ToolsSIP check
Did it matter when you started?
A great deal, over ten years. A Nifty 500 SIP started in 1997 and held ten years returned a median 25.3% a year after costs and tax; one started in 2010, 9.3%. Over twenty years the start mattered far less: every start year from 1995 to 2006 landed between 11.9% and 15.2%. And the 1990s had the longest wait: a SIP begun in Aug 1999 spent 3 years 2 months below the money put in.
Written for SIP investors wondering whether their start date was lucky or unlucky. Data to 6 Oct 2026.
By start year
Each cell is the median after-tax XIRR of the SIPs started in that calendar year's months and held the given number of years. Everything here is measured: published total-return indices, a G-sec index rebuilt from market yields, and RBI's domestic gold. The mixes start in 1996, when the G-sec yield series begins; the Nifty 50 in 1999.
Ten-year SIPs, by the year they started
Median after-tax XIRR, per cent a year. Darker is higher; colour saturates at 25%
How to read it. Read along a row. The Nifty 500's best ten-year start years were 1996, 1997, 1998: SIPs that bought through the slumps of the late 1990s and early 2000s and sold into the boom that followed. Its worst were 2003, 2006, 2010. Gold's best were 2001, 2002, 2016. G-sec moved least: 5.2% to 9.3%.
Ten-year SIPs, by the year they started. Heat table, 7 rows by 22 columns. Values from 3 to 25.
Twenty-year SIPs, by the year they started
Median after-tax XIRR, per cent a year; start years 1995 to 2006
How to read it. Twenty years smooth out most of the luck of the start. The spread across start years shrinks to a few points for every portfolio, though twenty-year windows from 1995 add up to a single independent period, so this is one history's evidence.
Twenty-year SIPs, by the year they started. Heat table, 7 rows by 12 columns. Values from 7 to 15.
After-tax XIRR across all start months from each portfolio's first. Read 'independent' before the percentiles: overlapping windows are few independent pieces of evidence.
| Portfolio | Years | Windows | Independent | 5th pct | Median | 95th pct | Worst (start) | Below FD | More |
|---|---|---|---|---|---|---|---|---|---|
| Nifty 500 | 5 | 321 | 6 | −1.7% | 14.2% | 39.5% | −9.9% (1996-10) | 17% | |
| Nifty 500 | 10 | 261 | 3 | 9.3% | 13.8% | 24.9% | 3.4% (2010-04) | 0% | |
| Nifty 500 | 15 | 201 | 2 | 10.5% | 14.0% | 17.5% | 6.1% (2005-04) | 0% | |
| Nifty 500 | 20 | 141 | 1 | 11.9% | 14.0% | 15.4% | 10.7% (2000-04) | 0% | |
| Nifty 50 | 5 | 268 | 5 | 4.3% | 13.3% | 36.9% | −4.7% (2015-04) | 8% | |
| Nifty 50 | 10 | 208 | 2 | 8.8% | 12.6% | 19.8% | 3.4% (2010-04) | 0% | |
| Nifty 50 | 15 | 148 | 1 | 9.6% | 12.5% | 15.5% | 6.2% (2005-04) | 0% | |
| Nifty 50 | 20 | 88 | 1 | 10.7% | 12.7% | 14.1% | 10.0% (2006-10) | 0% | |
| 60/20/20 | 5 | 305 | 6 | 4.7% | 12.4% | 26.9% | −0.6% (1996-10) | 8% | |
| 60/20/20 | 10 | 245 | 3 | 9.3% | 12.8% | 20.7% | 5.6% (2010-04) | 0% | |
| 60/20/20 | 15 | 185 | 2 | 10.3% | 12.5% | 16.4% | 8.1% (2005-04) | 0% | |
| 60/20/20 | 20 | 125 | 1 | 11.8% | 13.0% | 14.6% | 11.4% (2006-10) | 0% | |
| 60/40 | 5 | 305 | 6 | 5.5% | 11.8% | 25.2% | 0.6% (2015-04) | 8% | |
| 60/40 | 10 | 245 | 3 | 9.3% | 11.7% | 19.8% | 5.9% (2010-04) | 0% | |
| 60/40 | 15 | 185 | 2 | 9.9% | 11.7% | 15.0% | 7.8% (2005-04) | 0% | |
| 60/40 | 20 | 125 | 1 | 10.4% | 12.3% | 14.4% | 9.8% (2006-10) | 0% | |
| Equal thirds | 5 | 305 | 6 | 6.5% | 11.6% | 19.2% | 4.0% (1996-10) | 3% | |
| Equal thirds | 10 | 245 | 3 | 8.2% | 11.9% | 17.3% | 6.5% (2010-04) | 0% | |
| Equal thirds | 15 | 185 | 2 | 9.6% | 11.4% | 15.0% | 8.6% (2005-04) | 0% | |
| Equal thirds | 20 | 125 | 1 | 10.6% | 11.7% | 12.7% | 10.4% (2003-11) | 0% | |
| Gold | 5 | 321 | 6 | −1.8% | 9.2% | 23.6% | −4.2% (2011-01) | 33% | |
| Gold | 10 | 261 | 3 | 3.5% | 9.3% | 19.2% | 2.4% (1995-06) | 13% | |
| Gold | 15 | 201 | 2 | 7.7% | 10.0% | 15.2% | 7.2% (2007-11) | 0% | |
| Gold | 20 | 141 | 1 | 8.6% | 9.5% | 13.3% | 8.4% (1999-05) | 0% | |
| G-sec, 10-year (synthetic) | 5 | 305 | 6 | 3.1% | 6.4% | 18.2% | 0.6% (2003-08) | 17% | |
| G-sec, 10-year (synthetic) | 10 | 245 | 3 | 5.4% | 7.0% | 8.9% | 4.6% (2001-11) | 8% | |
| G-sec, 10-year (synthetic) | 15 | 185 | 2 | 6.7% | 7.2% | 8.0% | 6.4% (2011-10) | 0% | |
| G-sec, 10-year (synthetic) | 20 | 125 | 1 | 6.7% | 7.3% | 8.3% | 6.6% (2006-10) | 0% |
By start month
The same story month by month, set against the FD SIP on the same instalments. Each point is one whole SIP.
