Seven checks, in order of how much they change what you own. The two everybody runs - the name and the expense ratio - are the two cheapest, and the one nobody runs is where the exposure lives.
14 min read · Level 2
An hour with a holdings file is worth more than a year of comparing expense ratios, and almost everybody does the opposite - because the fee is one number on the front page and the holdings are a spreadsheet with several hundred rows. I want to be careful here, though, because the obvious conclusion from that sentence is wrong. Fees deserve the attention they get for a reason that has nothing to do with importance in the abstract: a fee is knowable in advance, applies with certainty, and compounds, whereas any conclusion you reach from the holdings file is a judgement that may turn out to be mistaken. Choosing on cost is choosing on the one variable you can actually be right about. The correct position is therefore not that holdings matter more than fees, but that they answer different questions and the order is fixed. The holdings decide whether this fund is the exposure you want at all, and no fee is low enough to fix a wrong answer there. Only once two funds deliver substantially the same exposure does the fee become the decision - and at that point it becomes the whole decision, as the 2,738.91 above shows on portfolios that were identical. Most people run the second test without ever running the first, which is how somebody ends up owning the cheapest available version of something they never meant to own.
Seven checks, ordered by how much each one changes what you actually own. The order matters: most people run the two that are printed and cost nothing to check - the name and the expense ratio - and stop there, never reaching the first, where the exposure actually lives.
One. The holdings, by weight. Fund A holds 500 companies with 35.8% in its ten largest and 4.9% across its smallest 250. "Diversified" is a claim about the count and concentration is a claim about the weights, and only the second one describes how the holding will behave.
Two. The distance between the name and the holdings. Two funds, the same theme in the title:
Fund E: 28 holdings, ten largest 68.9%, expense ratio 0.60%, turnover 94% Fund F: 310 holdings, ten largest 22.4%, expense ratio 0.12%, turnover 8%
A one per cent fall across each fund's ten largest holdings moves Fund E by 0.689% and Fund F by 0.224%. On the same event, E moves 3.08 times as far. The names are interchangeable; the instruments are not.
Three. The rules it follows. An index is a product with a written methodology: what is eligible, how it is weighted, when it is reconstituted, what happens to a company that breaches a threshold. A fund that tracks an index has delegated selection to whoever wrote those rules, and it is worth knowing that they are rules rather than a description of a market.
Four. The expense ratio, and what it excludes. It is printed, it is certain and it is not the total cost. The fund's own dealing costs sit outside it and rise with turnover; the spread you pay to get in and out sits outside it; any platform charge sits outside it. Lesson five puts all four layers together.
Five. Tracking difference, and its very different cousin tracking error.
Fund A: index +7.00%, fund +6.91%. Tracking difference -9bp against a 5bp fee. Fund D: index +7.00%, fund +6.88%. Tracking difference -12bp against a 20bp fee.
Fund A finished 4 basis points further behind than its fee alone explains - dealing costs, cash held for redemptions, sampling rather than full replication. Fund D finished 8 basis points ahead of what its fee implied, which usually means it earned something the holdings did not produce, most often by lending its securities for a fee. That income is real and it carries exposure to whoever borrowed them.
Tracking error is a different measurement entirely: the variability of the gap, not its size. A fund that is reliably 20 basis points behind every year has a large tracking difference and a very small tracking error. Confusing the two is common enough that the words are sometimes used interchangeably in marketing material, and they mean opposite things.
Six. Size and survival. A fund with 40m of assets charging 0.15% collects 60,000 a year, which does not pay for an audit, a custodian and a compliance function. Small funds are commonly supported by a fee waiver with an expiry date, merged into a larger sibling, or closed - and a closure is a forced sale on a date somebody else chose, with whatever tax that triggers.
Seven. Liquidity against your own size. Wanting 250,000 of a fund whose underlying basket trades thinly is a different proposition from wanting 2,500 of it. Size the order against the holdings, as lesson three set out, not against the ticker's volume.
