Something I Noticed in The Tbond Futures Roll
Alex Castaldo writes:
Last Friday was the day to roll long positions in ZB futures: to sell
the December futures (which are nearing expiration) and buy the March futures instead. I noticed something a little puzzling. For the last 2 years the far away (new contract) future was cheaper than the nearby one (the old contract). But last Friday it was the opposite:
2PriceDate old contr new contr oc price nc price roll cost
05/30/2018 ZBM8 ZBU8 145 14/32 144 19/32 - 27/32
08/30/2018 ZBU8 ZBZ8 144 31/32 144 7/32 - 24/32
11/29/2018 ZBZ8 ZBH9 140 4/32 139 16/32 - 20/32
02/28/2019 ZBH9 ZBM9 145 4/32 144 15/32 - 21/32
05/30/2019 ZBM9 ZBU9 153 2/32 152 14/32 - 20/32
08/29/2019 ZBU9 ZBZ9 166 2/32 165 8/32 - 26/32
11/27/2019 ZBZ9 ZBH0 160 3/32 159 10/32 - 25/32
02/27/2020 ZBH0 ZBM0 168 16/32 167 15/32 -1 1/32
05/28/2020 ZBM0 ZBU0 178 21/32 177 2/32 -1 19/32
08/28/2020 ZBU0 ZBZ0 176 14/32 174 25/32 -1 21/32
11/27/2020 ZBZ0 ZBH1 173 28/32 175 1/32 +1 5/32
As long as short term interest rates (repo rates) are positive, it would seem that an object delivered 3 months further away should be cheaper than the same object delivered 3 months sooner. (The good old Time Value of Money). Which makes me think that the Cheapest to Deliver for ZB March 2021 must be different from the CTD for ZB December 2020 if the March is priced higher? But I am not sure if this explanation is correct. And I find it disturbing that even though I traded tbonds for a while I do not fully understand some of the basic mechanics. Do you have any insight? Why did the price difference (technically know as the Roll Cost) flip like this?
George Zachar writes:
On Bloomberg, pull up USZ0 and USH1 CMTY DLV.
The cheapest to delivers did change:
Z0 = 4.5% '36
H1 = 5.0% '37