← All experiments
EXPERIMENT 03 · 8.5-YEAR BACKTEST

What if you only bought QQQ's top 20 holdings?

QQQ holds ~100 stocks, but a handful of mega-caps do all the work. We skipped the ~80 laggards, owned only the top 20, rebalanced quarterly — and ran $10,000 through 8.5 years of real data.

QQQ buy & hold
$45,083
$10,000 → 19.5%/yr · worst drop −35.1%
Top-20 holdings
$59,740
$10,000 → 23.6%/yr · worst drop −38.9%

The setup

The 8.5-year result

The top 20 turned $10,000 into $59,740. Buy-and-hold QQQ turned it into $45,083 — a $14,657 gap, about 4 points of annual return per year.

The mechanism is concentration, not stock-picking genius: the top 20 average 67% of QQQ's weight — two-thirds of the index. In a regime where a handful of mega-caps did almost all the work, owning more of the winners and none of the ~80 laggards was the right structural bet. We also ran an equal-weighted top-20 as a sanity check: $57,270 — it still beat QQQ, so this isn't a weighting trick. It's the tail that lost.

Year by year

The top 20 won 6 of 9 years. QQQ won 3 — and notably, one of them was 2022.

YearQQQ buy & holdTop-20 holdingsWinner
2018*−7.4%−5.2%Top-20
2019+38.4%+43.1%Top-20
2020+46.0%+52.4%Top-20
2021+29.2%+35.3%Top-20
2022−33.2%−37.2%QQQ
2023+55.9%+72.0%Top-20
2024+27.7%+42.6%Top-20
2025+21.0%+21.0%QQQ
2026*+17.2%+15.5%QQQ

*2018 measured from March 20; 2026 through August 31.

The price paid

You were paid about 4 points of annual return for taking genuinely more pain. There's no version of this trade where the extra return is free.

The honest caveats

The verdict

The top-20 won, and it wasn't close: $59,740 vs $45,083. The mechanism is concentration, not stock-picking genius — the top 20 are two-thirds of QQQ, and in a regime where a handful of mega-caps did almost all the work, owning more of the winners and none of the ~80 laggards was the right structural bet. The price was real: the concentrated book fell harder (−38.9% max drawdown vs −35.1%), so this is higher return for genuinely higher pain, not a free lunch.

Methodology. 34 quarterly holdings snapshots (SEC filings, Invesco archives, archived index data via Wayback Machine), each dated on or before its rebalance date; rebalance = first trading day after the third Friday of Mar/Jun/Sep/Dec (one holiday shift: 2018-12-25 → 12-26). Yahoo Finance dividend/split-adjusted daily closes, 2018-03-20 → 2026-08-31. Index weights renormalized to 100%; portfolios drift between rebalances. Cash earns 0%. No fees, no taxes, no spreads. Re-run it yourself — the rules above are the entire recipe.
📥 FREE: The 10 What-Ifs Every Investor AsksOne page · ten questions · real-data answers 📸 See this experiment as a carousel + reel@market.whatif on Instagram