What is a SIP investment?

Strip the branding and a SIP is three pieces of plumbing. First, a registration: you name a scheme, an amount, a frequency and a date. Second, a mandate: a NACH instruction (National Automated Clearing House, the NPCI system that carries recurring debits) authorising the fund's collection account to pull that amount from your bank each cycle, up to a ceiling you approve when you sign it. Third, a recurring purchase: each time the debit lands, the money buys units of the scheme at the applicable NAV, and the units sit in your folio exactly as if you had walked in and bought them one by one. Nothing about the units records how they were bought. Two investors holding the same fund, one through a SIP and one through a lump sum, own identical units with identical rights and identical risk.

That is the sense in which a SIP is not a product. It has no NAV, no return, no portfolio and no risk profile of its own; it inherits all four from whatever you point it at, and what a mutual fund is therefore decides everything this page cannot. The choice of instalment date changes almost nothing, and if the chosen date falls on a holiday the purchase simply moves to the next business day. What the schedule contributes is narrower and more valuable than the marketing suggests: it converts a stream of income into a stream of purchases without asking you, twelve times a year, whether now is a good time.

Two registration options extend the same machinery. A perpetual SIP is registered without an end date, so the mandate runs until you cancel it instead of lapsing after a fixed term. A step-up SIP raises the instalment automatically on a set schedule, by a fixed rupee amount or a fixed percentage each year, so the contribution tracks a rising income. The one hard constraint is the mandate ceiling: debits above the amount you originally authorised fail regardless of intent, so a step-up path needs a mandate signed above its final step, not its first.

The date on your SIP is a debit instruction, not an NAV promise. Since 1 February 2021, under SEBI's uniform NAV rules (circulars SEBI/HO/IMD/DF2/CIR/P/2020/175 of 17 September 2020 and SEBI/HO/IMD/DF2/CIR/P/2020/253 of 31 December 2020), units in every scheme are allotted at the closing NAV of the business day on which the money is actually credited to the scheme's bank account before the cut-off time, irrespective of the amount and irrespective of when the order was placed. Until then, purchases under ₹2 lakh received the NAV of the day the application was time-stamped, whether or not the money had arrived; the 2021 change made realisation the test for every rupee, and it applies to each SIP instalment individually.

AMFI's own illustration of the rule uses a ₹5,000 monthly SIP dated the 10th: the investor receives the 10th's NAV only if the instalment reaches the mutual fund's bank account before 3:00 pm that day. If the NACH debit goes out on the 10th but the money settles into the scheme's account on the 11th, the units are priced at the 11th's NAV. The purchase cut-off is 3:00 pm for equity and most other schemes, and earlier, 1:30 pm, for liquid and overnight funds.

Three practical consequences follow. First, the NAV on your statement can differ from the NAV of your SIP date, and that is not an error by the fund. Second, the settlement lag is one more reason the popular hunt for the best SIP date is empty: you cannot fully control which day's NAV you receive, and across years the differences wash out to noise. Third, the rule favours no one by design; it exists so that a buyer whose money has not arrived cannot capture a day's NAV ahead of investors whose money has, which is exactly what the old time-stamp regime permitted at the margin.

Rupee-cost averaging is usually presented as a benefit. It is more honest, and more useful, to present it as a division. A fixed instalment divided by a moving price buys a moving number of units: when the NAV halves, the instalment buys twice the units; when it doubles, half. The buying is automatic and inverse to price, which is the entire content of the phrase "you buy more units when markets are low".

The consequence for your cost is exact, not approximate. After a series of equal instalments, your average cost per unit is total money divided by total units. Total money is the instalment times the number of months. Total units is the instalment divided by the first NAV, plus the instalment divided by the second, and so on. Divide the first quantity by the second and the instalment cancels out: what remains is the harmonic mean of the purchase prices. The average of the prices themselves, the number a chart would show you, is the arithmetic mean. A standard mathematical inequality says the harmonic mean of unequal positive numbers is always below their arithmetic mean, and equal only when every number is identical. A fixed-rupee schedule therefore pays less per unit than the period's average price whenever the NAV moves at all. Not because anyone timed anything: because dividing a fixed numerator by a varying denominator loads the unit count towards the cheap months.

The schedule paid ₹42.71 a unit against an average price of ₹44.00, a gap of ₹1.29, about 3 percent, and the month that did the heavy lifting was the ₹32 month, when the same ₹5,000 collected 156.25 units. Six clean months make the arithmetic visible, but they flatter it. Run the same fixed instalment across a path with the shape a real one has, thirty-six months of uneven legs, false recoveries and a serious drawdown, and the gap is smaller than almost every explainer implies.

