There are two kinds of numbers on a mutual fund factsheet. One kind describes what happened. The other describes what will happen.
Past return, alpha, Sharpe ratio, rank within category — all describe the past, and are noisy estimates of quantities that may not persist. Expense ratio describes the future: it is contractually specified, charged daily against assets, and will be deducted next year regardless of what markets do.
That asymmetry is why cost has repeatedly emerged as the single most reliable predictor of future relative fund performance.
Sharpe’s arithmetic
William Sharpe’s 1991 note makes the underlying argument in about two pages, and it does not depend on any empirical claim.
Divide all investors in a market into passive (holding the market portfolio) and active (everyone else). Passive investors, by construction, earn the market return before costs. Since the two groups together hold the entire market, the active group in aggregate must also hold the market portfolio, and therefore must also earn the market return before costs.
After costs, the active group must earn less than the market, by exactly the amount of its costs — which are higher than the passive group’s.
This is arithmetic, not a hypothesis about market efficiency. It holds whether or not markets are efficient, whether or not skill exists. It does not say no active manager can outperform; it says they can only do so at the expense of other active managers, and that the average active rupee must underperform by its cost disadvantage.
The practical reading: in the aggregate, cost is not one input among many. It is the term the arithmetic guarantees will show up.
The empirical version
Morningstar’s long-running fee study has repeatedly found expense ratio to be a stronger predictor of future fund performance than its own star rating. Cheapest-quintile funds outperformed priciest-quintile funds across every asset class and time period examined, with success rates roughly double.
The SPIVA scorecards approach it from the other end and find the same asymmetry noted in the persistence literature: persistence at the top of the performance table is weak, but persistence at the bottom is meaningful — and the mechanism is cost. High fees are persistent in a way that good stock picking is not.
The Indian arithmetic
Direct plans in India were mandated by SEBI in January 2013. A direct plan and its regular counterpart hold the identical portfolio, managed by the identical manager. The only difference is that the regular plan’s expense ratio includes distributor commission.
The gap is typically in the range of 0.5% to 1.2% a year for equity schemes. Compounded over a long horizon, that difference is not marginal:
At an 11% gross return over 20 years, ₹10 lakh grows to roughly ₹80.6 lakh. Take 1% a year off — a 10% net return — and the same ₹10 lakh grows to about ₹67.3 lakh. The 1% cost differential consumed over ₹13 lakh, more than the original investment.
This gap is larger than the dispersion most fund-selection effort is trying to capture. Switching from a regular to a direct plan of the same scheme is, for most investors, a bigger improvement than any amount of fund selection between schemes.
What an expense-ratio screen does and does not do
It is a blunt instrument and should be understood as one.
Sorting purely on cost sorts toward index funds and ETFs. That is not a bug — index funds are cheap because they do less, and the whole argument above says doing less is where the reliable edge is. But it means the screen is closer to “build a low-cost index-tracking core” than to “find good active managers cheaply.”
Cost is not comparable across categories. A liquid fund charging 0.15% and an equity fund charging 0.6% are not competing. Ranking across a mixed universe by cost alone ranks by category, not by value for money within a category. If you want the latter, screen within a category first.
A very small fund is a different risk. Schemes with tiny asset bases carry closure and merger risk that has nothing to do with their portfolio, and their stated expense ratios are sometimes subsidised in a way that does not last. An AUM floor is a crude but effective filter for this.
Expense ratio is not the total cost. It excludes the fund’s own transaction costs and the securities transaction tax it pays on portfolio turnover, both of which come out of NAV. Two funds with identical stated TERs can differ meaningfully in total drag if one turns its portfolio over three times a year and the other once.
Rebalance almost never
An expense ratio changes when the AMC files a change, which is rarely and usually by a few basis points. The screen’s output therefore barely moves.
That is the point. Rebalancing a cost screen monthly is pure friction against a signal that has not changed — and in a mutual fund context, friction includes exit loads and short-term capital gains tax. Twice a year is generous. Annual is defensible.
This is the general principle stated in its cleanest case: match rebalance frequency to how fast the signal changes. A monthly rebalance on an annual signal is a cost programme with a strategy attached.
Try it
The Low-Cost Large Fund Core template screens schemes with a published expense ratio and assets above ₹500 crore, holds the twelve cheapest equally weighted, and rebalances twice a year.
Run it beside the momentum rotation template over the same window. One screens on the noisiest available signal at high turnover; the other on the most certain one at almost none. Comparing them is the cleanest available demonstration of which kind of number is worth building a portfolio on.
Further reading
Cost compounds the same way outside funds. Brokerage charges in India: the full cost stack does the equivalent arithmetic for a direct equity trade, itemised down to the exchange fee.