The split decides more than the pick

A friend spent his entire Sunday comparing two large-cap funds. Spreadsheet open, expense ratios lined up, five year charts side by side. He wanted to know which one to pick.

I told him he was sweating the small decision and skipping the big one.

The big one is the split. How much equity, how much debt, how much gold. Brinson and his co-authors put numbers on this back in 1986, and the finding has held up through every market since. The mix you choose explains roughly ninety percent of how much your portfolio's returns swing around over time. Which specific fund you drop into each sleeve is the smaller lever, not the larger one.

Now the part the brochures quietly mangle. That ninety percent is not "ninety percent of your return comes from asset allocation." Ibbotson and Kaplan cleaned this up in 2000. It is about variability, how bumpy the ride feels over the years, not about where the magnitude of your return came from. Look across different funds instead of across time and the split explains closer to forty percent of why yours landed differently from mine. On average across everyone, the policy mix accounts for roughly the whole long-run level, because active picking is close to a zero-sum game before costs.

The cleanest way to see it is to hold the index in every sleeve, so the picking is identical, and just move the mix. All equity returns about 10.4 percent and makes you sit through a 52 percent drawdown to get it. A 40 40 20 mix returns about 8.5 percent and never falls more than 18 percent. Two points of return between them. Thirty four points of drawdown. Same holdings. The split did all of it.

So the honest version. You can spend a weekend on the pick and one Tuesday afternoon on the split. The Tuesday afternoon decides more. The brochure sells the pick. The math pays the mix.

What your asset allocation actually decides:

One real year, same indices, five mixes

Here is the test from the opening on actual NSE data, with no fund picking at all. Each sleeve is an index. Equity is the Nifty 50. Debt is the Nifty Composite G-sec Index. Gold is an NSE listed gold ETF. The window is one year, 30 Sep 2025 to 30 Sep 2026.

Over that year the Nifty 50 fell 8.09 percent, the G-sec index rose 2.81 percent and the gold ETF rose 26.85 percent. Mix them and the result depends almost entirely on the weights.

An 11 point gap between all equity and 40/40/20, from the split alone. I would not read that year as a verdict on any mix. It was a bad year for Indian equity and a strong one for gold, and a different year can reverse the order. The lesson is the size of the lever, not its direction.

Moving the lever yourself

Now give equity a one point better fund, a strong result for a picker: change the equity return from minus 8.09 to minus 7.09. At 40 percent equity the portfolio moves by 0.4 points. Move 20 points from debt to equity instead and it moves by about 2.2, more than five times as much. Take those 20 points from gold and it moves by about 7. That is the ratio the friend's Sunday spreadsheet was missing.

How to read your own split

Brinson's method, still used to grade professional funds, breaks a portfolio's result into two parts. The allocation effect is what the weights did, given the index return of each sleeve. The selection effect is what the funds did against their own index. The free Learn module on performance measurement works through this attribution on a rupee example, with a quiz and a certificate.

For your own portfolio the first pass is short. Write down today's split across equity, debt and gold, compare it to the split you meant to hold, and decide whether the gap was chosen. The asset allocation calculator gives a starting split from a horizon and a risk level, and the cash dial, not the stock list shows why risk level should change the mix rather than the names.