Every time someone truly grasps how income tax hits a coupon bond, the same idea is born. It’s good. It’s elegant. And it’s almost always wrong.
The idea is this: income tax on Tesouro Direto (Brazil’s government bond platform) follows the regressive table, counted from the purchase date. A coupon received within 180 days pays 22.5%; from 181 to 360 days, 20%; and so on until it settles at 15% after two years. Since the NTN-F (Brazil’s fixed-rate treasury bond with semiannual coupons) pays on fixed dates — 1 January and 1 July — the number of days between your purchase and the first coupon depends on when you buy. Pick the date well, and the first coupon lands in a cheaper bracket. Free money, just for buying on the right day.
The logic looks flawless
And on paper, it is. If you buy just after a coupon date, the next payment only arrives almost a full semester later — 181 to 184 days down the road — and escapes the 22.5% bracket, dropping to 20%. If you buy on the eve of a coupon, that first payment lands within days, full of tax. Same NTN-F, same issuer, same contracted rate — and a tax difference you created yourself, purely by the date on the receipt.
The reasoning is seductive because it combines two things the Brazilian investor loves: a tax rule you can exploit, and a sense of control over something that usually feels like luck. It’s the kind of optimisation that makes a good thread on Twitter and leaves everyone feeling clever.
The problem is that nobody who proposes the idea ran the numbers all the way through. We did.
First obstacle: the window is a handful of days a year
The 22.5% bracket applies to coupons received within 180 days. For the first coupon to land in the 20% bracket, you have to buy the bond just after a coupon date, so the next payment sits more than 180 days out — that is, almost a full semester ahead.
We swept all 365 purchase dates in the year. The result is discouraging: on 360 of them, the first coupon arrives within 180 days or less and pays the full 22.5%. Only five days deliver the 20% bracket — 1 January and 1 to 4 July (the second half of the year, July to January, runs 184 days, which opens a little four-day window). Of those five, three are normal trading days: 2, 3 and 4 July. So yes, you can hit it — the window exists and is tradeable, it’s just tiny.
You can hit the date. What you can’t do is make it worth anything — and that’s where the idea falls apart.
Second obstacle: even captured, the prize is irrelevant
Suppose you manage it — you bought on the magic date and caught the 20%. What’s it worth?
We modelled a ten-year NTN-F, R$ 10,000 invested, a contracted rate of 13% a year, comparing the best possible purchase date against the worst. The difference in the final balance, after ten years:
| Worst date | Best date | |
|---|---|---|
| 1st coupon in the bracket of | 22.5% | 20% |
| Net over 10 years | R$ 28,756 | R$ 28,872 |
| Return | 11.14% p.a. | 11.19% p.a. |
One hundred and sixteen reais. Over ten years. Less than half a percentage point on the accumulated total, or 0.045 point a year — four and a half hundredths. The effect only touches the first three or four coupons, because from the fifth onward they’ve all passed 720 days and pay 15% regardless of when you bought. It’s an advantage that’s born small and dies fast.
One hundred and sixteen reais isn’t zero. But it’s the kind of number that vanishes on the first day of price volatility in the bond, and that costs more in calendar anxiety than it earns in tax saved.
What the mirage hides
Here’s the part that matters, and it’s why running the numbers was worth it.
While you aim at the R$ 116 of timing, there’s a decision sitting right in front of you that’s worth thirteen times more — and that one you capture for free. It’s the choice between having the coupon or not. Deferring the tax to a single event at maturity, by buying a zero-coupon instead of the coupon bond, saves about R$ 1,491 in the same example: 0.56 point a year, against the 0.045 of timing.
| Optimisation | Worth | How you capture it |
|---|---|---|
| Defer the tax (zero-coupon) | R$ 1,491 · 0.56 pp/yr | Choosing the bond. For free. |
| Nail the coupon date | R$ 116 · 0.045 pp/yr | A ~3-day tradeable window a year |
Deferral is large and automatic. Timing is tiny and nearly unreachable. They differ along two dimensions at once — size and capturability — and confusing one for the other is the classic mistake of spending all your energy on the screw while the beam hangs loose.
It’s no coincidence that the useless optimisation is the more tempting one. It’s concrete, it has a clear rule, it gives a binary answer — I bought on the right day or I didn’t. The decision that matters is more uncomfortable, because it demands admitting that maybe you shouldn’t hold a coupon bond during accumulation, full stop. Morgan Housel would call it what it is: we prefer the problem we know how to solve to the problem we need to solve.
