TLT and interest rates

For a rise of 1 percentage point in every Treasury yield, the value of the TLT model portfolio changes by an estimated -13.61%. For a fall of 1 percentage point, it changes by an estimated +16.84%.

Estimates by full repricing of the model portfolio's payments on the Treasury par yield curve for . The model portfolio is built by rule from the Treasury securities outstanding on . It is not the fund's holdings.

Sources: U.S. Department of the Treasury, daily par yield curve rates and Monthly Statement of the Public Debt. The model is this site's.

For the fund itself, see the issuer's page for TLT. This site is not affiliated with or endorsed by the fund's issuer.

TLT model portfolio on the Treasury par yield curve for November 13, 2026
MeasureFigure
Duration of all payments together14.92 years
Average of the securities' own durations14.92 years
Convexity313.65
Yield of all payments together5.52%
Weighted average maturity25.80 years
Number of securities40
Maturity band20 years or more to maturity
Kind of securityCoupon-bearing Treasury notes and bonds
WeightingIn proportion to the amount outstanding of each security
Securities outstanding as ofOctober 31, 2026
Curve dateNovember 13, 2026

The first duration takes all the payments of the model portfolio together at one yield, and it is the one the duration-only estimate uses. The second is the average of the securities' own durations, which is how a fund's published duration is built up.

TLT: the fund's published duration and the model's like-for-like figure
MeasureFigure
Published duration14.75 years
Published as ofNovember 12, 2026
Model: average of its securities' own durations14.92 years
Model as ofNovember 13, 2026
Model against published1.2% above

The issuer's figure describes the fund on the issuer's date, the model's figure describes every security its rule selects on the curve date, and the known differences between the two are listed in the method section of this page.

TLT model portfolio: the same move in every Treasury yield, on the par yield curve for November 13, 2026
Move (basis points)Price change nowDuration-only estimateConvexity contribution12-month return
-200+37.78%+29.83%+6.27%+42.42%
-150+26.73%+22.37%+3.53%+31.62%
-100+16.84%+14.92%+1.57%+21.91%
-50+7.96%+7.46%+0.39%+13.18%
00.00%0.00%0.00%+5.32%
+50-7.16%-7.46%+0.39%-1.77%
+100-13.61%-14.92%+1.57%-8.18%
+150-19.42%-22.37%+3.53%-13.97%
+200-24.67%-29.83%+6.27%-19.21%

The price change now is by full repricing of the model portfolio's payments. The 12-month return is a total return: it is for the whole 12 months, not annualized, with the move in yields in place at their end. Payments received during the 12 months are added as cash and are not reinvested. The duration-only estimate and the convexity contribution are the two terms of the usual shortcut; the full repricing is the answer.

The largest positions of the TLT model portfolio

The 10 largest of the 40 securities of the TLT model portfolio by model value, on the par yield curve for November 13, 2026
CUSIPCouponMaturityWeight by model value
912810UK24.750%May 15, 20553.85%
912810UU05.000%May 15, 20563.83%
912810UW65.125%August 15, 20563.78%
912810UR74.750%February 15, 20563.76%
912810UP14.625%November 15, 20553.70%
912810UG14.625%February 15, 20553.68%
912810UE64.500%November 15, 20543.66%
912810UM84.750%August 15, 20553.66%
912810UA44.625%May 15, 20543.57%
912810UC04.250%August 15, 20543.21%

The model portfolio has 30 more securities, 40 in all. These are positions of a model portfolio built by rule. They are not the fund's holdings.

The TLT model compared with the fund's daily moves

This site compared the daily moves of the TLT model with the fund's daily price moves, taken from exchange closing prices, from November 2, 2023 to November 13, 2026, over 713 trading days. The days taken to be the fund's distribution days were left out, because the amounts paid are not among this site's records. The model's moves explained 96 percent of the variation in the fund's daily moves. For a 1 percent move of the model the fund moved 0.98 percent on average.

What was measured for TLT
MeasureFigure
First dayNovember 2, 2023
Last dayNovember 13, 2026
Trading days713
Share of the variation in the fund's daily moves that the model's moves explained (R²)96%
The fund's average move, in percent, for a 1 percent move of the model (slope)0.98

Measurement run on . No fund price is stored or shown on this site.

