ETFs are the correct answer for maybe 70% of investors. For the other 30%, a custom index — direct ownership of the underlying stocks, under a rule you wrote — is quietly better. Here’s the side-by-side on fees, tax, control, and complexity, plus the five specific scenarios where custom indexing genuinely beats buying SPY.
TL;DR
- ETFs are cheaper and simpler; custom indices are more tax-efficient and more configurable.
- Tax-loss harvesting at the stock level is worth 30–100 bps of after-tax return per year in taxable accounts above ~$50k.
- Exclusions, tilts, and thematic precision are impossible inside an ETF wrapper. A custom index handles them natively.
- Under ~$10k, or in tax-sheltered accounts (ISA / 401k / SIPP), an ETF usually wins on simplicity alone.
The structural difference in one line
An ETF is a wrapper around a basket. A custom index is the basket. You hold every underlying stock in your own account under your own cost basis.
| ETF | Custom index | |
|---|---|---|
| What you own | Shares of the fund | Each stock, directly |
| Cost basis | One, at fund level | One per holding |
| Rebalance | Fund manager's schedule | Your schedule |
| Overrides | Impossible | Per-stock, any time |
| Transparency | Holdings disclosed periodically | Continuous — you wrote the rule |
Fees: not actually close
ETF fees famously kept falling for twenty years — 3 bps on VOO, 10 bps on VTI. The sticker price looks unbeatable.
Custom indexing has no fund wrapper, so it has no management fee. The cost is the platform charge plus the implicit bid-ask on rebalance trades. On Arithmos that nets out at 0–30 bps/year depending on the plan.
For most readers this is a wash. The real gap shows up on the next axis.
Tax: the silent 50–150 bps/year
When you own an ETF, a losing position inside the fund doesn’t help you at tax time. You pay tax on the fund’s distributions and on your eventual sale of the fund shares.
When you own a custom index, each stock is its own tax lot. When any single position falls, you can harvest the loss by selling it and buying a statistically similar name — staying in the market while banking the loss against gains elsewhere in your portfolio.
This is the reason the original direct-indexing shops (Parametric, Aperio) exist. The wealth-advisor world has known about it for three decades; retail only recently got the infrastructure to do it.
Control: exclusions and tilts
The second real advantage is configurability. Three examples you cannot express in an ETF:
- Concentration offsetting.You work at a biotech. You already own too much biotech via your equity package. You want “SPX-500, but ~0% healthcare.” An ETF can’t do this — you’d have to sell SPY and buy four sector ETFs to approximate it. A custom index can.
- Values-based exclusions.Exclude tobacco, thermal coal, any company flagged by your ethical screen of choice. ESG ETFs get close; they don’t let you define the screen.
- Thematic precision.“Profitable European defence primes and their top-tier suppliers” is a real view. No ETF tracks it. A rule can encode it.
Liquidity and complexity
ETFs win decisively here. One ticker, one click, deep intraday liquidity.
A custom index at 40 holdings needs 40 trades to rebalance (handled as a basket order on any modern broker — one allocation, broker slices it). Complexity’s lower than it sounds, but it’s still higher than pressing the “buy VOO” button.
Five scenarios where custom wins cleanly
- Taxable account above ~$50k. Tax-loss harvesting alone pays for the product several times over.
- You work in a concentrated sector. Build the rest of your portfolio with that sector deliberately underweighted.
- You have a specific thesis an ETF doesn’t track.“US onshoring beneficiaries in mid-cap industrials” isn’t on any product shelf; it’s a rule you can write.
- You want real values-based exclusions. Specific, not the vague “ESG-aware” blend.
- You want to understand what you own. An ETF prospectus is dense. A rule is one paragraph you can re-read whenever you’re unsure.
When an ETF is still the right answer
- Small account size (under ~$10k). The tax-alpha case doesn’t clear the complexity hurdle.
- Tax-sheltered accounts (ISA, 401k, SIPP, Roth IRA). No tax-alpha to harvest.
- You actively want set-and-forget. One ticker is beautifully simple.
- You’re happy with the exact exposure a broad-market ETF already gives you — no exclusions, no tilts.
The hybrid play (most people)
You don’t have to pick one. The common pattern is:
- ETFs in tax-sheltered accounts — because tax-alpha is irrelevant here, pay for simplicity.
- A custom index in taxable brokerage — capture the tax-alpha and bake in whatever tilts and exclusions you want.
- Bonds / cash / illiquid stuff elsewhere — neither vehicle is the right tool.
Build a custom index from a sentence →
FAQ
Does a custom index always beat an ETF?
No — that’s not the claim. Identical underlying holdings produce identical pre-tax returns. The edge is post-tax: loss harvesting at the stock level, plus the ability to tilt away from exposures you already have elsewhere.
How many stocks does a custom index need?
20–50 is the practical sweet spot. Below 20 you’re stock-picking with extra steps; above 100 you’ve just rebuilt the benchmark at higher cost.
What about tracking error?
A direct-indexed S&P 500 tracks the benchmark within ~0.5% a year in practice (Parametric’s published numbers). The tax alpha dwarfs the drag.
What’s the minimum viable account?
Around £1,000 / $1,000 with fractional shares. Tax-loss harvesting only starts meaningfully paying for itself around $50k in a taxable account.