Because the arithmetic of what a fee has to buy is the most clarifying calculation in this subject, and almost nobody runs it.
Take a fund charging 0.85% whose holdings substantially match an index available at 0.05%. The fee difference is 0.80 percentage points a year. Now ask how different the portfolio actually is.
If 85% of the weight matches the index, the differing slice is 15%. That slice must add 0.80 / 0.15 = 5.33 percentage points a year to break even.
If the differing slice is 40%, it must add 0.80 / 0.40 = 2.00 percentage points a year.
Read that as a test rather than as a verdict. The question a fee poses is not "active or passive" - that argument is had properly in active-vs-passive and is not repeated here. The question is whether the fee is proportionate to how different the portfolio is. A genuinely distinct portfolio charging an active fee is a coherent proposition you can accept or reject on its merits. A portfolio that is 85% the index charging 0.80 points over the index is asking a very small slice to do a very large amount of work, and the holder is paying a fee on the 85% that is doing nothing different at all. That is closet indexing, and it is visible from the holdings file in twenty minutes.
The same arithmetic protects the other direction, which matters just as much. A fund whose holdings genuinely differ from any cheap alternative cannot be dismissed on its expense ratio, because there is no cheap alternative delivering that exposure. Comparing fees between funds that hold different things is meaningless, and it is the single most common mistake made by people who have correctly learned that fees matter.
There is one more reason the holdings file outranks everything else on this list. Every other check on it is a summary somebody produced. The holdings are the primary document - the same distinction verification-vs-assumption draws - and a summary can be accurate, out of date, or selective, without any way to tell which from the summary itself.
Educational, not personalised financial advice. Nothing in the Investor Academy is a recommendation to buy, sell or hold anything — what is right for you depends on your own circumstances and the rules where you live.
Three funds, all offered as a way to own a broad market, appraised on the seven checks.
Fund A Fund E Fund H Holdings 500 28 500 Ten largest 35.8% 68.9% 35.8% Expense ratio 0.05% 0.60% 0.42% Turnover 3% 94% 4% Tracking difference -9bp n/a -47bp Assets 2,050m 190m 88m
Fund E is not a comparator at all. It holds 28 companies with two thirds of the money in ten of them and tracks no index, so a tracking difference does not exist for it. It is a concentrated portfolio in a wrapper, which may be exactly what somebody wants - but the fee comparison against Fund A is meaningless, because the exposures are unrelated.
Fund H is the interesting one. It holds the same 500 companies at the same weights as Fund A. Its tracking difference is -47 basis points against a 42 basis point fee, so it lags by 5 basis points more than the fee alone accounts for - an ordinary result, and nothing is wrong with it. It is simply more than eight times the price of an identical portfolio.
Suppose the basket returns 7% a year, with each fund's fee charged on the year-end value, on 10,000 held for twenty years.
Fund A: 1.07 x (1 - 0.0005) = 6.9465% a year -> 38,311.71 Fund H: 1.07 x (1 - 0.0042) = 6.5506% a year -> 35,572.80 Difference 2,738.91
That is 7.15% of Fund A's ending figure, lost to a difference in packaging cost on an identical set of companies. It is also the cleanest demonstration available that the holdings and the fee are separate questions: here the holdings are identical, so the fee is the entire decision, and in the comparison with Fund E the holdings are so different that the fee decides nothing.
That figure assumes a return in order to demonstrate the mechanism. Nobody knows what the basket will do, and the fee difference applies whatever it does - which is the only reason it is worth spending an hour on.
Mistaking the number of holdings for diversification. Five hundred holdings with 35.8% in ten is not what "five hundred" suggests to somebody who has not looked. The count is a marketing number and the weights are the instrument.
Comparing expense ratios between funds that hold different things. A fee is only comparable when the exposure is. Two funds with different holdings and different fees pose a question about which exposure you want, and the cheaper one is not the answer to it.
