Treasury Futures Margin Over/Understated?

AndrewJackson · May 10, 2011

Hi there,

I have been reading these forums off and on for the last couple months. I thought that you guys might be able to give this question a shot. It is a scenario based question. Here it goes.

Assume you are long 1 10 year treasury futures contract at 105 on Monday. On Tuesday, treasury futures rally 106. The cheapest to deliver bond has a conversion factor of 0.9 and it is nearly guaranteed to remain the cheapest to deliver. On Tuesday, your account will be credited $ 1000. Here is where the question comes in. If I intended to take delivery of the bond, the change in value of the cheapest to deliver bond would actually be equal to (106-105)*.9*1000 = 900. Why the difference? If I am long the treasury future, I get an extra $ 100. On the flip side, if futures fall by 1 whole point, I would have an extra loss of $ 100. What are the implications of this and is there any trading strategies that can be done to profit from it? Thanks

Martinghoul · May 11, 2011

Do you want the answer or do you want to read a book that can make you understand all this, so that you can figure all this stuff out on your own? Also, why do you talk about margin in the thread's title? Your question doesn't appear to have anything to do with margins.

AndrewJackson · May 11, 2011

Quote from Martinghoul:

Do you want the answer or do you want to read a book that can make you understand all this, so that you can figure all this stuff out on your own? Also, why do you talk about margin in the thread's title? Your question doesn't appear to have anything to do with margins.

My question really boils down to, why does the exchange set the profit/loss (variation margin) on the futures = to soly the change in the futures price when in fact the actual profit of being able to take/make delivery on the bond = change in futures * CTD CF? The only thing I can think of, is that making the profit/loss on the futures contract a constant amount per point is the only practical way for the exchange to set up. Most commodity futures (taking aside the time value of money) have the profit or loss on the contract = to the actual change it will take to purchase the commodity upon delivery. Can you see where I am coming from now?

Martinghoul · May 12, 2011

Quote from AndrewJackson:
My question really boils down to, why does the exchange set the profit/loss (variation margin) on the futures = to soly the change in the futures price when in fact the actual profit of being able to take/make delivery on the bond = change in futures * CTD CF? The only thing I can think of, is that making the profit/loss on the futures contract a constant amount per point is the only practical way for the exchange to set up. Most commodity futures (taking aside the time value of money) have the profit or loss on the contract = to the actual change it will take to purchase the commodity upon delivery. Can you see where I am coming from now?

Well, you're incorrect. Change in futures * CF is your fwd PNL out of delivery date. You will need to "discount" to obtain the spot PNL. The reason the commodities types can get away with it is that it's a more reasonable (although still overly simplistic and wrong, but it's practical) assumption that the "discount" rate and the fwd price of the asset are independent. Not the case for bond futures. So the issue is precisely the "time value of money", which you can't take aside when you're talking about interest rates.

As a note, the CBOT/Eurex/TSE/LIFFE bond futures design isn't the only way. ASX Aussie bond futures work differently, for example, but they're an exception, rather than the rule.

Finally, I really can't recommend "Treasury Bond Basis" by Burghardt, Belton et al enough. It's a must read, if you wanna understand the mechanics of this mkt properly.

AndrewJackson · May 12, 2011

Quote from Martinghoul:

Well, you're incorrect. Change in futures * CF is your fwd PNL out of delivery date. You will need to "discount" to obtain the spot PNL. The reason the commodities types can get away with it is that it's a more reasonable (although still overly simplistic and wrong, but it's practical) assumption that the "discount" rate and the fwd price of the asset are independent. Not the case for bond futures. So the issue is precisely the "time value of money", which you can't take aside when you're talking about interest rates.

As a note, the CBOT/Eurex/TSE/LIFFE bond futures design isn't the only way. ASX Aussie bond futures work differently, for example, but they're an exception, rather than the rule.

Finally, I really can't recommend "Treasury Bond Basis" by Burghardt, Belton et al enough. It's a must read, if you wanna understand the mechanics of this mkt properly.

Thanks Martinghoul. Obviously, spot profit would have to include a pv of the forward profit. The futures will have a beta greater > 1 (cf of ctd over 1) or beta < 1 (cf of ctd less than 1) with respect to the ctd bond. This is due to the way gains/losses on the contract are settled on the futures contract.

Martinghoul · May 12, 2011

Quote from AndrewJackson:
Thanks Martinghoul. Obviously, spot profit would have to include a pv of the forward profit. The futures will have a beta greater > 1 (cf of ctd over 1) or beta < 1 (cf of ctd less than 1) with respect to the ctd bond. This is due to the way gains/losses on the contract are settled on the futures contract.

Yep, that's just the mechanics, specific to bond futures.

In general, all futures, including commodities, should have PNL/variation margin calculated as (futures tick value * futures move in ticks, rather than being based on the move of the underlying. So I imagine, actually, in this sense bond futures are the same as all others.