Two honest footnotes belong to this arithmetic, and they are the reason the rest of this guide is about behaviour rather than division. The gap is real but modest: 1.56 percent on the path above, single digits for any realistic volatility, and nothing that deserves the word strategy. And it is measured against the average price of the window, never against the final price. Had the NAV ended below ₹102.55, the position would be under water, harmonic mean and all. Averaging tells you that you bought efficiently relative to the path. It says nothing whatever about where the path ends, and it offers no protection at all if the path ends low.

So when does the schedule actually help? Sort every market into three shapes and the answer falls out.

In a sustained rise, the SIP ends behind a lump sum by construction. The first instalment buys at the cheapest price the window will ever offer; each later one buys higher. Run the same six-month arithmetic up a smooth rise from ₹40 to ₹50 and the SIP accumulates 670.55 units at an average cost of ₹44.74; the same ₹30,000 invested on day one would have bought 750 units at ₹40. The staggered buyer paid nearly ₹5 a unit more for arriving gradually. That is not a defect; it is the fee the schedule charges for spreading an entry risk that, in this path, turned out not to bite.

In a choppy, sideways market the harmonic gap is the whole benefit, and the worked table above is exactly this shape: down, bottom, back up. The instalments bought at ₹32 and ₹40 are what dragged the average cost below the average price. A single entry in that market is a lottery on which day you happened to pick; the schedule collects the below-average cost without picking.

In a sustained fall, the honest statement is that the SIP loses money while buying well. Run ₹5,000 monthly down a straight decline from ₹50 to ₹30 and the folio collects 773.05 units at a falling average cost of ₹38.81; at ₹30 those units are worth about ₹23,191 against ₹30,000 paid, a loss of about 23 percent. A lump sum at ₹50 would be down 40 percent, so averaging cushioned the fall; it did not repeal it. Units accumulate while value falls, and every instalment is a fresh purchase into a declining asset. Any explanation that treats "you buy more units when markets fall" as pure good news is describing this third panel and calling it the second.

Set the averaging arithmetic aside for a moment, because at 1.56 percent it is not why the schedule earns its place. A SIP is a behavioural device, and it addresses exactly two of the things that reliably destroy retail outcomes. Neither of them is stock selection, and neither of them is market risk.

The first is timing. Deciding when to buy is the decision retail investors get wrong most consistently, and the reason is structural rather than intellectual. The moment that feels safest to invest is the moment after a long rise, when confidence is high and prices are too; the moment that feels most dangerous is the moment after a fall, when prices are low and every instinct says wait for clarity. Clarity, when it finally arrives, arrives priced in. A fixed date and a fixed amount remove the decision entirely. Not by making it well, but by not making it at all, which for a decision most people get backwards is the stronger outcome.

The second is consistency. Of everything an ordinary investor controls, the amount invested and the number of years it stays invested dominate; and both are functions of whether the contributions actually keep arriving. An auto-debit converts an intention that must survive twelve monthly renewals of willpower into a default that must survive one act of cancellation. Defaults win. That is the whole behavioural claim, and it is a real one, but note how narrow it is: the schedule improves the odds you finish, not the return you earn while finishing.

Read the right-hand column and the honest shape of the product appears. A SIP fixes one problem outright, half-fixes a second, and leaves three untouched. That is still a good trade, because the two it reaches are the two that most often decide whether a retail investor is still invested in year seven. But it is a much smaller claim than the one usually made for it, and the half-fixed row is where almost all of the damage happens. Our guide to the psychology of market decisions works through why the instinct to stop arrives precisely when stopping is most expensive.

Everything above describes a schedule that runs. The interesting question is what happens to the one that does not, because that is the common case and it is where the mechanism inverts. Take the same illustrative 36 month path, and give the investor the reaction almost everyone reports having: at month 16, five months into a decline that is now 22 percent deep and showing no sign of ending, they pause the SIP. Not out of panic, and not to redeem. They pause it sensibly, to wait for things to settle, and they resume at month 25 once the recovery is visible.

Nine instalments are skipped. They happen to be the nine cheapest months of the entire path.