Where the same energy actually pays off
If coupon timing is noise, where should the attention you were going to spend on it go instead? Fixed income has a well-defined hierarchy of levers, and timing sits at the absolute bottom of it. It’s worth seeing the whole order, because almost everyone optimises from the bottom up.
| Lever | Worth (approx.) | Effort |
|---|---|---|
| Choose the right product — defer the tax with a no-coupon bond, a fund without come-cotas (Brazil’s twice-yearly automatic fund tax) | 0.5 to 0.75 point a year | One decision, at purchase |
| Hold past two years — 15% instead of 22.5% on the whole gain | up to 7.5 points of tax on the profit | Just don’t sell early |
| Don’t pay a steep fee or churn the portfolio for no reason | fractions of a point, recurring | Low |
| Nail the coupon purchase date | 0.045 point a year | High (watching a calendar) |
Read it top to bottom and notice the pattern: the levers worth the most are the ones that take the least effort. Choosing a no-coupon bond is a single decision, made once, and it’s worth ten to twenty times more than timing. Holding the bond past 720 days to drop from 22.5% to 15% income tax saves up to 7.5 points on the entire profit — and the only “action” required is not selling on impulse before maturity. These are the laziest decisions on the list, and by far the most lucrative.
The 720-day step deserves an aside, because it’s the lever most people lose without noticing. The regressive table falls in steps of 2.5 points: at 181 days (from 22.5% to 20%), at 361 (to 17.5%), and at 721 (to 15%). Selling a bond on the 700th day instead of waiting until the 721st costs 2.5 points of tax on the whole two-year gain — and someone who redeems a CDB (fixed-income bank certificates, similar to CDs in the US) on the 179th day out of haste hands over the same 2.5 points they’d have kept by waiting 48 hours. None of these steps show up on the redemption screen warning you to “wait a little longer.” Each is worth multiples of coupon timing, and the only skill required is to check how many days are left before clicking sell.
Timing, at the bottom, is the opposite: it earns four hundredths of a point a year and demands the vigilance of a calendar. The investor’s intuition tends to invert the two ends, and not out of stupidity. The decision at the top forces you to admit an uncomfortable thesis — maybe you shouldn’t have this product in your portfolio. The one at the bottom offers a concrete little task, with a clear rule and a binary answer. It’s always more comfortable to fuss with the screw than to face the loose beam.
And the distance between top and bottom doesn’t shrink as wealth grows — it widens. On a R$ 10,000 investment, choosing the right bond is worth about R$ 1,491 and timing, R$ 116. Multiply by fifty, for R$ 500,000: the product choice now runs close to R$ 74,550, and timing, R$ 5,792. Both grow, but the thirteen-to-one ratio doesn’t budge a millimetre. The more money you have, the more expensive it gets to have optimised the wrong thing.
And the most ironic part is that the lever at the top and the one at the bottom compete for the same thing: your attention, which is finite. Every hour spent choosing the best day to buy is an hour not spent on the only question that moves the needle — should this bond even be in my portfolio? Timing doesn’t complement the decision that matters. It steals focus from it.
What this actually means
The regressive table isn’t a game to be won on the calendar. The gain from nailing the purchase date of a coupon bond exists, but it’s on the order of half a percentage point over an entire decade, concentrated in the first coupons and capturable only in a window of a few days at each semester’s turn. Chasing it is optimising the noise.
If you caught yourself thinking about which day to buy so the coupon lands in a better bracket, the instinct is right — early tax really does cost. It’s just that the right answer to that instinct isn’t to pick the date. It’s to step back and ask why you’re buying a coupon bond during accumulation at all, when the zero-coupon defers the whole tax at no cost, with no reinvestment risk and no calendar to nail. Model both paths through the Retirement Calculator with your horizon and watch the difference show up where it’s large, not where it’s easy to measure.
Coupon timing is the rare optimisation that gets more appealing the less you calculate. All it took was calculating.
This article is for general educational purposes and does not constitute investment advice. The figures are illustrative and assume a hypothetical ten-year NTN-F with a contracted rate of 13% a year, purchase and holding to maturity, and reinvestment of coupons. The income-tax rates described are those in force in 2026 and the exact schedule depends on the purchase date and the bond’s payment dates. Confirm the rules with Tesouro Direto and consult a qualified financial adviser before deciding.