1
$\begingroup$

I have been trying to explore the possibility of replacing my IEF (10 years treasury ETF) positions with ZN (10 years treasury futures) for better leverage.

Reading the posts here, I understand that their return should be similar, except for the drag of financing cost. Not an issue for me as I will hold some short-term treasury to compensate for that.

In order to prove that their returns are indeed similar, I got their prices from the two sites below and computed the cumulated return.

https://www.nasdaq.com/market-activity/futures/zn/historical

https://finance.yahoo.com/quote/IEF/history?p=IEF

It turns out that they are very different and the difference (approximately 4% a year) is far beyond the financing cost. Please let me know what I am missing here. Am I fundamentally wrong about how I use the futures prices? Thanks.

enter image description here

$\endgroup$
0
4
$\begingroup$

I ran some quick simulations and the differences don't seem particularly drastic:

enter image description here

The black line above is the cumulative total return (inclusive of dividends) of IEF. The yellow line is the so-called "excess return" index for TY (aka ZN), which is the cumulative return of buying and holding TY contracts. To compute this index, I assume that you buy and hold the front-month TY contract until a week before the delivery month, at which point you roll into the next contract. Finally, the green line is the "total return" version, which is simply TY's excess return index with return on cash added back (i.e., it assumes that your futures positions are fully collateralized). The annualized return difference between the two total return indices is <70 bps (I used fairly conservative cash return assumptions and the differences will be even smaller for most institutional investors).

The blue line is likely what you retrieved from NASDAQ. It's simply the rolling front-month TY contract prices. The problem is that this series doesn't properly account for the roll between contracts – if you roll from a contract priced at 120 to another priced at 119, you don't lose a dollar, but that's what that time series would suggest.

The underperformance of TY is to be expected, since IEF tracks the 7- to 10-year part of the curve, while TY has generally tracked the shortest bonds in the delivery basket thanks to the low yield environment. The chart below shows the time to maturity of the cheapest-to-delivers of the front-month TY contracts since 2010.

enter image description here

For an alternative perspective, the next chart compares TY's duration against the duration of Treasuries in the 7-10y sector (based on index-weights).

enter image description here

$\endgroup$
5
  • $\begingroup$ (Finallly an answer from someone who knows what they are talking about, and even does some calculations her/him-self (thank you) instead of guessing). The authors of the zero-rated answers should seriously consider improving or deleting their answers. $\endgroup$
    – noob2
    Dec 31 '20 at 7:57
  • $\begingroup$ Really great answer. Perhaps I should mix the ZN with the Ultra 10 to get the right duration. I would love to perform the same simulation. May I ask if you are using proprietary data for the charts, if not, do you mind sharing the source? Thanks. $\endgroup$
    – Usal
    Dec 31 '20 at 16:57
  • $\begingroup$ @Usal Glad it helps! Yes using a combo of TY & UXY would reduce tracking error. You need contract level pricing instead of these "generic" series. We use data from CME, but I surmise a lot of sites have it (e.g., barchart off the top of my head). $\endgroup$
    – Helin
    Dec 31 '20 at 19:15
  • $\begingroup$ adding back the return on your cash collateral makes the two series equivalent for the purpose of the backtest but if your goal is leverage you'll want to have a look at what the margin and margin rates are compared to the futures. This is a big benefit to futures as you can get institutional leverage and rates versus leveraging etf's. $\endgroup$ Jan 3 at 22:15
  • $\begingroup$ @EdwardWatson Yes thank you, I should've noted this. Although you'll be surprised at how many institutions (particularly endowments and pensions) use futures in an unlevered (i.e., cash collateralized) fashion. $\endgroup$
    – Helin
    Jan 4 at 4:56
1
$\begingroup$

IEF as an ETF will also have management costs. Also the duration of IEF is lower since it is holding a basket of 7-10 Yr US Treasuries vs a 10 Yr note future, which is a future on just the 10Yr Note (actually a 10Yr 6% Note). There may be some optionality, such as Cheapest-to-deliver, at play with the future.

Also, you will incur roll risk and costs of the futures if your strategy is to hold the position for an extended period of time. The ETF will also incur costs to as bonds roll down and no longer meet the maturity requirements but these costs will be different from the roll costs, which will occur quarterly if you are investing in the near futures.

$\endgroup$
3
  • $\begingroup$ For ZN, I believe the duration should be similar as the deliverable maturity is also 6.5 to 10 years, not strictly 10 years unlike the Ultra 10 (TN). If the ETF also has the management fee, should it have a lower return compared to the future? $\endgroup$
    – Usal
    Dec 30 '20 at 14:28
  • $\begingroup$ Not necessarily. The management fee will tend to lower the returns but may not offset the portfolio effects. If they ETF tends to be on a more favorable part of the curve over the time frame, the ETF may still outperform despite the management fee. $\endgroup$
    – AlRacoon
    Dec 30 '20 at 15:24
  • $\begingroup$ Thank you for your insight. Still not sure if it explains the big gap between the two returns. The futures' cumulative return is just above one for the past 10 years. Probably something wrong with how I use the future data. $\endgroup$
    – Usal
    Dec 30 '20 at 15:51
0
$\begingroup$

In addition to adding the return for your invested cash you have to roll the futures every quarter buy selling the front month and buying the back month a few days before the delivery cycle starts. You'll need some different contract specific data to do that. That should get you pretty close.

$\endgroup$
2
  • $\begingroup$ I have replaced the link for ZN data. I am not sure if I am calculating the return incorrectly or if it is simply a wrong piece of data for my purpose. $\endgroup$
    – Usal
    Dec 30 '20 at 14:35
  • $\begingroup$ @Usal - the new link does not appear to show a continuous, back-adjusted contract for ZN. $\endgroup$
    – user42108
    Dec 30 '20 at 23:35
0
$\begingroup$

You need to include the Roll Yield https://www.investopedia.com/terms/r/roll-yield.asp#:~:text=Roll%20yield%20is%20the%20return,premium%20to%20longer%2Ddated%20contracts.

Future contracts are usually rolled 4 times a year. When you do the roll, there is a price difference between contracts, and you need to include that. CBOE has a "Pace of the Roll" Tool which is really useful for such problems.

With roll yield added, adjustment for duration and considering the interest cost for the two cases, the return is the same.

$\endgroup$
1
  • 1
    $\begingroup$ Pace of the Roll (which is a CME/CBOT tool, not CBOE) shows the shift in volume or OI. I don't think it's much use for constructing a continuous, back-adjusted contract (you could just look at settlement prices around FND if that's your goal). $\endgroup$
    – user42108
    Dec 30 '20 at 23:37

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.