Forums  > Pricing & Modelling  > Dynamic subordination index tranches  
     
Page 1 of 1
Display using:  

day1pnl


Total Posts: 47
Joined: Jun 2017
 
Posted: 2018-06-01 10:39
Anyone in here been trading and modeling dynamic subordination in synthetic tranches? I know its quite common in cash land (callability and or other tranche paydowns). Would you price a step up subordination feature as a series of forward starting tranches and if so how do you hedge it?

Ill start of. I buy protection on iTraxx MAIN with the following subordination schedule

year 1 0-1%
year 2 1-2%
year 3 2-3%
year 4 3-4%
year 5 4-5%

Of course at this point it's ridicolously undefined what i mean by this.

I could buy 5y protection on the 4-5% and sell 4y 4-5% protection against it, and so on for the other tenors prior to maturity. If there are 5% losses in year 1, the 4-5% tranche will have already become equity and also subsequently wiped out. So in a long/short combo of tranches, the attachment points would in general be "floating" (because the attachment on the forward starting leg depends on defaults in first periods)

Another way is to trade it with "fixed" attachments. I buy a series of "truly" forward starting tranches, where last leg would be i buy 1y protection on 4-5% tranche starting 4 years from now (which isnt the same as above).

Price this as a 5y 4-5% tranche with fixed-to-floating recovery, having recovery fixed at 100% in year 1 to 4 prior to the forward starting date, and then reverting back to "normal" floating recovery. Similar for the other tenors. Do you agree?

How would you have approached this?

Any thoughts appreciated

day1pnl


Total Posts: 47
Joined: Jun 2017
 
Posted: 2018-06-10 21:13
Among structured credit clients it seems to resonate that for mezzanine tranches, the probability Q(L > K) that losses will exceed the attachment point is too high. They are therefore consistently short this probability (i.e. sell mezzanine protection).

Thats one way for a client to set his tranche attachment points. Another investors thought process could be something like "I want to be exposed at the x%-quantile in the loss distribution and I want it with X amounts of leverage => (i.e. find me a strike where Q(L > K)=x%)".. To target this he just call up your dealer/structurer and make him buy protection there at the relevant attachment points.

However its quite clear that he's only getting the "correct" quantile in the loss distribution at the maturity point. If you want to have exposure to a certain quantile in the loss distribution on a running basis (which would make sense), then at the starting date you would have to have a series of forward starting tranches or some dynamic subordination feature.

Thats raison d'être for the dynamic subordination.

Now I think these two papers can help me get going - does anybody have access to them?

"Step it Up or Start it Forward" - Prasun Baheti, Roy Mashal and Marco Naldi (J. Fixed Income 2006)
"Forward‐Starting CDO Tranche" - Prasun Baheti Roy Mashal Marco Naldi (J. Fixed Income 2010)


Cheng


Total Posts: 2856
Joined: Feb 2005
 
Posted: 2018-06-12 13:36
Among structured credit clients it seems to resonate that for mezzanine tranches, the probability Q(L > K) that losses will exceed the attachment point is too high. They are therefore consistently short this probability (i.e. sell mezzanine protection).

I would suggest a much simpler reason. Clients are probably not allowed to invest in equity tranches (risk management doesn't like being the first in the firing line, even if the spread overcompensates you for that), so the junior mezz is the next possible alternative. Otoh you want to maximize the spread you pocket so junior mezz it is. I have never heard (outside quant circles) that anyone buying that stuff thought about loss distributions ans probabilities.

"He's man, he's a kid / Wanna bang with you / Headbanging man" (Grave Digger, Headbanging Man)

Cheng


Total Posts: 2856
Joined: Feb 2005
 
Posted: 2018-06-12 13:39
Forward-start tranches (of whatever flavor) were just a way to increase the spread you could pocket while arbing the rating. I have never seen such a thing printed. The issue imho is liquidity. 5y is reasonably liquid (at least is used to be), 3y... weeeelll... and don't even think about 1y, 2y and 4y. So your forward loss distribution is probably messed up to a certain extent and hedging is nigh on impossible. Not exactly a nice position to be in.

"He's man, he's a kid / Wanna bang with you / Headbanging man" (Grave Digger, Headbanging Man)

day1pnl


Total Posts: 47
Joined: Jun 2017
 
Posted: 2018-06-20 18:21
thx for pitching in.

Though the tranches are not rated, name of the game is of course still to enhance the spread you can pocket by more closely tracking the expected loss rate with your attachment points

there are a few ways in whihc i can construct a loss distribution in the invisible 1y, 2y, and 4y "fibers" of my corr surface. At the end of the day, it relies on some more or less bold assumptions and so will forward loss distro, agreed. And wonder anomalies will be more pronounced in the forward loss distribution than in the constructed "spot" loss distributions...

day1pnl


Total Posts: 47
Joined: Jun 2017
 
Posted: 2018-06-20 18:21
Edit: Double post
Previous Thread :: Next Thread 
Page 1 of 1