Robo-Advisors and Model Portfolios

What is actually inside one

Strip away the interface and every mainstream robo is the same four components:

1. A risk questionnaire mapping you to one of roughly five to ten model portfolios. Typically a handful of questions about horizon, income stability and reaction to losses.
2. A model portfolio of index ETFs, usually broad equity plus bonds, occasionally with tilts (value, small-cap, gold, REITs) that the provider believes in.
3. Automation: recurring contributions, fractional investment, and rebalancing on a schedule or bands.
4. Extras, market-dependent: tax-loss harvesting (US, where it is genuinely valuable), goal projections, and human advisers at higher tiers.

The portfolios themselves are rarely the differentiator. Compare five robos' 60/40 models and you find the same building blocks in slightly different proportions: which is a compliment to the industry's honesty and a warning about paying for "proprietary allocation."

The cost arithmetic, stated plainly

Total cost is the robo fee plus the underlying funds. A typical structure: 0.25% platform fee plus 0.15% average fund TER, so around 0.40% all-in, against roughly 0.15% for the same portfolio held directly.

The 0.25% gap, compounded on €100,000 over 25 years at 7%:

• DIY at 0.15%: about €522,000
• Robo at 0.40%: about €492,000
• Difference: roughly €30,000

That is the honest price tag, and any robo assessment that skips this number is marketing. But the comparison is only valid against a DIY portfolio you would genuinely have built and maintained: which brings us to the number on the other side of the ledger.

The counterfactual that decides it

Chapter 64 established the behaviour gap at roughly 1–1.5% a year on average. If a robo prevents even a third of that for you, it has paid its 0.25% fee several times over. So the real question is not "is 0.40% expensive?" but "what would I actually do without it?"

Three honest self-assessments:

• If your alternative is a savings plan into one multi-asset fund, held for decades: the robo is a worse deal. You already have automation and internal rebalancing for 0.10–0.25% all-in (Chapter 27), with no second layer of fees.
• If your alternative is a two-to-three-fund portfolio you rebalance annually per Chapter 63: the robo is probably a worse deal too, costing ~0.25% for a twenty-minute annual task you have already demonstrated you can do.
• If your alternative is cash in a savings account, an unmanaged pile of past enthusiasms, or a portfolio you sold in the last crash: the robo is very likely the best financial decision available to you, and the fee is trivially worth it.

The mistake this chapter wants to prevent is the sophisticated reader dismissing robos on fee grounds and then not building the portfolio either. Being right about the arithmetic while owning nothing is the worst outcome in the comparison.

Where robos genuinely add value

• Tax-loss harvesting, in the US. Automated harvesting across many positions can add real after-tax value in taxable accounts, plausibly covering the fee in volatile years, though estimates vary and the benefit shrinks as unrealised losses are exhausted (Chapter 52's mechanics). Outside the US, this feature is usually absent or far less valuable.
• Genuine friction removal. Fractional investing, automatic allocation across accounts, no order tickets. For someone who finds Chapter 15 tedious rather than interesting, this is the whole point.
• The crash intermediary. Some robos add human contact or nudges during drawdowns. Given Chapter 64's evidence, an intervention that stops one panic sale in thirty years may be worth more than every fee ever paid.
• Sensible defaults for people who would otherwise take advice from worse sources. The realistic alternative to a robo is often a bank product at 1.5% with an entry load (Chapter 3), not a Boglehead portfolio.

Where they disappoint

• Fee stacking at scale. A percentage fee that is trivial at €10,000 is €1,500 a year at €300,000, for work that does not get harder. Robos with flat-fee pricing are structurally better for large balances; check the crossover point.
• Model complexity as differentiation. Some providers add eight to twelve funds, tilts and thematic sleeves that mostly serve to look like expertise (Chapter 27's warning about fund count applies, plus higher internal costs).
• Questionnaire risk assessment is crude. Five questions cannot establish what Chapter 26 says matters: your euro-denominated worst case and whether you would hold through it. Many people are placed a rung too aggressive because the questionnaire asked about attitudes rather than showing them numbers.
• Cash drag and product choice. Some allocate meaningful cash at low internal rates, or use in-house funds. Both are worth checking.
• Switching costs. Leaving a robo means selling a multi-fund portfolio, which is a taxable event in a taxable account. Easy to enter, less easy to leave.

Model portfolios: the same idea, unbundled

Adjacent to robos are model portfolios: published allocations you implement yourself, offered by providers, publications and advisers. They give you the allocation without the automation or the fee. Two uses, one caution.

Useful as a sanity check against your own build, and as a starting template for someone who wants Chapter 27's ladder pre-specified. The caution: a model portfolio is a recommendation with no accountability and no knowledge of you, and the more elaborate it is, the more likely it is expressing the publisher's views rather than the evidence. Judge one exactly as you would judge yourself: fund count, total cost, and whether every position has a reason you could write down (Chapter 61's gates).

A worked example

Two savers, both €60,000, both 70/30.

• Nils already runs a €500 savings plan into an 80/20 multi-asset UCITS ETF and has rebalanced nothing in four years because the fund does it internally. Total cost: 0.25%. A robo at 0.50% would give him the same portfolio, the same automation, and a second fee. Verdict: no.
• Ines has €60,000 in a savings account, has twice started investing and stopped, and sold everything in a previous downturn. A robo at 0.55% all-in gets her invested this month, automated, with rebalancing she will never have to authorise. Her fee costs her roughly €330 a year; the alternative has cost her a decade of returns. Verdict: obviously yes — and if in five years she finds the annual routine easy, she can migrate.

Same product, same fee, opposite answers, decided entirely by what each of them would otherwise do.