Dow Jones Industrial Average
S&P Dow Jones Indices; tracked by the SPDR Dow Jones ETF (DIA, State Street) — 30 price-weighted blue chips
The classic 30-stock blue-chip benchmark, price-weighted (so high-priced names dominate). Now carries the AI mega-caps after Nvidia joined in Nov 2024 alongside Microsoft and Apple — a more value/industrial-tilted read on the market than the Nasdaq-100.
- Reconstructed — Figures are third-party estimates rebuilt from filings, not reported results.
Top holdings
~top weight — Highest price-weighted Dow component
top weight — Large price-weighted health insurer
high weight — AI / Azure mega-cap in the Dow
high weight — Retail blue chip
high weight — Industrial / data-center buildout proxy
mid weight — Added to the Dow Nov 2024 — the AI compute leader, low Dow weight due to price-weighting
mid weight — Payments blue chip
mid weight — Consumer-AI mega-cap
Performance
July 2026: DJIA +0.3% price, 4th straight monthly gain; DIA ~+0.4% TR (derived, see unverified) — July 2026 was the Dow's best relative month of the year purely because of the structure this entry critiques: DJIA +0.3% vs Nasdaq -3.2% and SOX -20.6%, the only major index up. The price-weighted AI-hardware underweight (no memory, no WFE, no neoclouds; NVDA ~2.3%) was the reason. Provider figures differ only by as-of date (SSGA quarter-end 30 Jun vs providers 31 Jul), not methodology.
Recent moves
Nvidia replaced Intel in the Dow (Nov 2024); the index is more AI-exposed than its industrial reputation suggests but stays value/blue-chip tilted vs the Nasdaq-100.
Our take
A blue-chip baseline rather than an AI vehicle — price-weighting means even Nvidia is a small slice, so the Dow understates the AI trade vs cap-weighted S&P 500 / Nasdaq-100. Useful as the conservative end of the index spectrum.
Two things the marketing won't foreground.
First, price-weighting makes the portfolio an artifact of share-price convention, not economics: the top two weights (GS 12.10%, CAT 9.91%) are far smaller by market cap than names weighted a quarter as much, and any split silently re-sizes exposure. Second, the June 29 2026 Alphabet-for-Verizon swap is the methodology partly indicting itself. S&P's affirmative case was Alphabet's scale, share price and business breadth — AI, cloud, healthcare tech, digital advertising. But it also stated 'Verizon represents only one-half of one percentage point of the DJIA due to its lower share price': a large, liquid telecom became too small to keep on arithmetic, not fundamentals. Add discretionary selection (30 names, no published quantitative screen, WSJ editors on the panel) and 'passive index' is the wrong model — this is an actively curated 30-stock portfolio at index economics. The November 2024 turnover, Nvidia for Intel and Sherwin-Williams for Dow Inc., reads as momentum-chasing after the fact.
Thesis
The Dow is not a blue-chip strategy — it is a bet on share-price levels. Goldman Sachs (12.10%) and Caterpillar (9.91%) are ~22% of DIA, while Nvidia, among the world's most valuable companies, sits near 2.1–2.3%. Nothing in that reflects a view on those businesses; it reflects that GS trades above $1,000 and NVDA does not. A 4-for-1 Nvidia split would cut its Dow weight ~75% with zero economic event. Buyers think they own 30 quality large-caps; they own a committee-picked list sorted by share price, at 0.16% (Jul 2026).
Price-weighted index of 30 US large-caps, tracked by DIA (State Street, inception Jan 14 1998; $45.0B net assets as of Jul 20 2026). Weight = share price ÷ sum of member prices, scaled by the Dow Divisor. Constituents are set by a discretionary committee (S&P Dow Jones Indices + Wall Street Journal editors) on reputation and market relevance — not a rules-based screen. The committee also monitors that the highest-priced member stays under roughly 10x the lowest.
Assessment
- Genuine long history and name recognition — the divisor mechanism has kept a continuous, comparable series through a century of splits and substitutions.
- AI exposure stays underweight versus cap-weighted peers by construction — Alphabet joined only in June 2026 (~4.0%), Microsoft ~4.6% — leaving real industrial and financial weight.
- Committee discretion permits qualitative judgment a mechanical screen can't make — sector-representation balance is an explicit selection input.
- Deep liquidity and a mature options/futures ecosystem around DIA make it a low-friction expression of large-cap US beta.
- Price-weighting has no economic justification. It is an 1896 hand-computation shortcut that survived on inertia; nobody designing an index today would choose it.
- A stock split changes your exposure with zero underlying event. Nvidia splitting 4-for-1 would cut its Dow weight ~75% overnight — a real risk given persistent 2026 split speculation.
- 0.16% for 30 committee-picked stocks is roughly 5x a cap-weighted S&P 500 tracker, for a narrower, structurally odd portfolio. The fee is not defended by the methodology.
- Only 30 names, committee-selected with no published quantitative rule — single-name and single-committee-decision risk a 500-stock rules-based index does not carry.
- DIA's unit-investment-trust structure cannot reinvest dividends between distributions or lend securities, creating a structural cash drag versus modern open-end ETFs.
Record
DIA NAV annualized returns as of June 30 2026: 1Y +20.28%, 3Y +16.85%, 5Y +10.57%, 10Y +13.49%. Attribution is unambiguous: this is US large-cap beta minus 0.16%, with no manager skill or leverage to attribute. The interesting spread is 5Y (+10.57%) versus 10Y (+13.49%) — the 5Y window contains the 2022 drawdown and the period in which cap-weighted indices were carried by AI mega-caps the Dow structurally underweighted. That gap is the price-weighting drag showing up in the record, not a stock-picking failure. The strong 1Y and 3Y numbers coincide with financials and industrials — precisely the high-share-price names the methodology overweights — performing well, which is luck of the arithmetic, not design.
Risks & fit
- Concentration by accident: ~22% in two names (GS, CAT) chosen by share price, giving financials and industrials outsized mechanical influence over the index.
- Split risk is exposure risk — a constituent's corporate action can materially re-weight the portfolio with no change in fundamentals and no rebalancing signal to holders.
- Committee/key-decision risk: 30 slots filled by discretionary judgment, with turnover that has historically added names after their run rather than before.
- Persistent structural underexposure to the largest-market-cap companies, so the index lags cap-weighted benchmarks in mega-cap-led regimes.
- Fee drag compounds: ~13bp/yr above the cheapest large-cap alternatives, plus UIT cash drag, on an otherwise commoditized exposure.
The critique fails if price-weighting is a durable, unintentional value/anti-momentum tilt rather than noise. Concretely: if over a full market cycle DIA's risk-adjusted return matches or beats a cap-weighted S&P 500 tracker net of its ~13bp fee gap, the 'arbitrary weighting' argument loses force — the arbitrariness would be delivering real diversification. Two others: if S&P Dow Jones Indices moves the Averages to a cap- or equal-weighted basis, the structural objection retires; and if committee changes demonstrably lead rather than lag the market-cap ranking, the momentum-chasing charge is wrong.
Analytically, DIA is a coherent expression of one narrow thing: exposure to 30 large, established, mostly non-technology US companies with an accidental tilt toward high-share-price financials and industrials. It is not a coherent expression of 'the US market', 'blue-chip quality', or 'AI exposure' — a broad cap-weighted index expresses the first directly and DIA only obliquely; the second has no methodological support; the third is structurally suppressed.
0.16% gross expense ratio (SSGA, Jul 2026) — roughly 5x the cheapest cap-weighted S&P 500 trackers (~0.03%). No performance fee. The UIT structure adds unquantified cash drag from dividends not reinvested between distributions.