What this calculator measures
It chains 98 published calendar years of S&P 500 total return, from 1928 to 2025, and runs your dirhams through them. Total return is the change in the index price plus dividends reinvested. Over the whole series that compounds to x13,990.86, or 10.23% a year; from 1957, the year the S&P 500 itself began, x1,085.68 or 10.66% a year. Of the 98 years, 72 were up and 26 were down.
The distinction between total return and price return is the one that ruins most published versions of this arithmetic. In 2024 the index returned 23.31% on price and 1.71% in dividends, for 25.02% in total, and the price-only figure is lower in every modern year. Take that one dividend leg away from every year of the default window and x19.21 becomes x11.93: about 38% less, on the same thirty years. That is an illustration of the scale rather than a price-return series, because we publish the total return column and not the year-by-year dividend column. Tables headed "total return" that carry price-only numbers are common, and they are wrong by roughly that much.
- Total return, dividends reinvested. Not price only. The reinvestment convention is the official total return index's own: dividends go back in as they are paid.
- Nominal, before inflation. We publish no inflation-adjusted figure, because we have not sourced a price series to adjust it with. A nominal result overstates what the money would buy.
- Calendar years, close to close. Each figure runs from the last trading day of the prior year to the last trading day of the year named. Not rolling, not twelve-month trailing.
- A single lump sum. Money in on day one and out at the end. Modelling a monthly contribution against an annual series needs an assumption about when inside each year the money arrives, and that assumption changes the answer without being sourced from anything, so this tool does not make one.
One thing the series is not: a product. Nobody has ever received the index. It has no expense ratio, no platform fee, no commission and no minimum ticket, and every vehicle that tracks it has some of those. The deductions further down are the ones this tool models, and they are not the complete list.
The formula used
Three steps, and the order matters, because two of the three costs are charged once and one is charged every year. Dirhams buy dollars, the dollars compound through the series, and the dollars are sold back for dirhams.
U = A / (P × (1 + cin))
V = U × ∏ (1 + (ry − d))
B = V × P × (1 − cout)
- A is the dirhams you invest, and B the dirhams you end up with.
- P is the dirham peg, 3.6725 to the dollar, which does not move.
- cin and cout are the conversion costs, each charged once.
- ry is year y's published total return.
- d is the annual drag: dividend withholding plus fund cost, in percentage points.
- ∏ is a product over every year in the span, not a sum and not an average.
- n is the number of years, used only to annualise: the compound rate is (B/A)1/n − 1.
Check the tool's opening state by hand. AED 10,000 at the start of 1996, sold at the end of 2025, with the conversion costs and the withholding it opens with. The buy rate is 3.6824 dirhams to the dollar, so AED 10,000 becomes $2,715.61. Chained through 30 years of total return less 0.51 percentage points a year, that multiplies by x16.6810 to $45,299.11. Sold at 3.6641, it comes back as AED 165,978.
Two figures, and they are not interchangeable. The total return over that span is 1,559.78%, which is what the whole period did. The annualised figure is 9.82%, which is the compound rate that produces the same result. Neither is the average of the annual returns: over the whole series those returns average 12.10% while the compound rate is 10.23%. An arithmetic mean of annual returns is always the higher number and is never a growth rate, so anywhere you see one presented as the other, the result is overstated.
Where the series comes from
Two sources sit behind these 98 figures, and this page treats them differently because we reached them differently. The one we fetched ourselves is Aswath Damodaran, NYU Stern, "Historical Returns on Stocks, Bonds and Bills": the histretSP.xls workbook, downloaded on 26 September 2026 and read out of the file itself. Every figure in the published column is checked against it.
The column we publish is not that one, though. It follows Slickcharts, "S&P 500 Annual Returns" (details table), a tabulation of the S&P 500 total return index that S&P Dow Jones Indices maintains, and the copy of that table these figures come from is dated 26 September 2026. We publish it rather than the workbook because it uses the convention a tracker actually follows: dividends reinvested as they are paid, not once at year end. Read the citation carefully. It is a tabulation of the index owner's data rather than the index owner's own publication, which we could not reach, so the corroboration below is not a formality. It is the reason these figures are on the page at all.
