wilmott.com

I'm trying to price with the same curve two kind of IRS:- fixed rate vs. EURIBOR6M- fixed rate vs. EURIBOR3Mbut I'm facing a problem. The fixing of the EURIBOR3M gives me the effective zero rate of my curve for maturity 3M. The fixing of the EURIBOR6M gives me the effective zero rate of my curve for maturity 6M. This implies the forward rate for the period 3M->6M but this value is inconsistent with the market value of this foward rate given by futures and fras.If I forget about this inconsistency I have a bad pricing for the IRS with EURIBOR3M. Of course, if I do not consider the EURIBOR6M fixing and build it using the 3M fixing and the forward market rate, I make mistakes in the IRS with EURIBOR6M.One solution is to use two different curves for EURIBOR6M and EURIBOR3M. But is there some way to solve the problem while still using one single curve?

I was hoping for someone who knows exactly what he is talking about to respond to your question, but since this person does not seem to exist, I have to offer my own pitiful suggestion of what could be going on. Currently you are seeing a spike in 3M vs 6M basis in EUR which causes the discrepancy between the forward rate implied by deposit rates and by futures. If you incorporate this basis in your curve building procedure, the problem will hopefully disappear.Dont ask me the specifics...I am not a curve building expert unfortunately.

Yes, but I think that Incorporating the 3M vs 6M basis is like having two curves, one for 3M and the other for 6M.Indeed, to calibrate the basis before t = 1 Y (where you do not have market data) you just define it to be compatible with the 2 curves obatained from the available sets of market data for 3M and 6M separately.Even if I restrict the problem to just one floating rate, that is I magine to have only 6M vs. fixed rate, I'm facing a problem.- When the start date is in the next week I would like to have a first fixing that is nearly equal to the EURIBOR6M fixing of today.- When the start date is later I would like to have the first fixing that comes from a curve built with standard instrumentsTo satisfy these requirements I build the curve with the fixing of today, 6M fras, and than IRS.The problem is that when I have some stub period (expecially at the beginning), that needs interpolation between some other rate (like EURIBOR3,4,5 M) I obtain from my 6M curve very bad estimates for them.I think the only way to solve completely this problem is a "very intelligent" pricing machine that have 3M, 4M, 5M, 6M curves and use them all together to price a 6M vs. fixed rate irs with irregular periods. But this pricing machine is not compatible with many commercial software that expects as input only a single zero rate curve. Moreover, using more than one curve seems like violating no arbitrage constrains.Any suggestion?

Unless you want to spend the rest of your life solving this, I'd recommend you just stick with two different curves (e.g. 3m curve and a basis curve). In the current environment, and this is my strongly held personal point of view, there is NO way to build a curve that will reprice fixings correctly. Believe me, I have tried.For example, to illustrate another problem you will run into, why are you assuming that next week's EURIBOR 6m fixing is going to be close to today's? Most recently that had been patently untrue. In order to account for the very front-end term structure of LIBOR, you're gonna have to use some more curves.Basically, abandon hope and do what pretty much every bank on the street does, have 3m and 6m curves and play it by ear for the other stuff... Otherwise, if you do figure it out, let me know and I'll be the first to say you're a God!