Does leaning toward cheap companies pay?

On paper and in a fund

Cheap companies have beaten expensive ones on paper for a century. Companies that have been rising have kept rising for a few months more. Both are well documented. The question for a person with a brokerage account is what a fund delivers after it has paid to trade, and the answer is much less than the paper number.

A study of what funds actually delivered found the plain market fund returned 6.93% a year against 6.72% on paper: nothing lost. The value premium was 7.76% on paper and 2.84% in funds, so nearly two thirds vanished. Momentum was 9.48% on paper and 1.86% in funds. Value survives the trip into a fund. Momentum mostly does not, because a momentum fund has to replace its holdings every year.

What the lean added

The lean on this site is six funds: VTI 49%, VXUS 16%, VTV 15%, AVDV 10%, IDMO 5%, AVES 5%. Against a cheap index mix holding the same 65/35 split of US and foreign stocks, it added 0.80 points a year between 1990 and 2026, with a range of 0.36 to 1.31, and it was ahead in each of five separate eras.

That is a backtest, and the funds were chosen by someone who had seen the data. So I ran the same lean on 1981 to 1990, ten years the fund choices had nothing to do with. It added 0.89 points a year, range 0.44 to 1.34, and both halves of the decade were positive. The direction reproduces. The size is less certain: on some other reconstructions of the same decade it reads 0.5.

Most of the gain came from outside the US. The US value fund, VTV, is the least supported of the six, worth about 1 hundredth of a point a year per point of money in it. AVDV, which holds small cheap companies in developed markets outside the US, and IDMO, which holds foreign companies that have been rising, carried the most. The US value premium on its own, measured from 1994 when it was published, cannot be told from zero, and 79% of what it did earn came in the single year 2000. Across the US, other rich countries and emerging markets together it clears.

A warning about how easy this is to fake. A made-up series of random monthly returns, no edge at all, run over the same window, produced an apparent advantage of 2 points a year with a 53% worst fall. Thirty-six years of data cannot tell a small edge from noise. About thirty years of holding this lean would be needed to be confident. Hence "probably".

What losing looks like

US value has trailed the US market since September 2008, and by the middle of 2026 it was 54.3% behind. Foreign stocks trailed the US for 18 years and finished 69.0% behind. Neither has recovered. If you hold this lean, that is the stretch you are signing up for, and it will come with no warning and no way to tell whether the premium is late or gone. Do not hold it because you think you could sit through that. Hold it only if you have already sat through something like it and know how you behaved.

The momentum fund is the first to drop

IDMO holds foreign developed-market companies that have been rising. It is 5% of the portfolio and it is the weakest of the six. It turns over 105% of its holdings a year, so most of the tax shelter an ETF normally gives you is lost, which is why it goes in a retirement account. Momentum's advantage over the market of 2.4 points a year before tax shrinks to 0.6 after tax at current rates. And all three regions' momentum strategies crashed together in 2009, by 53% in the US, 37% in developed markets and 29% in emerging markets. I keep it because it is a different idea from cheapness, and two ideas are better than one. If you want a simpler portfolio, drop it first.

I did not add a US momentum fund. The best one, SPMO, was worth 0.02 points a year against the portfolio without it, at a 5% weight and with a range straddling zero, and it moves with IDMO closely enough that two momentum funds are about 1.2 ideas, not 2.

Why five ideas in one country beat one idea in three

Stacking ideas only helps if they are different ideas. Value in the US, value in Europe and value in Asia are one idea, and together they count as about 1.5 independent bets out of 3. Five different ideas in one country counted as 5.52 out of 5, because some of them lean against each other. That is why the lean here holds three ideas that differ in kind, cheapness, profitability and momentum, plus one small-company fund, rather than the same thing on three continents.

If each idea has a 55% chance of beating the market and the ideas are unrelated, you need 5 of them for 60% odds, 18 for 70%, 45 for 80% and 105 for 90%. Nobody has 105 unrelated ideas. A realistic portfolio has three or four, which is why the claim I can make is "better odds", never "will beat".

Cheaper versions and dearer ones

VTV costs 0.03% and holds big cheap US companies. AVUV costs 0.25% and holds small cheap ones. Swapping one for the other made no difference the test could see, and most of AVUV's extra exposure is to smallness, which no market I measured could show paying anything. The price gap between cheap and expensive US stocks is wider now than it has been 81% of the time on record, which is a reason to hold what you have and not to enlarge it.