The two sources are genuinely independent rather than two copies of one table: the workbook computes each year's dividend yield against the index level at the start of the year and reinvests once at the end of it. They agree on the direction of the market in all 98 years. There is no year in which one says the index rose and the other says it fell.
- From 1957 on, the two are close. Year by year they sit 0.22 of a percentage point apart on average, and the widest single year is 1.13. That average counts every gap as a distance: the years one source runs high are not allowed to cancel the years it runs low, which would halve the figure and flatter both tables. The gap itself is the reinvestment convention, and we publish the one that matches how a tracker behaves.
- Before 1957, they are not. That average distance widens to 0.76 of a point and the worst year to 4.01 points, in 1933. This is the weak part of the series and the tool marks it.
- Two years were dropped and one was excluded. The tabulation runs from 1926, but 1926 and 1927 appear in only one of the two sources, so they are not here: a figure with no second source does not go into a published series. 2026 is excluded because it is incomplete, and a year to date is not a calendar-year total return.
The 1928 to 1956 stretch deserves its own warning, and the tool flags it whenever your start year reaches into it. The S&P 500 as such began in 1957. Before that the figures describe the index's 90-stock predecessor plus back-filled data, so that period is a reconstruction rather than a measurement. It is also where the two sources disagree most. Everything said about the modern era rests on two independent tabulations that never sit more than 1.13 points apart in any year; nothing said about the 1930s rests on anything that strong. Treat it as context.
The dirham leg: a fixed peg and a cost inside the rate
The dirham is pegged to the US dollar at 3.6725, set by the UAE Central Bank, and it does not move. That has a consequence worth stating plainly, because it is the opposite of what most currency commentary would lead you to expect: converting in and out at the peg itself leaves the dirham result exactly equal to the dollar result, because the peg cancels. A UAE resident holding a dollar-denominated S&P 500 fund is taking no currency bet at all, unlike a euro or sterling investor. There is also no such thing as a dirham return series distinct from the dollar one, which is why this page does not publish one.
What is left is not a risk but a charge: two conversions, one on the way in and one on the way out, taken inside the rate rather than billed as a fee. That is the cost this tool exists to show, and we have measured it on real UAE accounts rather than assuming a spread. The range is wide and it depends on the broker, not on the market.
| Broker | What we measured on the conversion |
|---|---|
| AvaTrade | None visible in our test: AED 400 in became $108.93 and $108.85 out became AED 399.752, both at the 3.6725 peg. AvaTrade warns that deposits in another currency "may result in charges for conversion to the account's currency" |
| Amana | Measured on one deposit: AED 91.75 credited as USD 24.99, an implied 3.6715 against the 3.6725 peg, so no mark-up on Amana's side. Amana states a fixed 3.67. The bank's side was not checked |
| Sarwa | Not published. We measured ~0.27% on a 200 AED deposit (200 AED credited as $54.31) and ~0.23% on the withdrawal ($52.32 paid out as 191.70 AED) |
| Pepperstone | About 1% on one internal USD-to-AED transfer, filled at 3.635775 on 26 August against the 3.6725 AED peg. That is a single internal conversion in one direction, not a measured average, and it is separate from the AED funding rail, which moved our money both ways at the peg with no fee. Pepperstone reports 0.01% |
Now the point the arithmetic makes, and it is not the one the table suggests. A conversion spread is charged once each way, so it never compounds. On the tool's opening state it costs AED 832, which is 0.50% of the result, and it would be that same 0.50% on a one-year hold as on this 30-year one. The annual deductions below do compound, and on the same settings they take AED 25,321, about 30 times as much. Show a reader the currency cost and stay quiet about the dividend tax, and every individual number stays true while the emphasis misleads.
The costs the index does not carry
The series reinvests dividends gross and charges nothing. A UAE resident tracking it pays for both. These are the deductions the tool models, and each one is an editable input rather than a hidden constant.
- US withholding on dividends, and this is the big one. The IRS lists no US-UAE income tax treaty, so US dividends paid to a UAE resident face 30% withholding (IRS Publications 515 and 519), and the W-8BEN signed at onboarding does not reduce it. The default drag is that rate applied to a dividend yield we can source: 30% of 2024's 1.71% dividend leg, which is 0.51 percentage points a year. Read what that default is. It is one year's dividend yield applied to every year of whatever span you pick, because we publish the total return column and not the year-by-year dividend column, so it is an approximation of the drag rather than a year-by-year calculation. It is also an approximation that errs one way: 2024 was a low-yield year by the standards of this series, and dividends were a larger share of the return through most of it, so on a long span this understates the withholding rather than overstating it. The input is editable and the arithmetic is the same at any figure you put in it.