Ten-year SIPs by the month they started
After-tax XIRR, per cent a year
How to read it. The Nifty 500 line falls from the mid-20s for late-1990s starts to around 10% for starts in 2006 to 2010, then recovers. The FD line, after tax each year, stays between 4.0% and 6.0% throughout.
Ten-year SIPs by the month they started. Line chart, Feb 1995 to Oct 2016. Nifty 500: latest 11.0%, range 3.4% to 31.6%. Gold: latest 19.5%, range 2.4% to 22.9%. G-sec, 10-year: latest 5.5%, range 4.6% to 10.2%. FD: latest 4.1%, range 4.0% to 6.0%.
Twenty-year SIPs by the month they started
After-tax XIRR, per cent a year; start months from 1995 to 2006
How to read it. Over twenty years the lines flatten and mostly keep their order: equity, then gold, then G-sec, then FDs. The exception: for 2006 starts, gold's median beat equity's, carried by gold's run of recent years.
Twenty-year SIPs by the month they started. Line chart, Feb 1995 to Oct 2006. Nifty 500: latest 11.3%, range 10.7% to 15.8%. Gold: latest 13.4%, range 8.4% to 14.9%. G-sec, 10-year: latest 6.6%, range 6.6% to 8.7%. FD: latest 4.6%, range 4.6% to 5.8%.
Against gold, G-sec and FDs
For each start year, the share of the Nifty 500's ten-year SIPs that finished behind the same instalments in gold, in the synthetic G-sec, or in FDs.
When a ten-year Nifty 500 SIP finished behind
Share of that year's start months, after tax
How to read it. Darker is more often behind. Gold beat equity for most starts in 2002, 2003, 2015, 2016; G-sec only for some 2010 starts (42% of them); FDs almost never. Over all start months, equity finished behind gold in 23% of ten-year windows.
When a ten-year Nifty 500 SIP finished behind. Heat table, 3 rows by 22 columns. Values from 0% to 100%.
The FD is 364-day T-bill primary yield (91-day before April 2005), after slab tax and cess each year. Before 1996 deposit rates were set by RBI and the same at every bank; from 1996 they are the rates of five major public sector banks.
Share of all start months from 1995 in which the Nifty 500 SIP finished behind each rival, after tax.
| Years | Windows | Behind gold | Behind G-sec | Behind FDs | More |
|---|---|---|---|---|---|
| 5 | 321 | 38% | 22% | 17% | |
| 10 | 261 | 23% | 2% | 0% | |
| 15 | 201 | 15% | 2% | 0% | |
| 20 | 141 | 9% | 0% | 0% |
How long under water
The statement value (after costs, before tax) against the money put in, for every start month from 1995. The longer record adds the slumps of the late 1990s and the dot-com bust of 2000 to 2003.
Share of SIPs under water, by how long they had been running
Per cent of start months from 1995 (1996 for the mixes)
How to read it. A five-year-old Nifty 500 SIP was under water in 9% of start months on this longer record. Gold took longest to recover: 11% still under water at five years, SIPs begun in 1995-96 that waited out gold's long slide to 2001, and SIPs begun in 2010-12 near its previous peak.
Share of SIPs under water, by how long they had been running. Heat table, 7 rows by 7 columns. Values from 0% to 37%.
The longest a SIP stayed below the money put in, FDs and inflation
Longest run of consecutive month-ends, any start month from 1995
How to read it. The Nifty 500's longest stretch under water was 3 years 2 months (from Aug 1999). Gold's was 5 years 7 months, and it stayed below FDs for 9 years 9 months. Equal thirds never stayed under water for more than 1 year.
The longest a SIP stayed below the money put in, FDs and inflation. Bar chart of 7 items. Below the money put in: highest Gold at 67, lowest Equal thirds at 12.
