Rebalancing a concentrated book

You built a portfolio on purpose. You decided how much to put in each position — that decision was your real control over risk — and for a while the book looks exactly as you designed it. Then prices move. Winners swell, laggards shrink, and without you buying or selling a single share, the portfolio slowly becomes a different portfolio from the one you chose. The bet you sized at a fifth of the book is now a third; the one you sized at a fifth is now a twentieth.

The question this module asks is: how do you pull the book back toward the shape you chose — without, in the process, wrecking the very theses that made it grow? That "without" is the whole difficulty. The crude version of rebalancing — always sell what went up, always buy what went down — is easy to state and quietly destructive, because it can mean selling your best businesses simply for the crime of doing well. Good rebalancing is a more careful thing: keeping your risk where you chose it, while leaving your winners room to keep winning.

When you build the book you assign each holding a target weight — the share of the portfolio you decided that position deserves, chosen from your conviction and how much risk you were willing to carry in it. Say five positions at 20% each of a ₹10,00,000 book. That number was never arbitrary; it was your answer to "how much of my money should one surprise in this name be allowed to move?"

Prices then do what prices do, and the weights drift — wander away from their targets as some holdings rise and others fall. This matters for one reason above all: drift silently changes your risk. A holding that grows from 20% to 38% of the book has nearly doubled how much a single bad surprise in that one name can hurt you. You did not decide to take that larger risk; the market handed it to you, and doing nothing is a decision to accept it.

This is where the free lunch of the earlier modules comes back. Spreading risk across positions only keeps protecting you if the spread is maintained — a book that has drifted into one giant holding has quietly given the free lunch back. Rebalancing is the maintenance that keeps the diversification you paid for from decaying. It is not about squeezing out more return; it is about making sure the risk you are carrying is still the risk you chose.

The naive method is to reset every position to its exact target whenever it moves. This is a mistake, and an expensive one. In India especially, every trim of a gain can trigger capital gains tax — a tax on the profit when you sell — and every trade carries brokerage and charges. Chasing an exact 20% on every wiggle produces a stream of small taxable trades that eat returns while adding almost no real risk control, because a position at 22% is not meaningfully more dangerous than one at 20%.

The better method is band rebalancing — doing nothing until a position drifts outside a pre-set band around its target, and only then trimming or topping up. Set a band of, say, target ±5 percentage points: a 20% position is left entirely alone between 15% and 25%, and you act only when it pokes above the upper band or below the lower one. This cuts the number of trades dramatically while still catching the drifts big enough to matter for risk. It also removes you from the decision in the heat of the moment — the band was set in a calm hour, and it simply tells you when to look.

There is a gentler alternative that avoids selling altogether: rebalance with new money. If you are adding savings to the book anyway, direct the fresh cash to the positions that have fallen below their band, topping up the laggards without trimming the winners at all. This sidesteps the tax on gains entirely, because you are only buying. For a portfolio that is still growing through regular contributions, this is often the cleanest way to hold your shape.

Walk a real drift. illustrative You started with five positions at 20% each of ₹10,00,000 — ₹2,00,000 apiece. A strong year later the book is worth ₹12,50,000, but not evenly: your best holding, A, has run to ₹4,37,500 — 35% of the book — while a laggard, B, has slipped to ₹1,50,000, just 12%. With a ±5 point band, A has breached the 25% upper band and B has fallen through the 15% lower band. The other three sit quietly inside the band and you leave them completely alone.

Now the careful part, and it is not automatic. Before trimming A, you ask the question the rule cannot answer for you: is A's thesis intact? If the reasons you bought it are stronger than ever, you do not blindly dump it back to 20% — you trim enough to bring the concentration back to something you can bear, perhaps to the top of the band rather than all the way to target, keeping most of a winner you still believe in. If A's thesis has quietly broken, this is no longer a rebalance at all; it is the beginning of an exit, and it should not stop at 20%. The same 35% weight calls for two completely different trades depending on the reason.

This is the tension every thoughtful investor feels here, and it is worth naming honestly. Mechanical rebalancing says "sell what rose"; the wisdom of letting winners compound says the opposite. The resolution is that you trim a winner only to control concentration risk, never as a reflex to punish it for winning — and you top up a laggard only if its thesis still holds, never merely because it is cheaper than before. The band tells you when to look; the thesis tells you what to do when you look. Obey the first blindly and you will sell your best businesses; ignore the first entirely and you will let one holding grow into a risk you never chose.

Rebalancing is a maintenance tool, and it is easy to ask it for things it cannot give.

It cannot tell you whether a holding is good or bad. It works purely on weights — how big each position has become — and knows nothing about the businesses. That is why the thesis check has to sit beside it: the band decides when a position's size needs attention, never whether the position deserves to exist.

It cannot improve a portfolio of poor holdings. Rebalancing keeps the risk you chose where you chose it; if what you chose was weak, tidier weights just give you a well-organised weak portfolio. It is a discipline for maintaining a sound book, not a rescue for an unsound one.

And it is not free, so it should not be done for its own sake. Every trim of a gain can trigger tax, and every trade carries a cost. The friction is the reason for bands and for using new money rather than sales — and the reason that "am I rebalancing to control a real risk, or just to make the numbers tidy?" is always worth asking before you place the trade.

The same traps catch people maintaining a book.

1. Rebalancing on a hair-trigger. Resetting to an exact target on every wiggle piles up small taxable trades that cost real money and control almost no additional risk. Bands exist precisely to stop this.

2. Trimming winners mechanically. "Always sell what rose" quietly sells your best businesses for succeeding. The only honest reason to trim a winner is that its size has become a concentration risk you did not choose — not that it went up.

3. Topping up losers by reflex. "Always buy what fell" can pour good money into a holding whose thesis has broken. Add to a laggard only if the reasons to own it still hold, never merely because it is cheaper.

4. Forgetting tax and cost. In India a trimmed gain can be taxed and every trade carries charges. Ignoring this friction is how rebalancing silently loses more than the risk it controls is worth — which is why funding top-ups with new money is often the cleanest route.

5. Confusing a rebalance with a sell. Trimming an intact winner for risk and exiting a broken business are opposite acts that can look identical on the screen. The weight tells you to look; only the thesis tells you which one you are actually doing.