Do cheaper, smaller, more profitable companies pay more?

Cheapness lost 63% of itself on the way to a real account. Momentum lost 80%. The market lost nothing. And the fund column is before the fund’s own fee.

There is no mystery in that pattern. Holding the whole market requires no trading at all. Buying the cheap companies and avoiding the dear ones requires constant trading, and buying recent winners requires it every month. The trading eats the idea. Two further findings point the same way: across 215 strategies that banks built and then sold to clients, the typical one lost 73% of its apparent quality once real money ran through it; and the 2.43-point advantage momentum showed over the market before tax comes to 0.60 points after tax at current rates, and slightly negative at historical ones.

What the raw ideas are, and what they measure

The academic versions are paper portfolios that buy one end of a ranking and sell the other short. Value ranks companies by their accounting net worth divided by their stock market price, and buys the cheap-looking end. Size ranks by how big the company is, and buys small. Profitability ranks by operating profit against net worth, and buys the consistently profitable. Nobody can buy these directly. They are gross of trading costs, borrowing costs, fees and tax.

Pooled across the US, other rich countries and emerging markets, and measured against zero:

• Cheap companies: 4.74 points a year, 1994 to 2025, against a smallest detectable effect of 3.35 points. It clears, narrowly.
• Profitable companies: 2.53 points a year, against a floor of 2.62 on its much shorter twelve-year window. Below what the test could see. We could not tell.
• Small companies: 0.33 points a year against a floor of 2.47. Nothing there.

The general result matters more than any of the three. On the public data that everyone uses, no advantage below about 2.6 points a year can be signed at all. Adding more countries does not rescue it, because countries move together: three regions of the value idea are worth about one and a half independent looks, not three.

And here is what noise looks like on this data. We ran a purely random series with the same length and the same jumpiness through the identical machinery. It produced an apparent advantage of 1.98 points a year, a 53.2% fall from its peak, and 247 months under water. Anyone hunting a two-point effect on a series this short will find one whether or not it exists.

Our own leaning funds, with the caveats attached

The four funds we lean with came to 0.79 points a year against a cheap index fund over 427 months, range 0.30 to 1.32. That clears its floor. It clears it by about 10%, and on a like-for-like comparison you would need thirty years of holding to establish it, not the thirteen we once printed.

Two things belong beside that number. First, 86% of it comes from series that begin in 1990 and sit outside the US, which is the shortest data we hold and none of it was kept back for a fresh test. Second, charging each fund the shortfall we actually measured on it drops the 0.79 to 0.30, below its own floor. The real developed-market value funds you can buy have run 2.3 to 2.9 points a year behind their own benchmarks over the 55 months they have existed.

The one to drop first

One of the four funds we lean with is a momentum fund, and momentum is the weakest thing on this site. It has the largest gap of anything measured between what it promises on paper and what anyone has actually collected: 9.48% a year on paper against 1.86% that real mutual funds delivered per unit of exposure, before their own fees.

Tax takes much of what is left. Momentum’s 2.43-point advantage over the market before tax comes to 0.60 points after tax at current rates, and turns slightly negative at historical ones.

The reason for both is trading. The momentum fund replaces 105% of its holdings in a year, against 4% for the international value fund sitting beside it. Every one of those trades pays a spread and, in a taxable account, creates a tax bill. A reader who wanted to hold one fewer leaning fund should drop the momentum one first.

So the honest summary is that the cheap-companies idea is the only one of the three we can sign, that it is worth much less to a fund holder than to a paper portfolio, and that the market itself is the thing that arrives intact.

What would change our mind

Twenty more years of data outside the US, held back rather than mined. Or fund records long enough to measure what real value funds deliver over a full cycle, instead of the 55 months we have.

Where these numbers come from

• Kenneth French’s public data library of monthly returns for US, developed ex-US and emerging market stock factors, through December 2025.
• Patton and Weller, "What You See Is Not What You Get: The Costs of Trading Market Anomalies", Journal of Financial Economics, 2020.
• Suhonen, Lennkh and Perez, on 215 bank-sponsored strategies before and after they went live.
• Israel and Moskowitz, on what momentum costs a taxable investor.
• Our own measurement of the exposures four real funds deliver, from their filings, run over 427 months to May 2026.