Capital gains tax, and the calendar that decides it

What the holding period actually decides

How long you held an asset before selling it decides which of two tax treatments applies to the same gain. For listed equity and equity-oriented funds the line sits at 12 months. For one large category of funds, the line no longer exists at all.

Nothing else about the transaction changes. Same scheme, same purchase price, same sale price, same profit in rupees. Cross the line and the gain is long-term; fall short of it and the identical gain is short-term, and the two are charged under different sections at different rates.

That is unusual, and worth pausing on. Most of the tax code responds to how much you earned. This part responds to when — and “when” is the one variable in the whole calculation that you control completely and for free.

The citations moved recently, so it is worth being precise about them. The Income-tax Act 1961 was repealed with effect from 1 April 2026 and replaced by the Income-tax Act 2025. The rates barely moved; the section numbers all did. Short-term gains on listed equity now sit in s.196 (formerly s.111A), long-term gains in s.198 (formerly s.112A), the holding-period definition in s.2(101), and the special rule for debt-oriented funds in s.76 (formerly s.50AA). If you are filing for FY 2025-26 — April 2025 to March 2026 — that year is still governed by the old Act, which is why both numbers appear here.

The line sits in a different place for each asset

The single most common error is to learn one holding period and apply it everywhere. There is no general rule — there is a rule per asset class, and one class where the rule was withdrawn.

Two things about that table are worth saying out loud. The blanks are deliberate: a plausible-looking rate copied from a stale article is exactly the defect this library exists to avoid, so where we have not verified a figure against the 2025 Act we leave the cell empty and list the question at the bottom of our own file. And the annual threshold of ₹1.25 lakh is annual and aggregate — it applies once across all your qualifying long-term gains for the financial year, not once per scheme and not once per sale.

The two tests are not opposites, and a fund can fail both

Here is the part that even careful investors get wrong, because the two tests read like a pair when they are not.

A fund is equity-oriented under s.198 if it holds more than 65% of its total proceeds in domestic equity shares. A fund is a Specified Mutual Fund under s.76 if it holds a mirror-image proportion in debt and money-market instruments — the two thresholds are set at the same level, pointing in opposite directions.

Set at the same level, they leave a gap. A scheme holding, say, half its book in equity and a third in debt satisfies neither test. It is not equity-oriented, so the 12.5% route and the annual threshold do not apply. It is not a Specified Mutual Fund, so s.76 does not apply either. It falls into ordinary capital-gains treatment, which is a third answer, not a rounding of the other two.

The gap does not depend on getting the debt-side figure exactly right, which is worth saying because we have flagged that figure as unverified. A book that is half equity is not majority-equity on any reading, and it is not majority-debt either. Something has to tax it, and that something is neither of the two routes most articles describe.

Now the part that is genuinely non-obvious, and it is about how fund houses respond. Several hybrid categories are built so that their gross holding of domestic equity shares — the directly held stocks plus the cash leg of arbitrage and hedged positions — stays above the equity-oriented threshold, even while the scheme's net exposure to the market is far lower. The result is a fund that is defensive in substance and equity-oriented in law at the same time. That is industry practice rather than a rule, and we have listed the underlying question — whether the test reads gross holdings or net exposure — in our open questions. But it explains something otherwise puzzling: why a conservative-sounding scheme insists on a stock book it then hedges away.

Schemes that do not do this can land in the middle band. Multi-asset allocation funds holding equity, debt and gold in stated proportions are the common case, and a scheme can move across the line as the manager shifts allocation. The category name on the factsheet is a SEBI classification; the tax test is an Income-tax Act arithmetic test on what the scheme actually holds. They are different questions and they do not always agree.

The practical consequence: for any hybrid scheme, the tax treatment is a fact about the portfolio, not about the label — and it is the one fact a scheme's marketing is least likely to lead with. In practice the scheme information document is where a fund house states the treatment it expects the scheme to attract; whether SEBI requires that disclosure is an open question on our list. That statement, not the category name, is the thing to read, and it is the reason two hybrids sitting side by side in the same shortlist can be taxed differently.

Where the calendar stopped mattering

Until 1 April 2023, holding a debt fund past a threshold moved it to a gentler treatment. Then the rule was withdrawn for a defined class of funds, and it is that withdrawal which s.76 of the 2025 Act carries forward.