The lines are identical until month 16 and never converge again. The investor who kept buying finishes with 1,755.22 units; the investor who paused finishes with 1,240.49, having forfeited 514.73 units, 29 percent of the total, all of them at prices below their own average cost. Because both committed the same money, the cleanest way to read the damage is per unit: the continuing investor paid an average of ₹102.55, the pausing investor ₹108.83. The pause raised the cost per unit by 6.12 percent.

Set that number beside the one from the previous section and the argument closes itself. Rupee-cost averaging contributed 1.56 percent. The pause cost 6.12 percent. A single nine-month pause, taken for entirely reasonable-sounding reasons, destroyed roughly four times the benefit the averaging mechanism had spent three years accumulating. The schedule's whole advantage lives in the months an investor least wants to fund it.

The third row is the one worth dwelling on, because pausing and exiting are separated by a much thinner membrane than investors expect. Cancelling a SIP stops future debits and does nothing else; the units already bought stay invested. But the state of mind that cancels the debit in month 16 is the same state of mind that redeems the folio in month 22, and the second act is the one that converts a decline into a permanent loss. That investor paid ₹111.14 a unit, sold at ₹78.60, and never returned. Nothing in the mechanism failed. The mechanism was switched off.

The comparison that follows from the map is often posed as a duel, and is better posed as a question about the shape of your money. If a large amount already exists, the mechanical facts are these: in any window where the NAV rises more or less steadily, the lump sum finishes ahead, because every later instalment pays more than the first; in any window where it falls, the schedule loses less. Across long historical windows, studies comparing the two in Indian and global equity markets have generally found the lump sum finishing ahead more often than the staggered entry, for the unglamorous reason that markets have spent more time rising than falling, and instalments still waiting their turn are capital out of the market while the trend runs. Stated carefully: spreading the entry is not a return strategy; it is protection against a bad first day, paid for by sitting partly in cash while the market decides.

For most people the duel never arises. Income arrives monthly, so the realistic alternative to a SIP is not a lump sum; it is irregular manual purchases, or cash accumulating in a savings account until courage and free time coincide. Against those alternatives the SIP is simply the version of you that never forgets, never hesitates, and never waits for a clarity that does not come.

The last row is the honest core of the whole product. The averaging arithmetic is real but small; the allotment rules are mechanical; what a SIP mostly buys is the removal of twelve timing decisions a year from a person who would otherwise take them under stress.

Because the SIP is a schedule, everything about changing it is administrative, not financial. Miss an instalment and the fund house levies no penalty, deducts nothing from your folio, and leaves your existing units untouched; that month's purchase simply does not happen. The cost of a failed debit sits on the banking side: if the NACH mandate bounces for insufficient balance, the bank can charge a mandate-dishonour fee, a few hundred rupees at many banks, plus tax, per bounce. Fail repeatedly and the registration itself lapses: most fund houses cancel a SIP automatically after around three consecutive failed instalments, after which continuing means registering afresh. The units bought so far are unaffected throughout.

Pausing is a formal facility at most fund houses and platforms, typically for a set number of instalments, and cancelling is free and takes effect within a processing cycle. What cancellation does is narrower than most investors assume: it stops future debits, nothing else. The units you already hold stay invested and keep moving with the NAV until you place a separate redemption, which is its own decision with its own possible exit load and tax consequences. Stopping the purchase and exiting the investment are two different acts, and conflating them, usually in a falling market, is how a pause becomes an accidental exit.

Changing the amount is a registration change: platforms typically modify the registration, or cancel and re-register, and a step-up SIP handles the common case automatically. The constraint to plan for is the mandate ceiling discussed earlier: a step-up that will one day debit ₹15,000 needs a mandate authorised at or above ₹15,000 on day one, not at the opening instalment.

The table assumes nothing about markets; it is the contribution schedule alone. A 10 percent annual step-up puts 59 percent more capital to work across ten years than a frozen instalment does, purely because the instalment tracked an income that was itself rising. For an investor whose salary grows, leaving the SIP amount at its starting figure is a slow, silent reduction in real saving, and it is the most commonly neglected lever on the whole machine.

One consequence of the schedule confuses more investors than any other, and it is a direct result of the fact that a SIP is not a single purchase. Every instalment starts its own holding period. The registration date means nothing to the tax code. What matters is the age of each individual tranche of units on the day you sell, and on redemption from an equity-oriented scheme, units are drawn on a first-in-first-out basis: the oldest units leave first.

So a SIP that has been running for three years is not a three-year-old holding. It is 36 holdings of 36 different ages, and only the ones older than twelve months have reached long-term status.

