Skip to content
All library documents

Distinguishing Pure Carry from Expected Returns in Treasury Futures

Article Quant Q&A · Author: Yannick

Summary

The document explains why treasury futures carry cannot be obtained simply by subtracting an implied repo rate from a deliverable bond’s yield. Under a strict definition, pure carry consists of known cash inflows and outflows from holding and financing an asset. A futures position has neither the bond’s coupon receipts nor the financing cost during the contract period, since these effects are reflected in its pricing, so its pure carry is treated as zero.

A broader interpretation may include expected futures convergence toward spot, bond rolldown, delivery-option decay, and convergence from relative mispricing. These components can contribute to expected returns if market conditions remain stable, but they are not all tangible or deterministic; delivery mechanics and other contract features can affect convergence. The response recommends considering these components separately. It gives conceptual relationships rather than a worked calculation or a specific yield estimate, and the broader carry label is not used consistently across market participants.

Key ideas

  • Pure carry for a financed bond position reflects coupon income net of financing cost.
  • Treasury futures prices already account for the underlying bond’s carry, so the futures holder does not receive coupons or pay repo financing while holding the contract.
  • A strict definition therefore assigns zero pure carry to the futures position itself.
  • A broader expected-return view may include convergence, rolldown, delivery-option decay, and relative mispricing.
  • These broader components are less certain and should be assessed separately.

Tags

Full text
# Determine the carry of a treasury bond futures contract?


# Determine the carry of a treasury bond futures contract?












Hi fellow financial market enthusiasts. I'm trying to understand my options as a retail investor. I want to leverage a cash bond portfolio but my broker does not allow that, so I want to use futures instead. How can I determine the carry of a treasury bond futures?

I can read from other posts that you need to do [CTD Yield] - [Implied Repo Rate].

So in our case right now on the ZF contract, we do 2.60 - 0.65 in order to get around 1.95% yearly carry if I buy a contract?

See attached snapshot of the CME treasury analytics page.

Can you help me understand this please. It doesn't need to be exact but at least a ballpark yield percent with .05 margin of error is fine.

Update: Thanks for the answers, really appreciate the help. Theoretically, as pointed out by Daniel below: because taking delivery only takes place in 3 months time the position would not accrue any income till futures delivery date.

Still, it doesn't mean the futures doesn't have a positive carry. This paper from CME illustrate my point for S&P500 futures: http://www.cmegroup.com/education/files/S-and-P-500-Implied-Financing.pdf - refer to "Positive and Negative Carry" section.

I assume same is valid for treasury futures: short-term interest rates being cheaper than the yield on the underlying bonds, resulting in a positive carry.

Update2: To further illustrate my point, I went through a rudimentary estimation exercise. According to my logic, measuring the difference in price of the ZF contract (mar18) when the yield is the same should give us a good approximation of the carry.

I'm looking for a more convenient way of doing this calculation.

## Answer by Helin (score 11, accepted)

https://quant.stackexchange.com/a/38144

Based on the your comments, I believe the issue lies with what you consider to be "carry." The reality is that there's no consensus. So let's take mini steps.

We'll start with what rates guys consider as "pure carry." In this most classical and fairly strict definition, carry is the deterministic component of expected returns – you know exactly what it is before you enter the trade. It is also very tangible, involving clear cash inflows/outflows. In this sense, @Daniel's answer is 100% correct as is: forwards, be it a forward bond position or even a forward starting swap, have no carry (no "pure carry" anyways).

There are several ways to think through this. First of all, recall that the futures price (excluding the embedded switch options, forward/futures difference, and other technicalities) is $$\begin{align*} \text{Forward Price} &= \text{Spot Price} - (\text{Coupon Income} - \text{Financing Cost}) \\ &= \text{Spot Price} - \text{Pure Carry}.\end{align*} $$ This simple formula (and its equivalents) applies to all forward contracts. The futures price literally is the net result after the carry of the underlying has been removed.

Secondly, there is no tangible cashflows of any kind. By using a futures contract, you forgo coupon income from the underlying; nor do you pay a financing cost. Those have already been factored into the futures pricing.

Thirdly, as @Daniel has pointed out, carry basically provides you with a cushion – if carry is positive for a bond, yields can rise a little bit (by an amount equal to the difference between forward yield and spot yield) before you start losing money. For a futures contract, there's no such cushion at all. Yields start rising, you start losing money, because there's no coupon income to mitigate the capital losses.

This is not to say that there's no expected returns when you hold a futures contract and the world is static – now we're expanding the scope of the word carry. For clarify, I'll refer to this definition as "broad carry" – expected returns of an instrument when the world remains unchanged. When you allow for this broader definition, many things start to count.

For example, as you have pointed out, futures converge toward spot (by an amount equal to the underlying's pure carry). You can consider this to be a form of carry (I do!). Why isn't this "pure carry" though? Because it's neither tangible nor deterministic. It's not tangible because there are no real cashflows. It's not deterministic because bond futures allow for delivery one week after trading stops, so the classical futures/cash convergence may never happen.

Going further, bonds have expected rolldown returns that will flow through to futures – that could also count as a form of carry (some people do, others treat it as a separate concept). Bond futures also have an embedded delivery option, which can have time decay just like any other option. Bond futures may be mispriced relative to cash bonds, creating another source of convergence (toward fair value).

Anyways, I agree with @Daniel that strictly speaking, the carry of futures of zero. But if you're trying to think through what your expected returns might be if the world is unchanged, it's a much broader/messy definition. Depending how you trade and how you hedge, you are free to make discretionary decisions on what you want to count toward this "broad carry." It is probably best to list them out separately, so that you have a better idea about how reliable each component may be.

## Answer by Daniel (score 2)

https://quant.stackexchange.com/a/38059

If you think about carry as a cushion against a change in the forward yield then carry (in basis points) for the underlying bond equals with (coupon income of the bond - repo rate) / forward DV01 of the bond. (Carry could be also calculated as the forward yield - spot yield) That is how much the forward yield can rise before you start loosing money on the financed bond position.

We can apply the same logic for a bond futures contracts as well. Being long in a treasury futures means you are buying a treasury bond in 3 months time which is priced by the forward yield curve to get the present value (~or futures price) of the bond at the futures delivery date. Financing a futures contract is virtually zero and because taking delivery only takes place in 3 months time your position would not accrue any income till futures delivery date. Therefore your net carry is zero and your cushion against any changes in the forward YTM of the bond is zero.

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.