For units of a Specified Mutual Fund acquired on or after that date, the gain is taxed at slab rates, with the gain always treated as short-term. Held 11 months or 11 years, the answer is the same. The lever this whole article is about does not exist for that asset.

Which sounds like the end of the story and is not, because removing the holding-period lever leaves the other one untouched. A debt fund is still taxed only when you redeem, while interest on a deposit is taxed as it accrues whether you touch it or not. The rate is the same; the timing is not, and timing is what the next section is about. The comparison is worked through in full in debt funds against fixed deposits.

The trade-off, stated plainly: s.76 took away the reward for patience in that category. What survives is the reward for not churning — and those are not the same thing, which is why the change was smaller in practice than the coverage suggested.

Why deferral compounds, and what it actually costs to realise

Realising a gain does not cost you the tax — the tax falls due either way. It costs you the return on the tax for every year between paying it early and having to. That mechanism survives every rate change the next twenty Budgets can produce, so it is worth seeing the arithmetic once.

Take two portfolios, each starting at ₹10 lakh and each earning 10% a year for 20 years. The 10% is an assumption for the arithmetic, not a projection of anything. One never sells. The other sells everything each year, pays tax on that year's gain, and buys the identical holdings back the next morning — same securities, same return, only the calendar differs.

Suppose tax takes a tenth of every gain you realise. A tenth is not any Indian rate; it is a round number chosen so you can check the arithmetic without a calculator.

• The one who sells every year compounds at 10% less the tenth taken out of each year's gain — that is 9% net. Over 20 years, ₹10 lakh at 9% becomes about ₹56 lakh, and nothing further is owed.
• The one who never sells compounds at the full 10%. ₹10 lakh becomes about ₹67.3 lakh. Sell on the last day and a tenth of the ₹57.3 lakh gain is due — about ₹5.7 lakh — leaving roughly ₹61.5 lakh.

Same return, same tax fraction, same securities. The gap is about ₹5.5 lakh, and every rupee of it is compounding on money that would otherwise have left the base early.

That ₹5.5 lakh is the return on the tax and nothing else — the same tax fraction was paid in both columns. That is why the damage from frequent realisation scales with two things and only two: how many times you realise, and how many years remain for the removed money to have compounded. Realise once, at the end, and there is no gap at all.

The same arithmetic in a different frame is the subject of the time value of money — a rupee that stays in the base is not worth the same as a rupee that leaves it.

And the trade-off, because there always is one. Deferral is a reason not to churn. It is not a reason to keep an asset you would otherwise sell. An investor holding a deteriorating position purely to postpone a tax bill has let a fraction of the gain make the decision about the whole of it, which is the more expensive error in almost every case. Deferral is a tiebreaker, not a thesis.

A switch is a sale, even when nothing appears to leave

Now the failure mode that costs people money quietly, because on the statement it looks like nothing happened.

Moving from one scheme to another is not a transfer of units. Mechanically, the registrar redeems your units in the first scheme at that day's NAV and allots fresh units in the second. A redemption is a transfer. A transfer of a capital asset is a taxable event. The money never reached your bank account, and that is irrelevant — nothing in the charging section is keyed to your bank account.

Everything in this list is the same event:

• Switching between two schemes at the same fund house.
• Moving from the regular plan to the direct plan of the same scheme — a lower expense ratio by construction, and still a redemption on the day you make it.
• Switching from growth to the income-distribution option, or back.
• Each instalment of a systematic transfer plan, which is a small switch on a schedule.
• Each withdrawal under a systematic withdrawal plan, which redeems units to raise the cash.
• Every sell leg of a rebalance.

Two consequences follow, and the second is the one people miss. The gain to that date crystallises, so tax may fall due for a year in which you felt you did nothing. And the new units start a fresh holding period from zero. A holding you had patiently carried for 11 months, switched for a good reason, is one day old again.

None of that makes switching wrong, and the comparison is arithmetic you can actually do. On one side, the expense-ratio saving, which recurs every year for as long as you hold. On the other, the tax crystallised today plus any exit load, which is paid once. A recurring saving against a one-off cost resolves in favour of the switch as the remaining horizon lengthens, and against it as the horizon shortens — which is why the identical switch pays for itself several times over across 20 more years of holding, and does not across three more months. Run it on your own expense ratios and your own unrealised gain; the general shape does not settle any particular case. The error is not switching. The error is switching while believing it was a non-event.

A SIP is not one holding, it is many

The second recurring error, and it comes from an entirely reasonable intuition.

