange rates, prices, NAVs and distributions.

  1. 2026-10-22
    Currency, Base Dates and the Wrong Ticker Create the Illusion That Taiwan-Listed ETFs Lead
    On Google Finance, Taiwan-listed 00646 and 00662 appear to beat VOO and QQQ because the New Taiwan dollar fell 5.86% in a year. In US dollars, 00662 returned +18.82% and 00646 +12.75% YTD, within 0.5 points of QQQ and VOO. Not investment advice.
  2. 2026-10-20
    USD/TWD Exchange Rates and the US-Dollar Returns of Taiwan-Listed Overseas ETFs (00646 / 00662)
    The New Taiwan dollar fell about 5.53% in a year and about 3.10% YTD. In US dollars, 00646 returned +12.74% YTD (VOO +12.35%) and 00662 +18.82% over one year (QQQ +17.47%, +18.95% once dates are aligned). Research note, not investment advice.
  3. 2026-10-20
    NASDAQ-100 Three-ETF Market Data Check (QQQ / 00662 / CNDX)
    The "CNDX +18.90%" on Google Finance is a Toronto-listed TSX 60 ETF. The real LSE CNDX returned +17.63% over one year and QQQ +17.47%; 00662's +25.78% comes mainly from a weaker NT dollar and a different base date. Research note, not investment advice.
  4. 2026-10-20
    2026 YTD Market Data for Three S&P 500 ETFs (VOO / 00646 / CSPX)
    Google uses 2025-12-31 as VOO's YTD base but 2026-01-02 for 00646 and CSPX. On a common base, VOO is +12.35%, CSPX +12.87% and 00646 +16.35% in NT dollars; with dividends, VOO +13.33% vs CSPX NAV +12.71%. Research note, not investment advice.

RSI oversold crossover backtest

Full-history backtest of the RSI6-below-20 then RSI6-crosses-RSI10 signal on QQQ and VOO: main article, full data appendix and literature review.

  1. 2026-10-19
    RSI Oversold Crossovers Buy Low, but Rarely at the Bottom
    Full-history backtest of RSI6<20 then RSI6 crossing above RSI10: only 18.4% of QQQ and 38.3% of VOO signals were within 3% of the low; QQQ signals more than 20% below the 52-week high won 20%.
  2. 2026-10-19
    QQQ / VOO "RSI oversold + RSI6/RSI10 golden cross": full-history backtest
    Full data for 77 QQQ and 47 VOO signals. QQQ 6M mean +4.4%, win 69.7% vs. random +6.5%, 72.9%; VOO +8.5%, 85.1% vs. +6.3%, 80.2%, not significant. Research note, not investment advice.
  3. 2026-10-19
    Short-Period RSI Oversold Signals and RSI Golden Crosses in US Index ETFs (QQQ/VOO/SPY): Literature and Empirical Background
    Published RSI(2) oversold backtests on SPY/QQQ win about 70%-80% of trades yet lag buy-and-hold (Backtrex CAGR 1.7% vs 13.5%), plus RSI crosses, momentum and Wilder vs Cutler RSI. Research note, not investment advice.

Standalone articles

  1. 2026-10-24
    Leveraged ETF Return Convergence and the Market-Support Hypothesis: A Test Report
    Does SSO/QLD-VOO/QQQ spread convergence mark a rebound? 2006-2026 tests: only 5 of 24 troughs (20.8%) were in the band; R² 0.9975~0.9989 vs return, volatility, cost. Research note, not investment advice.
  2. 2026-10-21
    Returns Come From Beta, Not Distributions: A Strategy Playbook for XQQI, QQQI and TMGN
    QQQI returned 19.91% a year vs 23.63% for the Nasdaq-100; XQQI's 1.5x exposure made +16.26% vs +18.67%. Returns come from beta, not the 14–21% distributions. Not investment advice.
  3. 2026-10-17
    US-Domiciled ETFs or Irish UCITS? VOO vs CSPX and QQQM vs CNDX on Tax and US Estate Tax
    For non-US investors over 30 years: on cost alone CSPX beats VOO but QQQM beats CNDX; once US estate tax is included (above USD 60,000, up to about 40%), UCITS wins for both indices.
  4. 2026-10-16
    Quantitative Investing in US Stocks: Before Quant Trading, Quantify the Market Regime
    A quant framework for the US stock market: Market Score (0–100) from six weighted components, code-enforced Kill-Switch thresholds, credit > sentiment > valuation, and three backtest traps.
  5. 2026-10-16
    Signal Combinations Can Bracket a Bottom, but Cannot Pinpoint the Low
    No single signal catches the VOO/QQQ low: setup signals arrive 3 to 7 months early, confirmations lag 8 to 15 trading days or months. A three-type, staged-entry framework.