Confusing tracking difference with tracking error. One is how far behind the fund finished; the other is how much that gap wobbles. A fund can be excellent on one and poor on the other, and the words are used loosely enough in marketing copy that you should check which is meant every time.
Judging a fund on a single year of tracking difference. Securities lending income, the timing of an index reconstitution and one large corporate action can each move it by several basis points, and none of them says anything durable about how the fund is run.
Buying the biggest fund because it is the biggest. Size is evidence of survival and of the operational resources to keep running, both of which are worth something. It is not evidence about the holdings, and past flows are not a view.
Exchange-traded funds typically publish theirs daily and open-ended funds monthly, usually as a spreadsheet. The top-ten table on the factsheet is a summary chosen by the manager; the file is the fund. Note its date, because a monthly file read in the fourth week is describing a portfolio that has since moved.
Do it before you read anything else about the fund, so the figure is not coloured by the marketing. Then ask yourself the only question that matters about it: if those ten holdings had a bad year together, would the resulting fall in this position be one you could live with? That is a question about your own circumstances, and it cannot be answered by the word diversified.
It is a published document, usually around thirty pages, and it contains every decision that determines what the fund holds. If you are not going to read it, at minimum find out how the index weights its constituents and how often it is reconstituted, because those two facts drive most of what the fund does. Not reading it is a legitimate choice; believing you have therefore avoided somebody's judgement is not.
Take the fund's total return over a period and the index's total return over exactly the same period, and subtract. Then compare the gap with the expense ratio. If the fund is further behind than its fee, find out why; if it is ahead of its fee, find out what it earned that the holdings did not, because securities lending income comes with a counterparty.
The expense ratio is one of four layers. Write the other three down alongside it - the fund's internal dealing costs implied by its turnover, the spread you will pay on the round trip, and any platform charge - and only then compare it with an alternative. Lesson five works a full position through this.
A small, young fund with a waiver expiring next year is a different proposition from its headline fee, and a fund too small to pay for its own administration is a candidate for closure or merger. Both of those end in a disposal you did not choose, on a date you did not choose, with the tax consequence that follows.
Express the position you want as a share of the daily traded value of the underlying basket rather than of the fund. For a broad index fund the answer is negligible and you can stop. For anything narrow, a position that takes several days to build or unwind is a position whose exit price you do not control.
Before buying any fund, when deciding between two funds that look similar, at least annually on anything you already hold, and whenever a fund's return diverges from what you expected of its holdings.
Do not run this appraisal on a fund whose full holdings are not published - the absence is the finding - and do not use it to choose between exposures, which is an allocation question that lesson six and asset-allocation answer, not a question about which fund is better made.
Funds launched around a theme frequently find that the investable universe matching their stated criteria is much smaller than the theme itself. The resulting portfolio holds perhaps twenty-five to thirty names, of which a handful are large enough to absorb the fund's money at meaningful weight, and two thirds of the money ends up in ten holdings. Holders who bought for exposure to an idea have bought a concentrated position in a few companies that happen to be the largest way to express it. The disclosure is complete - every weight is in the holdings file from the first day - and the gap persists because the theme is what was marketed and the weights are what was bought.
When a theme is narrower than it sounds, the fund built on it is more concentrated than it sounds. Check the ten largest weights before deciding a thematic fund is diversified exposure to an idea.
Small funds often launch with a stated expense ratio supported by a waiver: the manager agrees to absorb costs above a level, for a period, after which the waiver may or may not be renewed. The headline figure is real while it lasts. What it signals is that the fund is not currently large enough to cover its own operating costs, which is a fact about its viability rather than about its pricing. Two outcomes follow commonly - the fee rises when the waiver expires, or the fund is merged or closed - and both arrive as a decision the holder did not make, with a disposal and its tax consequences attached.
Find the expiry date of any fee waiver and the fund's assets under management in the same sitting. A waived fee on a sub-scale fund is a temporary price on an uncertain vehicle.
Create a free account to read the rest and apply it to your own investing.