Hi, can someone please help me with some pointer about how the swap curve is constructed in bloomberg. to build the curve say i start with historic data and put just 1 swap instrument with a 2y maturity. i choose interpolation of of linear continuous. to be concrete lets take the curve date to be 02/09/1998 and put just one 2Y swap intrument which has a rate of 5.72% then what i expect is the discount factors of the curve will be such the fixed leg of a swap (which rate 5.72%) which i can define in the SWPM screen will be equal to the notional. this is what i do not get. if I take a huge notional (say 100 trillion ) to see many digits of the computations i do not get the exact result. the discount factors which i get from the cash flows are ball park okay but not very good (of the order of 10^-14 or so) when compared to my manual computation. this is an extremely simple case and whatever BBG gives sould be reproducible with very high precision. can somebody please help. thanks in advance

help bberg out and put a 3m depo in as well

in fact i have done that but still did not get good result

ok - give me the 3 mth rate and 2y swap rate, day count and freq

the 3m rate is 5.625% the 2Y rate is 5.72%the conventions are.FIXED FREQUENCY 6MFLOATING FREQUENCY 3MFIX CALENDAR NewYork LondonFIXED ACCRUAL CONVENTION fFIXED PAY CALENDAR NewYork LondonFIXED PAY CONVENTION FFLOAT ACCRUAL CONVENTION FFLOAT PAY CONVENTION FNOTIONAL 10,000,000,000,000.00the basis is act/360the dates and the corresponding day year fractions are (02/11/98) is the valuation date.11-Feb-98 11-May-98 0.24722222211-Aug-98 0.25555555612-Nov-98 0.25833333311-Feb-99 0.25277777811-May-99 0.24722222211-Aug-99 0.25555555612-Nov-99 0.25833333311-Feb-00 0.252777778BBG is giving the followingPayment Dates Payments(Rcv) Payments(Pay) Net Payments Discount Net PV dicount(=net PV/net) Reset Dates5/11/1998 0.00 -139,062,500,000.00 -139,062,500,000.00 0.986284 -137,155,185,698.88 0.986284481 03/17/19988/11/1998 286,000,091,791.2 net 3 -143,610,790,201.97 142,389,301,589.26 0.972321 138,448,094,454.50 0.972320904 06/17/199811/12/1998 0.00 -145,004,541,513.12 -145,004,541,513.12 0.958423 -138,975,735,619.20 0.958423331 09/17/19982/11/1999 286,000,091,791.23 -141,690,477,114.28 144,309,614,676.95 0.945033 136,377,364,147.04 0.945033111 12/17/19985/11/1999 0.00 -138,388,818,020.26 -138,388,818,020.26 0.932133 -128,996,843,195.45 0.932133427 03/17/19998/11/1999 286,000,091,791.23 -142,914,088,226.80 143,086,003,564.43 0.919000 131,495,983,988.13 0.918999628 06/17/199911/12/1999 0.00 -144,300,170,184.64 -144,300,170,184.64 0.905927 -130,725,432,445.00 0.905927084 09/17/19992/11/2000 10,286,000,091,791.20 -10,141,001,478,190.20 144,998,613,600.95 0.893331 129,531,754,368.85 0.893330985 12/17/1999what I am trying to reproduce in my manual calculation is the 7th column at high accuracy. i could get the first line of 7th column = 1/(1+5.625% * 0.247222222) this result is upto 10^-14 accuracy after that everything breaks down. the interpolation is linear in continuous zero thanks for any help in advance

sorry the formatting of the table messed up, could not figure out how to post with better formattried to attach picture, tried to attach excel everything failed.

ok - this what I get from my modelDate Discount factor Fixed flows PV Floating flows PV Net Net PV11/05/1998 0.986284481 139062500000.26 137155185699.13 -139062500000.26 137155185699.1311/08/1998 0.972320573 286000000000.00 278083683882.01 143614244145.66 139639084160.69 142385755854.34 -138444599721.3211/11/1998 0.95857063 143442143633.40 137499426049.89 -143442143633.40 137499426049.8911/02/1999 0.945031165 286000000000.00 270278913065.11 143270046041.06 135394658457.22 142729953958.94 -134884254607.8911/05/1999 0.932130268 138402287110.83 129008961040.00 -138402287110.83 129008961040.0011/08/1999 0.918994939 286000000000.00 262832552437.93 142931471257.48 131353298651.44 143068528742.52 -131479253786.4911/11/1999 0.906060081 142759382329.41 129348577494.72 -142759382329.41 129348577494.7211/02/2000 0.893322438 286000000000.00 255490217191.01 142587296321.06 127376431138.97 143412703678.94 -128113786052.04PV = 90256116.00%age opf notional = 0.0009%the fixed leg of the swap is 30/360 I think not ACT/360.

well you see you have a bigger difference compared to BBG. that is the issue which i am trying to resolve. for the last date.date your DF my DF BBG DF11/02/2000 0.893322438 0.893331018 0.893330985 so you see your model is further off from BBG than mine. but the point is both of us are off compared to BBG. This is what is confusing for such a simple model. All of us should get the numbers different by at most machine precision any thoughts ?

i could try tightening up the tolerance that I use when solving ... will let you know

I have noticed that the cashflows from the fixed leg don't match. if the swap if 30/360 then the cash flow for each period should be180/360 * 1e13 * 5.72% which is 2.86e11. bloomberg shows 286,000,091,791.23 but this is a minor issue

that is another mystery which I also observed. but any luck with the discount factors ?

no - i have tried various combinations of linear/log-linear/const and can't match BB's discount factors.

QuoteOriginally posted by: daveangelno - i have tried various combinations of linear/log-linear/const and can't match BB's discount factors.Is this a potential opportunity here? I.e. non sophisticated members pricing wrongly?

any help from any sophisticated members ?