the small-cap edge is a regime, not a rule

A friend forwarded me a small-cap fund chart on Sunday morning. Five years, straight up and to the right. His question was simple: if small beats large, why isn't my whole SIP sitting in small caps?

It is a fair question. The answer is the part the brochure quietly skips.

Small beats large is real, in the textbook. Banz documented the size premium in 1981 at roughly 5.2 pct a year. It built a thousand fund factsheets. Then the paper got read, the money came in, and the edge did what crowded edges do. Out of sample since, the premium reads closer to 1.5 pct. Around 71 pct of it has decayed.

You can see the same thing up close without any academic paper. Take the same equal-weight engine and the same fortnightly rebalance, change only the cap segment, and the cap curve flattens fast. Allcap 48.02 pct CAGR. Midcap 41.21. Smallcap 41.03. Microcap 40.56. The return barely moves as you go smaller. The drawdown keeps deepening, from minus 22.7 pct down to minus 26.4 pct. Microcap gives back almost half a point of return for four points more pain. (Backtest, illustrative, not a guarantee.)

Here is the part that matters. The premium that survived is not a reward for being clever. It is compensation for illiquidity and distress risk. You get paid to hold what other people cannot sell in a hurry. So it pays in one kind of weather and reverses in the other. Early cycle, liquidity loosening, small caps cheap and uncrowded, it works. Late cycle, risk-off, everyone already in, it hands the book back.

So size is a tilt, not the whole portfolio. Read the regime, not the twenty-year average. Pick a weight you can SIP through nine straight red months, then stop touching it.

Small beats large in the right weather. The rest of the time you are renting volatility and calling it a factor.

How factors actually behave: