Rebalancing harvests volatility

Friend pulled his CAS out over coffee on Saturday. He had been upset about his five-fund SIP underperforming the index by a margin he could not explain. Buy and hold, he said. Equal weight, five funds, started in 2021. Should have been the boring sensible thing.

I made him add the columns up.

The actual split today is 38, 17, 16, 15, 14. One fund had quietly become more than a third of the book on its own. Two years without a trade. The weights traded for him.

This is the part of buy and hold nobody puts in the brochure. You can pick a perfect equal-weight portfolio on day one and the market will resize it on you while you are not looking. The diversification is on the start statement. Not the current one.

So I drew him the napkin version.

Two assets, fifty fifty, total one hundred rupees. Asset A drops thirty percent, Asset B rises thirty percent. Total still one hundred. Both assets then revert to where they started. Buy and hold path ends at one hundred. Flat round trip, zero return.

Now run the same swing with a rebalance at the high. Sell fifteen rupees of B at the peak, buy fifteen rupees of A at the trough. When both prices revert, the portfolio ends at one hundred and ten. Same start prices. Same end prices. Ten percent harvested by the act of rebalancing.

The textbook name for this is a volatility harvest. The bigger the swings between holdings and the more they mean-revert, the bigger the harvest. The math is not opinion. It falls out of the geometry of compounding.

Cadence decides how much of it you actually capture. Short cadence catches mean reversion before drift sets. Long cadence lets it ferment. Annual is buy and hope. Trade off is small turnover cost in exchange for the harvest plus a portfolio that still looks like the rule sheet you bought.

Three rules I left him with over the second coffee.

One. Rebalancing is not churn. It is the discipline that keeps the portfolio you bought from being eaten by the portfolio you accidentally became.

Two. Without it you are a momentum trader by default. Just a slow one.

Three. If your weights have not been touched in two years, you do not own the portfolio you started with.

What the drift looks like on one page

Five funds at 20 percent each is the plan. The statement two years later is the portfolio. One fund at 38 percent means almost twice the planned exposure to whatever that fund holds, and the other four have each shrunk below their target.

The napkin, with your own swing

The ten rupees in the opening come from one formula. Start 50/50. Asset A falls by some percentage, asset B rises by the same. You rebalance back to 50/50 at that point. Then both return to their starting prices. The untouched portfolio ends at 100. The rebalanced one ends at 50 divided by (1 minus the swing) plus 50 divided by (1 plus the swing).

Try 10 percent and the harvest is about 1 percent. Try 50 and it is about 33. The harvest grows with the square of the swing, which is why it is real for volatile assets that take turns and close to nothing for calm ones.

The same test on real Indian data

The napkin is a best case. So I ran the plain version on actual NSE closes. Put Rs 10,00,000 into a 50/50 mix on 14 Oct 2024: Rs 5,00,000 tracking the Nifty 50 and Rs 5,00,000 tracking the Nifty Composite G-sec Index. One copy is left alone. The other is put back to 50/50 at every quarter end. Both are measured on 30 Sep 2026.

A gap of Rs 343 on Rs 10 lakh. No harvest worth the name. The Nifty 50 fell about 10 percent over the window and the G-sec index rose about 9.6 percent, so the move was mostly one way, not a swing that reversed. Government bonds also move far less than stocks, so the swings between the two were small.

What rebalancing did deliver here was the other job. The untouched copy ended with 45 percent in equity, not 50. The rebalanced one carried the risk that was chosen, every quarter. That is the honest summary: the harvest is a bonus that shows up when holdings swing hard and take turns. Holding the risk you picked is the part you get every time.

When the harvest is worth chasing

It needs three things together: holdings that swing a lot, swings that partly reverse, and costs low enough not to eat the gain. Equity funds in the same category fail the second test, because they rise and fall together. A stock basket rebalanced to equal weight passes more often, because individual stocks take turns leading. That is one reason equal weight books are rebalanced on a short, fixed cadence.

Costs are the brake. Every rebalance that sells equity can create short term gains, taxed at 20 percent for listed equity sold within 12 months, for transfers from 23 Jul 2024. The cheapest rebalance is usually new money pointed at the sleeve that is under target. The rebalance cost calculator prices a trade before you make it, and allocation drift shows the same drift on a 60/40.

The free Learn module on rebalancing strategies compares calendar, band and signal rules on a rupee portfolio, with the tax side of each trade.