Roundhill Magnificent Seven ETF
Roundhill Investments — Investment Committee (active)
Equal-weight exposure to the 'Magnificent Seven' mega-cap tech names in a single ticker, rebalanced to ~14.3% each quarterly. Actively managed structurally but low-turnover. Low fee for an active fund: 0.29% expense ratio. NOT leveraged (the 2x version is the separate MAGX ticker).
- Concentrated — Top disclosed positions are 50%+ of the disclosed book.
Top holdings
Equal-weight target; the dominant AI-compute name
Equal-weight target; AI/cloud (Azure + OpenAI)
Equal-weight target
Equal-weight target; Gemini/AI + search
Equal-weight target; AWS + commerce
Equal-weight target; Llama/AI + ads
Equal-weight target; the highest-volatility constituent, drifts between rebalances
Recent moves
Pure quarterly equal-weight rebalance back to ~14.3% each — no discretionary stock-picking; weights drift between rebalances with Nvidia/Tesla swinging most.
Our take
The single cleanest, cheapest way to own the Mag 7 as one position, and equal-weight meaningfully dampens single-name (Nvidia) concentration vs. a cap-weighted Nasdaq fund. The most 'investable' name on this list for a core AI/mega-cap allocation — far less speculative than the ARK suite. (Don't confuse with the leveraged MAGX.)
The label oversells the benefit. Equal-weighting seven names that all rise and fall on the same AI-capex and mega-cap-tech factor cuts idiosyncratic (NVDA-blowup) risk but does little for the systematic risk behind real drawdowns — in a selloff the seven correlate toward 1.
Mechanically it is a mild anti-momentum bet: each quarter it trims whatever led and tops up the laggards, so it structurally lags both a cap-weighted book and simply owning the winner during a single-name leadership run, and only earns its keep when intra-group leadership rotates. Two things marketing underplays: (1) it is genuinely 1x unleveraged — not the 2x daily-reset MAGX, which is where volatility decay lives; and (2) as of mid-2026 the exposure is delivered synthetically via total-return swaps backed by a ~51% Treasury collateral sleeve, so you carry counterparty risk and hold no actual shares — not the direct stock basket the name implies.
Thesis
MAGS packages all seven 'Magnificent Seven' mega-caps into one ticker, rebalanced quarterly to ~14.3% each, at a 0.30% fee. The pitch is that equal weight tames the single-name (NVIDIA) risk of a cap-weighted Nasdaq fund. The reality is it remains a 7-name, single-country, single-factor bet where every holding shares the same AI-capex / mega-cap-tech driver — so equal weight trims idiosyncratic risk but barely touches the systematic drawdown risk that actually decides outcomes.
Actively managed in form but mechanical in practice — no stock selection, just a quarterly reset to equal weight across AAPL, MSFT, NVDA, GOOGL, AMZN, META, TSLA. As of mid-2026 it delivers that exposure largely through total-return swaps (Goldman Sachs is a named counterparty in the holdings) backed by a ~51% Treasury / ultra-short collateral sleeve, so a screen shows ~23 line-items, not 7 shares. Unleveraged 1x — distinct from the separate 2x MAGX ticker.
Assessment
- One-ticker equal-weight Mag 7 at 0.30% — cheap for an active-branded fund, and it removes the discipline problem of rebalancing seven names by hand
- Quarterly trim of the crowded leader (NVDA back to ~14.3%) lowers single-name blow-up risk vs a cap-weighted Nasdaq book
- Genuinely unleveraged 1x — not to be confused with the 2x daily-reset MAGX sibling that carries volatility decay
- Low turnover and a quarterly cadence keep trading and tax drag modest for an active wrapper
- 'Equal weight' is sold as diversification, but the seven are near-single-factor (US mega-cap AI/tech); correlations spike in a selloff, so it dampens single-name not systematic risk
- Anti-momentum drag: quarterly trimming of the leader lags cap-weight or direct-NVDA in a persistent single-name leadership tape
- Synthetic swap implementation (mid-2026) adds counterparty and rollover risk and means no direct share ownership — a detail marketing does not lead with
- 0.30% buys convenience replicable for ~$0 by holding seven stocks at a zero-commission broker
- All seven are richly valued at once — no valuation dispersion within the basket to rotate into
Record
+18.75% trailing 1-year (incl. dividends, as of Jul 21 2026) and +36.61% average annual since the Apr 11 2023 inception — driven overwhelmingly by the 2023-26 mega-cap / AI rally, not by any stock selection. The equal-weight construction meaningfully lagged a pure-NVIDIA or cap-weighted book during NVIDIA's dominance: trimming the leader each quarter is a headwind when one name leads for quarters and a tailwind only when leadership rotates. No clean 3-year annualized figure yet — the fund is just over three years old and issuer pages still show 'As of TBD' placeholders.
Risks & fit
- Single-factor concentration — a mega-cap tech / AI de-rate hits all seven names together
- Swap counterparty and rollover risk from the synthetic implementation
- Anti-momentum construction underperforms if one name keeps leading
- Rich valuations across all seven holdings simultaneously
- Antitrust / regulatory overhang on GOOGL, AMZN, META and AAPL
If the seven names stay tightly correlated through the next real drawdown — all falling together on an AI-capex or rate shock — then equal weight's 'diversification' delivered little the reader couldn't get from any cap-weighted tech fund, and the premise is cosmetic. Conversely, if intra-group leadership rotates sharply (laggards like Apple or Tesla outrunning NVIDIA for several quarters), the equal-weight rebalance earns its keep and the anti-momentum critique is what's wrong.
Suits an investor who wants one-ticker, equal-weight Mag 7 exposure with hands-off quarterly rebalancing and accepts high single-theme (US mega-cap tech / AI) concentration as the whole point — not a diversifier despite the label. Those wanting leverage look to MAGX (2x, daily reset — carrying the volatility decay MAGS does not); those wanting broader mega-cap look to QQQ or XLG.
0.30% gross — cheap for an actively-branded fund and well below thematic-AI peers (0.68-0.88%), but a self-built seven-stock basket costs ~$0 at a zero-commission broker; the fee buys the wrapper, automatic rebalancing and the swap plumbing.