“I have been running this SIP for three years, so the whole thing is long-term.” It is not. Each instalment is a separate acquisition on its own date with its own clock. A 3-year SIP redeemed in full today contains units of every age from one day to three years, and the tax computation runs lot by lot, not on the total.

Redeem only part of it and a second question appears: which units were sold? The convention in practice is that the oldest units go first, which for a SIP means the units most likely to have crossed the line. We have flagged the statutory basis for that convention as an open question rather than asserting it here — but the shape of the point stands either way, because the answer determines the tax and it is not something you choose at redemption time.

The check that settles it is unglamorous and reliable: pull the statement of account from the registrar or the consolidated account statement, and read the dated list of purchases. That list, not the SIP start date, is what the holding-period rule operates on. The same is true of every index fund or ETF accumulated in pieces.

Which regime you are in, and what it does and does not change

Almost every Indian tax article written before 2026 assumes the reader is in the old regime. Most readers are not, because the new regime is now the default under s.202 — you are in it unless you actively opt out. The full comparison is its own article; what matters here is the narrower question of what a regime does to a capital gain.

The answer splits in two, and the split is not where most readers expect it.

The distinction to hold on to: the regime governs what you may deduct from ordinary income and what your slab table looks like. The holding period governs how a capital gain is classified. Those are separate machinery, and choosing a regime does not move the line at 12 months.

The last row is the one worth re-reading, because it is the exception that most regime-versus-regime comparisons omit. Once s.76 pushed debt-fund gains onto the slab table, those gains stopped being insulated from the regime choice. Equity gains under s.196 and s.198 carry their own flat rates and are unaffected. So the regime question does reach your portfolio — but only through the holdings whose gains are taxed as ordinary income, which is exactly the set the holding-period lever was taken away from. One change took the calendar lever off those gains and left them exposed to the regime lever instead. We have flagged both halves of this — that the flat rates ignore the regime, and that slab-taxed gains do not — in our open questions rather than treating them as settled, so read it as the shape of the thing and check it against your own slab before it decides anything.

Where deductions meet all this is in why people buy certain things. An equity-linked savings scheme is sold on the s.123 deduction, and that deduction is old-regime only. Its three-year lock-in, meanwhile, is a SEBI condition on the scheme and applies to everyone who holds it, deduction or no deduction. A reader in the default new regime who buys an ELSS is buying an equity fund with a lock-in and no deduction — which may still be a reasonable thing to own, but for entirely different reasons than the ones in the sales pitch.

Working out which regime leaves you better off is arithmetic on your own numbers: total the deductions you would actually claim under the old regime, and compare the tax on the two computations. This article does not tell you which to pick, and nobody who has not seen your figures can.

What this article is deliberately not telling you

Two things a reader arriving here may want, which we are not going to state.

Loss set-off and carry-forward. The Act allows capital losses to be set against gains, with restrictions on which class of loss may offset which class of gain, and allows unused losses to be carried forward for a limited number of years. Those restrictions and that number are exactly the kind of detail that every section renumbering breaks, and we have not yet verified them against the 2025 Act. So they are in our open questions rather than in this paragraph. Read the Act or ask someone who has.

Rates for assets outside listed equity. Property, gold, unlisted shares and offshore holdings each have their own holding period and their own rate, and the 2025 Act changed the numbering for all of them. None of those figures appears above, for the same reason.

This is not throat-clearing. A stale rate stated confidently is worse than a blank, because the reader cannot tell it is stale. The blanks are the honest version.

Working it out on your own numbers

The holding-period question is answered by dates, not by opinion: the date each lot was bought, the date it was sold, and which side of the line that puts it on. The registrar's dated statement of account is the primary document, and it is free.

For the prior question — whether a given holding period was long enough to be worth having in the first place — FNOTrader's Mutual Funds app runs SIP and lumpsum on the full AMFI NAV history, around 34 million NAV rows, and reports XIRR, invested against value, maximum drawdown and rolling-return distributions over the horizon you specify. A SIP simulation is, by construction, a dated list of instalments, which is the same shape as the lot register the holding-period rule operates on.

It does not compute your tax, and no tool can do that from NAV data alone — the answer depends on your own purchase dates, your regime and the rest of your return. FNOTrader is not a tax adviser and none of the above is advice on your position.

Common questions

Continue reading

More in Tax Planning · App: Mutual Funds · Definitions: glossary · Free tools: calculators · All: every article