- Fund cost, which we default to nothing and should not be read as zero. The index has no expense ratio and every tracker does. It is a figure per fund rather than per market, we have not sourced one, and inventing a plausible number would be worse than leaving the field at nought. Put your own fund's figure in: it comes off the return every year, so it compounds like the withholding rather than like the conversion.
- Platform costs, which are not modelled at all. Custody or platform fees, commission or spread on each purchase, minimum tickets and any inactivity fee. We have measured several of these per broker, in the reviews, and none of them is in this series or in this tool.
- Nothing else. The tool applies the withholding, the fund cost and the two conversions. Anything not on that list is not deducted, so the result is an upper bound on what a real account would have produced, not an estimate of it.
It is worth seeing the scale before trusting any headline. On the opening state the index itself returns AED 192,131 on AED 10,000. The annual deductions take AED 25,321 of that, about 13%, and the currency takes AED 832. The annualised figure falls from 10.35% to 9.82% on a drag of 0.51 points a year, which is what compounding a small annual leak over thirty years does.
Two figures we deliberately do not publish. A Gulf investor usually reaches the index through an Irish-domiciled UCITS tracker, which is taxed at a lower rate at fund level under the US-Ireland treaty rather than at the 30% above. We have not sourced that rate, so the tool offers no preset for it and this page names no number, and you should lower the withholding input yourself if you hold one. We also quote no tracker expense ratio for the same reason. A cost we cannot source does not become a cost we estimate.
Why one number is not a forecast
A single compounded figure hides the order the years arrived in, and the order is most of what a real investor experiences. The tool answers this directly: whatever span you pick, it also reports every window of that length in the series, so the window you chose can be seen as one draw rather than as the result. The figures in this section are the index itself, before the deductions above, while the tool's own panel applies whatever drag you have set, so its version of them sits a little lower.
- Ten-year holds. The series contains 89 of them, and 4 ended below where they started. The worst ran 1999 to 2008 at -1.38% a year; the best, 1949 to 1958, at 20.06%.
- 30-year holds. There are 69 of those. 1929 to 1958 compounded at 8.47% a year and 1975 to 2004 at 13.75%. Both are the same index, and the only difference between them is the year somebody happened to start.
- What a bad start actually feels like. Of the 10 calendar year ends from 2000 to 2009, 8 closed below the starting value. The first that did not was 2006, and 2008 put it back under. Sequence is not a footnote to this arithmetic, it is most of the experience of it.
The same series holds 26 down years out of 98, and the worst five were 1931 at -43.34%, 2008 at -37.00%, 1937 at -35.03%, 1974 at -26.47%, 1930 at -24.90%. Anyone contributing monthly or drawing money out gets a materially different result from the same set of annual returns rearranged, which is a second reason this tool models a lump sum and says so rather than quietly picking a contribution schedule for you.
Past returns are not a forecast. That warning applies harder here than on an assumed-rate calculator, because a historical series looks like evidence: it is measured, it is sourced, and it still says nothing about the next thirty years. Move the start year, watch the answer move with it, and treat the spread between the answers as the honest output.
Where to take this next
A historical return is only as useful as the account you would hold the tracker in. What it costs to buy, what the platform charges to keep it, and whether the firm is licensed to hold your money in the UAE all change the real result more than a decimal place on the index.
- Compound interest calculator, for a plan with monthly contributions and a rate you choose rather than one history handed you.
- Best trading platforms and apps in the UAE, compared on cost and on the entity that onboards a UAE resident.
- Tax on trading in the UAE, which sets out where the UAE's position ends and another country's begins.
- UAE regulation, covering who is licensed here and what each licence permits.
- Our methodology, for what we measure, what we source and how we keep the two apart.
This calculator is provided for education. It runs a sourced historical index series and the costs you enter, and it is not investment advice, not a recommendation to buy any index tracker, and not a forecast. The index is not a product, past returns do not predict future ones, and the result is nominal: before inflation, and before every cost listed above that the tool does not model.