XIRR calculator: the real annual return on irregular cashflows

Why absolute return and CAGR both fail a SIP

Suppose you started a SIP three years ago. You have put in ₹3.6 lakh over thirty-six instalments and the folio is worth ₹4.6 lakh today. What return did you earn?

The tempting answer is the absolute one: you are up ₹1 lakh on ₹3.6 lakh, so about 28%. That number is arithmetically correct and almost useless, because it silently assumes all ₹3.6 lakh was invested for the same length of time. It was not. The first instalment has been working for three years. Last month's instalment has been working for four weeks. Most of your money has been invested for far less than the full period, which means a 28% total gain represents a much healthier yearly rate than 28% divided by three.

CAGR does not rescue you either. Compound annual growth rate takes one amount, one start date, one end date, and works out the smooth yearly rate connecting them. It is the right tool for a lump sum you bought once and never touched. Feed it a SIP and it has nowhere to put the other thirty-five dates — you would have to pretend the whole ₹3.6 lakh went in on day one, which understates your return badly, or on the average date, which is a fudge.

XIRR is the general case that both of those are special cases of. It asks: what single annual rate, applied to each cashflow for exactly as long as that cashflow was invested, makes everything balance? There is no closed-form answer, so it has to be solved by iteration — the calculator above uses Newton-Raphson with a bisection fallback, the same kind of method a spreadsheet's XIRR function runs, and the same code the Klera app uses.

XIRR, CAGR and absolute return, side by side

Take one concrete case. You invest ₹1,00,000 on 1 January 2023 and another ₹1,00,000 on 1 January 2025. On 1 January 2026 the holding is worth ₹2,60,000.

Absolute return is ₹60,000 on ₹2,00,000 — 30%. Over three elapsed years that reads like roughly 10% a year, and it is wrong, because the second lakh was only invested for one of those years. CAGR on the total, treating ₹2,00,000 as if it all went in at the start, gives about 9.1% — also wrong, and wrong in the same direction. XIRR gets roughly 13.5%, because it charges the first lakh three years of growth and the second lakh only one.

The gap between those numbers is not a rounding difference. It is the entire question of whether your investments did well. And it explains why the "returns" figure your broking app shows can differ from the one on your fund house's statement: one may be quoting point-to-point NAV growth for the scheme, another your personal XIRR including the exact dates you bought. Both can be correct and mean different things. Before comparing two numbers, check that they are the same kind of number over the same dates.

Getting your own cashflows out of a statement

You need two columns: a date and an amount. For mutual funds, the transaction statement from the AMC or the consolidated account statement from CAMS or KFintech lists every purchase, SIP instalment, switch and redemption with its date and rupee value. That is your raw material. For stocks, the contract notes or the ledger from your broker do the same job.

The rule is simple. Every purchase is money leaving you, so it is a negative flow. Every redemption or dividend actually paid out is money coming back, so it is positive. Then add one final row dated today for what the holding is worth right now, also positive — this is the row people forget, and without it there is nothing for the purchases to be measured against. In the calculator above you never type a minus sign; the dropdown on each row marks it Invested or Received and the sign follows from that.

Two practical shortcuts. If you ran a plain monthly SIP with no changes, you do not need every instalment to get a usable answer — a handful of representative rows plus the closing value will land close, though the exact figure needs the exact dates. And if a fund was switched rather than redeemed, treat the switch-out as a receipt in the old scheme and a fresh purchase in the new one, otherwise the two folios contaminate each other.

Where XIRR calculations go wrong

The sign convention is the most common failure. Enter everything as positive and there is no rate that balances the flows, so a spreadsheet returns an error and a careless script returns nonsense. Get it exactly backwards — purchases positive, value negative — and you will often get a plausible-looking number with the wrong meaning attached.

Forgetting the closing value is the second. A list of purchases with no current value is all money in and no money out; there is nothing to solve for. The calculator above says so explicitly rather than showing a blank.

Then there is mixing units — some rows in rupees, some in thousands or in units of the fund rather than money — which produces a confidently precise and completely fictional rate. Using XIRR where CAGR would do is harmless but noisy: for a single lump sum held untouched, the two agree, and CAGR is easier to explain. And comparing XIRR figures across different time windows is the subtlest mistake of all. A 19% XIRR measured over eighteen months and a 12% XIRR measured over nine years are not rankable. The first one has barely been tested.

What XIRR does not tell you

XIRR is a summary of what happened to your money. It is not a summary of how sensibly it was invested, and it is worth being blunt about the gap.

It says nothing about risk. Two portfolios can post the same 14% XIRR while one drifted upward steadily and the other halved in the middle and recovered. If you would have panicked and sold during the second one, the identical XIRR describes an experience you could not have lived through. It says nothing about volatility, drawdown, or concentration.

It also cannot separate skill from conditions. A great XIRR earned entirely inside a bull run tells you the market rose, and very little about whether the fund manager added anything — which is why comparing against the scheme's own benchmark over the exact same dates is more informative than the raw number. And it says nothing about tax: XIRR is a pre-tax figure, and equity and debt holdings are not taxed alike, so two identical XIRRs can leave you with different amounts of money.

Finally, it is a description of the past. A three-year XIRR is a fact about three particular years that will not repeat. Use it to check whether a holding is doing what you bought it for, not to project what it will do next.

Doing this on your own portfolio

Rebuilding a cashflow table by hand for every folio gets old quickly. Klera computes XIRR on your tracked investments automatically, offline on your phone — it holds mutual funds, stocks, FDs, RDs, PPF, EPF, NPS, gold, real estate and crypto as holdings, and once your purchases are recorded the annualised return is derived from them rather than typed in. It is the same solver as the calculator on this page, ported from the app's own code. There is no account, no cloud sync and no server that ever sees your holdings, and the app does not pull live market prices for every asset class or calculate your tax — you update valuations yourself, and the maths runs on your device. Get Klera free on Android if you want the returns tracked rather than recalculated by hand each quarter.