Scenario for the TLT model portfolio

From -300 to 300. 100 basis points are 1 percentage point; a minus sign is a fall.

Optional. The 10-year yield of November 13, 2026 is 5.15%; a level here sets the move and the field above is not used.

Optional. With a figure here the result is the duration-only approximation for that duration and not the model portfolio's result.

For a rise of 100 basis points in every Treasury yield, the value of the TLT model portfolio changes by an estimated -13.61% now, or -$1,361 on $10,000.

Model portfolio for TLT (iShares 20+ Year Treasury Bond ETF): a rise of 100 basis points in every Treasury yield
MeasureChange in percentOn $10,000
Price change now, by full repricing-13.61%-$1,361
Duration-only estimate-14.92%-$1,492
Convexity contribution+1.57%+$157
Total return over 12 months-8.18%-$818

The total return is for the whole 12 months, not annualized, with the move in yields in place at their end. Payments received during the 12 months are added as cash and are not reinvested.

Break-even: a rise of about 37 basis points in every yield over the 12 months would leave a total return of zero. The figure is rounded to the whole basis point, so a rise of exactly 37 basis points leaves a total return close to zero.

Full repricing discounts every payment of the model portfolio again on the moved curve. The duration-only estimate and the convexity contribution are the two terms of the usual shortcut, from the model portfolio's duration of 14.92 years and its convexity of 313.65. Each figure in the table is worked from figures that are not rounded and is rounded once, so the two terms can differ in the last digit from the same steps done with the duration and the convexity shown here, and the dollars can differ from the percent shown applied to the amount.

The model portfolio is built by rule from the Treasury securities outstanding on October 31, 2026, valued on the par yield curve for November 13, 2026. It is not the fund's holdings.

What a move in yields does to the TLT model

On the Treasury par yield curve for November 13, 2026, a rise of 1 percentage point in every Treasury yield changes the value of the TLT model by -13.61%. A fall of 1 point changes it by 16.84%. On $10,000 those are -$1,361 and $1,684. The 10-year par yield on that curve was 5.15%.

TLT is the ticker of the iShares 20+ Year Treasury Bond ETF, issued by iShares (BlackRock). Its stated aim is to track the ICE US Treasury 20+ Year Bond Index, which the issuer describes as U.S. Treasury bonds with more than twenty years left to maturity. Rate Workbench is not affiliated with or endorsed by any fund issuer. The fund's price, holdings, yield and costs are on the issuer's page for TLT, and this page shows none of them.

A model, not the fund. The figures come from a portfolio the site builds from the Treasury's own list of securities outstanding on October 31, 2026. Maturity band: 20 years or more to maturity. Kind of security: coupon-bearing Treasury notes and bonds. Number of securities: 40. The model's yield is 5.52%. On the list for October 31, 2026 every bond in the band had first been sold as a 30-year bond, so "20 years or more" meant bonds with 20 to 30 years left.

Duration is well short of maturity. The bonds in the model mature in 25.80 years on average, weighted by value, but the model's duration is 14.92 years. Duration is what sets the size of the price move, and the coupons shorten it. A bond with 25 years left makes 50 more interest payments, and all but the last arrive before the principal does.

Convexity matters at this length. The model's convexity is 313.65. It is the reason the gain from a fall in yields is larger than the loss from an equal rise. For a move of a few basis points the difference is too small to show in figures rounded to two decimals, so the calculator can show a gain and a loss of the same size. For a rise of 1 point, duration by itself gives -14.92%, and the convexity correction is 1.57 percentage points. The table of moves on this page, "TLT model portfolio: the same move in every Treasury yield", shows both beside the full repricing, for moves of up to 200 basis points each way. Its duration-only column is worked from the duration in the row "Duration of all payments together" of the model table, before that figure is rounded, so for a move other than 1 point either way it can differ by as much as 0.01 from the figure shown in that row times the move in percentage points. Each cell is rounded on its own, so a row worked out from another row as shown can end in other last digits.

