Hi, can someone pleae help me with a jargon I keep coming across. in the ocntext of constructing discount curves or pricing derivates i come across this phrase "roll the discount curve". now is this a standard pharase which people use to mean something. i understand my question is very vague, but any help is greatly appreciated. thanks

Hi babu12,I'll offer my guess: to roll a curve from now to time t means to divide all the discount factors with the discount DF(0,t). Of course, you may want to discard the discount factors between now and time t, as they will represent discounts for things that happened in the past. Some people do accept curves that extend both in the future and the past, so that's up to you. The reason I said this is just a guess, is that people I work with don't use this term, and say "to forward" a curve instead. But since different institutions have different internal jargons, it is possible that "rolling" the curve is the equivalent term in some other places. Best of luck,V.

This is an idea originated by old school bond traders. It is a fairly non-rigorous exercise in playing a "what if" game. The game assumes the yield curve is upward-sloping (as it often is) and goes as follows : if the yield curve doesn't change whatsoever as time moves forward, what happens to the value of a fixed income instrument? As a bond or swap ages it will by definition be revalued in the future at successively lower yields if the yield curve is upward sloping. This what-if game is clearly simplistic as the laziest thing one can assume about the future yield curve is that it will not change whatsoever. A very old school game that is not very enlightening.Rolling the curve is also mixed up with the concept of the "carry" of a position. See the recent discussion in the technical forum on "The Carry of an IRS" (or somehting like that) for more details.

Better to work with the forward curve than the yield curve for consistency. As time progresses just chop off rates from what are now the past and then recalculate yields, bonds, etc. from this shifted curve. It's like saying that instantaneous forward rates are fixed, which of course in practice they aren't. P