The factor leg you can actually buy

A friend forwarded a smart-beta value fund last week. The pitch leaned on the value premium from the textbooks, something like five points a year over the market across decades of data. His fund had done maybe three points over the index, and he was ready to sell it for underperforming.

I told him to stop. He hadn't bought a bad fund. He'd bought one leg of a two-legged trade.

Here is the part the brochure skips. Every factor premium in the academic literature is a long-short spread. You go long the cheap stocks and short the expensive ones. Long the small, short the large. Long the recent winners, short the recent losers. The premium is the distance between those two baskets, not the return of either one on its own.

A plain mutual fund or a smart-beta index fund can't short anything. So it owns the long leg and leaves the short leg on the table. And how much of the premium actually lives on the long side depends entirely on the factor. The long leg keeps almost all of the size premium, around 60 percent of value, and only about half of momentum (Israel and Moskowitz, 2013). Same word, three very different trips from the paper to your portfolio.

Then there is the quieter problem. Whatever the long leg keeps still arrives wrapped in full market beta. A long-only value fund is the index plus a tilt, not the value premium standing alone. Strip the beta and the tilt is smaller than the sticker.

None of this makes factor funds bad. It just means you should know which leg you are buying and how much of the premium survives before you pay an active fee for it. Two or three points of genuine tilt over the index is worth having. Paying a long-short price for a long-only slice is not.

Two legs on paper, one leg in your account

The picture is simple once you draw it. The research portfolio holds two baskets at the same time and earns the gap between them. The market moves both baskets together, so the market cancels out. What is left is the factor and only the factor.

A fund that cannot short gets the right hand picture. Most of what it earns in any year is the market. The factor is the thin band on top, and that band is also smaller than the paper premium because the short leg's share never arrives.

The beta is printed on NSE's own factsheets

You do not have to take this on trust. NSE publishes a five year beta to the Nifty 50 for each of its factor indices. Beta tells you how much an index moves for each 1 percent move in the market.

None of them sits near zero, which is where a pure long short factor lives. The value and momentum indices swing more than the market. Quality swings less. Every one of them is, first, a way of owning Indian large and mid caps.

Strip the beta from my friend's three points

Here is the worked example, in the opening's own numbers. Say the index did 12 percent a year and the fund did 15, the three points my friend was unhappy with. Say the fund's beta is 1.1 and cash earns 6.5 percent a year. Both are assumptions for the arithmetic, not his fund's figures.

Step one: the part the market explains. Cash plus beta times the market's return above cash: 6.5 + 1.1 x (12 minus 6.5) = 12.55 percent. Step two: the tilt is what is left. 15 minus 12.55 = 2.45 points. The sticker said three. A fund with a beta above one earns part of its lead just by swinging harder in a rising year, and gives it back in a falling one.

Try a year where the index falls. Put the index at minus 10 and the fund at minus 9. The fund looks one point better, but with a beta of 1.1 the market alone would have taken it to minus 11.65, so the tilt was worth 2.65 points. Beta flatters a fund in rallies and hides it in falls. Read the tilt, not the gap.

The tilt is real, and it is not steady

The last five years on NSE show both halves of the opening. Each of these is a long only index, total return, measured against its own parent universe.

Value was 14.6 points a year ahead of its parent. Momentum was about one point ahead. Quality trailed by 1.9 points. All three still owned the market first, which is why a long only factor index can lag for years while the paper factor looks fine. The free module on introduction to factors sets out what each factor owns, with the P/E, P/B and beta of each NSE index side by side.

The same logic decides how a blend behaves. Two long only factors share the market, so they move together more than the research says they should. I worked through that on NSE data in the value and momentum post, and the case for holding several tilts at once is in multi factor stacking.