Case: pick the fund line-up from six factsheets

The cheapest-looking fund on a list is the one with the lowest ongoing charge, because that is the number printed largest. For our users it is rarely the cheapest to own.

S2 shows why. Inside the fund it is excellent: it trails its index by only 0.06%. But it pays dividends out, and as a US-domiciled fund 30% of them is withheld for a user like ours: 0.54% a year. It needs a separate foreign broker with a 0.40% platform fee and 1% per conversion. All-in over ten years: 1.20% a year.

S1, on the same index, trails it by 0.25%, but it accumulates, sits in the Zapas app and converts at 0.50%. All-in: 0.61%. The fund with a fee four times higher costs half as much to own.

The lesson isn't that S2 is a bad fund. It is that a fund's cost depends on who owns it, where and how.

A difference of 0.59 points sounds abstract. On a balance of $10,000 it is $59 a year: S2 costs about $120 and S1 about $61 a year all-in over a ten-year holding.

Two things make that gap bigger than it looks. First, it repeats every year and compounds: money lost to costs also loses everything it would have earned later, which is how a 1.3-point gap grows to $38,011 over 40 years in a $100-a-month plan earning 5% real before costs. Second, balances grow, so the same percentage becomes more dollars each year.

When you present a line-up to a team or to users, show both: the all-in percentage, so funds can be compared, and the yearly dollars on a typical balance, so people can feel it. Keep the holding period and assumptions next to the numbers.

An all-in ranking is the fair starting point: S3 0.54%, S1 0.61%, S5 1.13%, S2 1.20%, S4 2.40%, S6 2.70%. But a cost ranking compares price tags of things that may not do the same job.

S3 is the cheapest overall, but it holds short bonds: a different asset class with a different role, lower expected return and much smaller falls. It doesn't compete with S1; it complements it. S5 is cheap-ish, but it follows one sector with 60 companies, so its risk is nothing like a world fund's.

So use cost to choose within a role (which world stock fund, which bond fund), and use the role to choose which funds belong in the line-up at all. A cheap fund in the wrong role is still the wrong fund.

A line-up for long-term investors needs a few clear roles. In our model portfolios that means a broad world stock fund for growth and a bond fund for the part that must not fall much. Each role has a test before cost:

• world stock: thousands of companies, market-weighted, many countries;
• bonds: high quality, short or medium term, in the user's planned spending currency where possible.

Only funds that pass the role test go on to the cost comparison. S5 fails the world-stock test: 60 companies in one sector is a concentrated bet, whatever its price. S4 fails too: 45 picked stocks and 2.40% all-in, with no evidence that its early lead repeats. S6 mixes assets in hryvnias with a 3% entry fee and 2.70% all-in.

This order also protects the product from trends: a theme that is hot this year is a marketing story, not a role.

It is tempting to fill every slot on a screen. For an investing product, every extra fund is also an extra decision for the user, and extra decisions invite mistakes: chasing last year's winner, piling into a theme, switching often and paying the doors again.

A small line-up works when each fund has a clear job. One broad world stock fund and one high-quality bond fund can build portfolios from careful to growth just by changing the mix. A third fund earns its place only if it does a job the two can't, at a cost that fits.

Ask of any candidate: which role does it fill, does it pass the role test, and what is its all-in cost for our users? If no candidate passes, an empty slot is a decision, not a gap. Later modules may add a fund for the safe part in the user's spending currency.