What is certain and what is not. The par yields and the list of securities are official and dated, and the arithmetic is exact for the model. Two things are estimates: how well a portfolio built from a maturity rule stands in for the fund, and the scenario, in which every maturity moves by the same amount at the same moment. The scenario is a choice, not a forecast. Today's yields are on the Treasury yield curve today page, the six fund models sit side by side on the bond ETF interest rate calculator, and the zero rates and discount factors of that curve are on the Treasury curve lab.

Worked example: 1 point up and 1 point down on $10,000

The inputs are the TLT model on the par curve for November 13, 2026, with the securities outstanding on October 31, 2026. The amount is $10,000. The move is 1 percentage point, or 100 basis points, in every Treasury yield.

  1. Duration by itself. The model's duration is 14.92 years. Multiply it by the change in yield in percentage points, which is 1, and reverse the sign. Result for a rise: -14.92%.
  2. Convexity correction. Take half the convexity times the square of the change in yield as a decimal: ½ × 313.65 × 0.01 × 0.01. Result: 1.57 percentage points, with the same sign whether yields rise or fall. The result is worked from the convexity before rounding, so it can differ in the last digit from the product of the figures shown.
  3. Full repricing, yields up. Every payment of the model is discounted again on the curve moved up by 1 point. Change in value: -13.61%. On $10,000: -$1,361. The loss is smaller than step 1 says, because of the correction in step 2.
  4. Full repricing, yields down. On the curve moved down by 1 point the change is 16.84%, which is $1,684 on $10,000. The gain is larger than the loss in step 3. That asymmetry is convexity at work.
  5. Twelve months with no move. If every maturity still had today's yield a year from now, the model's total return over the 12 months would be 5.32%. Interest received during the year is added as cash and is not reinvested. The figure is for the whole period.
  6. Break-even. This is the move in every yield at which the 12-month total return of step 5 comes to zero: a rise when that return is above zero, and a fall, shown with a minus sign, when it is below. Rounded to the whole basis point: 37. If that reads "none found", no move within the limits of the calculation brings the return to zero.

How the TLT model is built and where it differs from the index

Selection and weights. The model holds every fixed-rate Treasury bond with 20 years or more left to maturity on the curve date, in proportion to its amount outstanding. Time left is counted in calendar months. Each bond contributes all of its remaining interest payments and its principal. A payment due on the curve date itself is not counted. The securities come from the Treasury's Monthly Statement of the Public Debt and the yields from its Daily Treasury Par Yield Curve Rates. The page lists the ten largest positions of the model and counts the rest.

The shortcut in symbols. Steps 1 and 2 of the example are the standard second-order estimate:

ΔV/V ≈ −D × Δy + ½ × C × (Δy)2

D is the model's modified duration, C its convexity and Δy the change in yield as a decimal, so 1 point is 0.01. Both are measured on all the model's payments together at their one bond-equivalent yield, which is twice the half-year rate. The estimate treats a move of the par curve as an equal move of that one yield. That is exact only on a flat curve, which is why the page gives the full repricing as its answer and the shortcut beside it.

Where the model differs from the index. The index rules below are those of the fund's summary prospectus, as read for this site on November 15, 2026.

  • The 20-year edge. The index takes maturities greater than twenty years. The model also includes a bond with exactly 20 years left.
  • Timing. The index fixes its members on the last calendar day of each month. The model selects by time left on the curve date from the latest month-end list, so a bond sold after October 31, 2026 is missing until the next statement.
  • Size. The index requires at least $300 million outstanding after Federal Reserve holdings. The model applies no minimum.

What a model of this kind leaves out. It knows nothing of the fund's fees, its cash, its exact holdings or its trading, and nothing of the premium or discount at which the fund's shares trade against its net asset value. It treats every bond as paying full half-yearly interest, leaves out any bond with a payment more than 30 years away because the curve ends there, moves every maturity by the same amount and leaves cash from payments uninvested. Taxes are left out.

The issuer's duration figure. The issuer's page for TLT showed a published duration of 14.75 years as of November 12, 2026. Taken as the average of its securities' own durations, which is how a fund's published duration is built up, the model's figure is 14.92 years on the Treasury curve for November 13, 2026, 1.2 percent above the published figure.

