If swap's fixing days, eg two business days, dont coincide with to libor's settle days, eg one business day, there will be two types of fwd rates one used for interest and the other for discount. which one should be used in the libor market model? whats the market convention? are they same? thx in advance!

I think the convexity adjustment that would arise from having a slight lag between the swap fixings and libor fixings is probably negligible compared to other factors in the model and that it would be reasonable to approximate by assume that the fixings coincide.

Thx AlexesDad, I think mismath always exists except for assuming actual accrual periods of floating leg coincide with periods of underlying libor rates. floating leg of swap will be priced at par if taking fwd rates projected from the swap curve under this assumption. any mismatch is nightmare for implementing libor market model cause it cant re-price vanilla swap perfectly as u suggested.

Neglect these kind of details:When pricing a vanilla swap these sort of adjustement change the value (upfront) by a few bps.If you were trading just swaps, these differences would seem huge (swaps trade tight). But you don't use these heavy models to trade simple swaps: the derivatives you will price with LMM trade at a margin big enough that make a few bps look like 0.

thx Clopinette. if as u said, two methodologies should be contained in pricing ir portfolio, one considering these sort of adjustements is used to price vanilla swaps and the other neglecting adjustements are used in libor market model to price complex products. analogously two swap curve shoud be bootstrapped according to different assumptions.

No you can't possibly bootstrap as many swap curves as you have models...There would be really no point, plus it would be costy.Try to see it this way: - There is only one swap curve: ==>The one that the swap desk uses- There are plently of models one can use for derivatives: ==>Each of them have weaknesses and strengths. Each of them will misprice something. For example Hull White will reprice swaps but will misprice plenty of swaptions. LMM may reprice swaptions but is a little imprecise one plain swaps.Again no one uses LMM to value swaps portfolios...

but which model shoud be choosed to risk-manage the portfolio containing both vanilla swaps and complex products (must resort to libor market model)? I think the most important thing in risk-managing a portfolio is to apply the same inputs and the same pricing methodology.

Usually the model choice is done by product rather than by portfolio. Because it makes more sense.So in the same portfolio you will have different products, priced with different models.Plus, it is likely that the swaps won't be in the same portfolio as derivs anyway.

If you use swaps to delta-hedging complex products, I guess, these shoud be included in the portfolio.

Generally speaking, when pricing derivatives, one should worry about what matters most, knowing that:1/ You will never get the perfect price or hedge.2/ You only have limited hardware and time to get your risk.Because of 1/ you can forget about a 2bps mispricing, because your product margin (which accounts for these pbs) is about 100 times that.About 2/: try to explain your trader that you have chosen to price all his swaps with LMM, and so, from now on, his risk will take 3 times longer and will be done using montecarlo. Explain as well that, following this decision, you will need more resources to work on stabilising the risk coming out of the montecarlo. See what he says....

ur right I agree...