On the illustrative schedule above, a redemption in month 36 finds 24 of the 36 instalments past the twelve-month threshold. By units, because the cheap months of the decline bought disproportionately many, that works out to 1,189.54 units at the long-term rate against 565.68 still short-term: 67.8 percent and 32.2 percent. An investor who assumed a three-year SIP was uniformly long-term would have mis-estimated the tax on nearly a third of the position.

The rates themselves belong to the fund, not to the schedule. As of 18 July 2026, gains on an equity-oriented scheme, meaning one holding at least 65 percent in Indian equity, are long-term after twelve months and taxed at 12.5 percent with total long-term gains up to ₹1.25 lakh a year exempt, while gains on units held twelve months or less are short-term and taxed at 20 percent. The framework covering debt and hybrid schemes belongs to the fund rather than to the schedule, and the taxation guide works the underlying classification through in depth.

Two practical consequences follow. First, an exit load, where a scheme levies one, is applied the same way, tranche by tranche, so a redemption soon after a long-running SIP can attract a load on the newest units while the oldest escape it. Second, the per-instalment clock is a quiet argument for not redeeming in a lump: a partial redemption drawn from the oldest units is treated very differently from one that reaches into the last twelve months of purchases.

A mechanism is best understood by the ways it fails. These are the six that actually occur, in rough order of how much they cost.

Five of the six are administrative or conceptual and cost little once understood. The first one is different in kind, and it is the reason this guide spends more space on behaviour than on arithmetic.

Read plainly, a SIP automates the least important decision in investing, when to buy, and leaves the most important one, what to own, entirely to you. The schedule will execute a monthly purchase into a well-chosen fund and into a mediocre one with identical discipline; it audits nothing, compares nothing, and notices nothing. Its arithmetic gives you the harmonic mean of the prices you encounter, worth a little over one percent on the path modelled here. Its mandate machinery gives you allotment at the NAV of realisation. Its psychology gives you distance from your own reactions, and that last one is worth several times the other two combined.

None of which touches the question of whether the scheme you chose deserves the next five years of your income. That question is upstream of the schedule and unaffected by it, and it is where the actual work sits: what the fund holds, what it charges, whether a low-cost index vehicle such as an exchange traded fund would do the same job for less, and what evidence would make you change your mind. That upstream judgement, why an investment should earn anything and what would falsify the case, is exactly what the method we teach is built around. Point a good schedule at a good decision and the machinery disappears into the background, which is where machinery belongs.

• SEBI, uniform NAV upon realisation of funds. Circulars SEBI/HO/IMD/DF2/CIR/P/2020/175 (17 September 2020) and SEBI/HO/IMD/DF2/CIR/P/2020/253 (31 December 2020): from 1 February 2021, purchase units in all schemes, including every SIP instalment irrespective of amount, are allotted at the closing NAV of the day funds are credited to the scheme's bank account before cut-off.
• AMFI, cut-off timings and the rule on applicable NAV. The industry body's explainer of the February 2021 change, including the worked illustration of a ₹5,000 SIP dated the 10th receiving that day's NAV only if the money reaches the fund before 3:00 pm. amfiindia.com
• AMFI monthly data, May 2026. SIP contribution of ₹30,954 crore for the month, 16 percent higher than a year earlier and the third consecutive month above ₹30,000 crore, from about 9.6 crore contributing SIP accounts; SIP-linked assets near ₹17.1 lakh crore, roughly a fifth of industry assets. amfiindia.com
• Mandate and instalment mechanics. NACH auto-debit mandates run on NPCI rails; fund houses levy no penalty for a missed instalment, banks may charge for a dishonoured debit, and most fund houses cancel a registration after about three consecutive failures. Practice as documented by fund houses and registrars; thresholds vary.
• Capital gains on equity-oriented mutual fund units, as of 18 July 2026. Following the July 2024 budget, gains on units held more than twelve months are long term and taxed at 12.5 percent with total long-term gains up to ₹1.25 lakh a year exempt; units held twelve months or less are short term and taxed at 20 percent. Each SIP instalment carries its own holding period and redemption is first-in-first-out. Rates and thresholds change from budget to budget. incometaxindia.gov.in
• The computed figures on this page. The 36 month NAV path, the units, the average cost and average price, the paused-schedule comparison and the holding-period split are all computed from one illustrative, hand-constructed price path stated in full in the figures. It is a model chosen for legibility, not a forecast, not historical data, and not a claim about any scheme.