Two model durations appear on this page because they answer different questions. The first treats the whole model as one stream of payments discounted at one yield, and step 1 of the example uses it. It is the row "Duration of all payments together" of the model table: 14.92 years. The second takes each bond's own duration at that bond's own yield and averages them by value. It is the row "Average of the securities' own durations": 14.92 years. On a flat curve the two agree exactly, and on a curve that is close to flat along the band they can print the same number to two decimals. They remain two calculations. The issuer calls its figure an effective duration and describes it as measured bond by bond and then combined for the fund, so the second model figure is the one set against it.

The two still describe different portfolios on different dates. The issuer's figure belongs to the fund's holdings on the issuer's date, and the model's to every bond in the band on the Treasury's list for October 31, 2026, valued on the curve for November 13, 2026. They are shown together for comparison, and the model is not fitted to the issuer's figure.

The model beside the fund's daily moves. Where the model's day-to-day moves have been compared with the day-to-day moves of the fund's price, the result is in the block headed "The TLT model compared with the fund's daily moves". A comparison of that kind covers one period, and it is about the direction and size of daily moves, not the fund's return or its price. The closing prices it uses are not published on this site. The block states the period and how the fund's distribution days were treated. Both can differ from fund to fund, so the figures of two funds are not a ranking.

Checked against reference values

Each row sets this site's calculation beside a reference value. The Check and Source columns say where the reference comes from: a figure an official source published, or a value worked from published figures, from a definition or from a formula. The page is not built when a row does not match.

Reference values next to this site's calculation
CheckReferenceThis siteResultSource
Modified duration of a 10-year bond priced at par at 5.15%, the 10-year Treasury par yield of November 13, 2026: closed form beside a repricing 1 basis point up and down7.74 years7.74 yearsMatchClosed form of a par bond's modified duration; the par yield from the U.S. Department of the Treasury, daily par yield curve rates, retrieved

Sources and freshness

Questions

Why do long Treasury funds fall so far when yields rise?

Because a long bond's payments are discounted over many years, a small change in the rate compounds into a large change in value. The TLT model's duration is 14.92 years, so a rise of 1 point in every yield moves its value by roughly that many percent before the convexity correction. Change in value by full repricing on the curve for November 13, 2026: -13.61%. A principal payment due in 25 years is discounted over 50 half-year periods, against 4 for one due in 2 years. The table of moves on this page carries the figures out to a rise of 200 basis points.

Does TLT gain more from a 1-point fall than it loses from a 1-point rise?

In the model, yes: 16.84% for the fall against -13.61% for the rise, on the curve for November 13, 2026. The difference is convexity, which is 313.65 for this model. As yields fall the sensitivity itself grows, and as yields rise it shrinks. The gap between gain and loss widens with the size of the move, as the table of moves on this page shows.

If yields stay where they are, what does the TLT model return in a year?

5.32% over 12 months, on the curve for November 13, 2026. The figure adds the interest the bonds pay during the year, kept as cash, to the change in their value as each bond moves a year closer to maturity on an unchanged curve. The model's yield is 5.52%. An unchanged curve is an assumption for the calculation and not a forecast, and the fund's own return also depends on its fees and its trading price.

Why is TLT's duration so much shorter than the maturity of its bonds?

Because a coupon bond pays part of its value long before it matures. The model's bonds mature in 25.80 years on average, while its duration is 14.92 years. Duration weighs the date of every payment by what that payment is worth today, so the interest paid in the early years pulls the figure down. A security with no coupons has a duration close to its maturity, which is the case of the EDV model.

How closely does the model follow TLT from day to day?

From November 2, 2023 to November 13, 2026, over 713 trading days, the model's daily moves explained 96 percent of the variation in the daily price moves of TLT, and for a 1 percent move of the model the fund moved 0.98 percent on average. A comparison of this kind covers one period and describes daily moves: it does not show the fund's return or its price. The period and the treatment of a fund's distribution days can differ from fund to fund, so it is not a ranking of one fund's model against another's.

Not investment advice. The tools show calculations on published data. They are not recommendations to buy, sell or hold anything.