Longest stretches and the deepest point below the money put in, with the start month of the SIP that had each.
| Portfolio | Below the money put in | Below FDs | Below inflation | Deepest point | Its start | More |
|---|---|---|---|---|---|---|
| Nifty 500 | 3 years 2 months, from a Aug 1999 start | 3 years 3 months, from a Jul 1999 start | 3 years 4 months, from a Aug 2010 start | −50% | Oct 2007 | |
| Nifty 50 | 3 years 4 months, from a Dec 1999 start | 3 years 5 months, from a Dec 1999 start | 3 years 5 months, from a Dec 1999 start | −48% | Oct 2007 | |
| 60/20/20 | 1 year 11 months, from a Jan 2000 start | 3 years 2 months, from a Aug 1999 start | 2 years 9 months, from a Jan 2000 start | −30% | Nov 2007 | |
| 60/40 | 1 year 11 months, from a Jan 2000 start | 2 years 9 months, from a Jan 2000 start | 3 years 4 months, from a Aug 2009 start | −29% | Oct 2007 | |
| Equal thirds | 1 year, from a Nov 2007 start | 1 year 11 months, from a Jan 2000 start | 1 year 10 months, from a Jan 2000 start | −16% | Dec 2007 | |
| Gold | 5 years 7 months, from a Sept 1995 start | 9 years 9 months, from a Jan 1996 start | 9 years 6 months, from a Feb 1995 start | −19% | Oct 1995 | |
| G-sec, 10-year (synthetic) | 1 year 6 months, from a Aug 2003 start | 4 years 5 months, from a Aug 2003 start | 7 years 3 months, from a Nov 2001 start | −10% | Nov 1997 |
How the series are built
Nothing on this page assumes a return. Two series had to be assembled, and both were checked against a published index before any result was computed, with the pass marks written down in advance.
Equity. NSE's Nifty 500 total-return index from January 1995 and Nifty 50 total-return index from June 1999, as published, with dividends reinvested: nothing is assumed about dividends, and the Nifty 500's runs 1.5 to 3.4 points a year above its price index in those years, as it should. One caution: the Nifty 500 (then the CNX 500) has a base date of January 1995, but NSE's index company was set up only in 1998, so its first years are most likely calculated after the fact, as NSE's back-tests of newer indices are. Its constituents were chosen with some hindsight.
G-sec, rebuilt from yields. A 10-year government bond bought at par each month: its coupon is that month's 10-year yield. A month later it is repriced at the new yield, with a month less to run, plus the coupon earned, and the money rolls into a new 10-year bond. This is the standard way to turn a yield series into a bond index (Swinkels 2019). The yields are RBI's month-end 10-year G-sec yields from May 1996. Against NSE's 5-year G-sec index from Oct 2001, its monthly returns had a correlation of 0.83 (pass mark 0.8); it returned 8.0% a year against 7.8%, and swung more (6.8% volatility against 3.8%), as a longer bond should.
Gold. RBI's monthly average domestic price in Mumbai, rupees per 10 grams, from 1990, placed mid-month with the days between filled at a constant rate, and linked in June 2005 to the lab's daily gold (WGC's domestic rupee price). Over 244 months of overlap the two moved together with a correlation of 0.964 (pass mark 0.95). Before 2005 RBI's domestic price is preferred to a world price converted at the exchange rate, because India's import duties changed and the converted price misses that.
What is not here. Index funds, gilt funds and gold ETFs did not all exist in the 1990s: their later costs stand in for what an investor would have paid. Tax follows the rules of each year, with indexation from FY1995-96; the surcharges of the late 1990s are not modelled. A version from 1979 with the Sensex, which needs an assumed dividend yield, is next.
What I predicted
Written into the specification on 8 October 2026, before any of these numbers existed.
S13 Held
The median 10-year Nifty 500 SIP XIRR differs by more than 8 points between the best and the worst start year from 1995 to 2016.
1997 starts: 25.3% at the median; 2010 starts: 9.3%. A gap of 16.0 points.
S14 Failed
In at least one start year from 1996 to 2016, the median 10-year Nifty 500 SIP had a lower after-tax XIRR than the synthetic G-sec SIP.
In no start year. The closest was 2010, when the Nifty 500 median was 9.3% against 8.5% for G-sec, and 42% of that year's start months did finish behind.
S15 Held
The longest stretch a Nifty 500 SIP started from 1995 spent below the money put in is longer than three years.
3 years 2 months, for a SIP begun in Aug 1999, just before the dot-com bust.
S16 Held
Over all start months from 1995, a 10-year Nifty 500 SIP beat the same SIP in gold in more than two thirds of windows.
It finished ahead in 77% of 261 windows. The losses bunch in 2002, 2003, 2015, 2016 starts.
Historical research on index portfolios, not investment advice. Returns are after fund costs; 'after tax' applies Indian capital-gains rules by date for a 30% slab investor and sells everything at the end. Most factor, mid-cap and small-cap index history before each index's launch date is back-tested by NSE. One 21-year sample, domestic assets only.
The workings
Data
Every chart has a table view, a CSV download and a link to its published JSON. Bundles are listed in data status.