I usually use ON Libor fixing to bootstrap the USD overnight curve from t+0 to t+1, then some improper quote as tomorrow-next on [t+1, t+2], then finally use OIS on [t+2, (t+2)+mY] with m= 1, 2, ..., 50YNext Mon 25-Aug-2014 is a London holiday, but not a NY holiday. To the best of my understanding OIS will all start on Wed 27-Aug-2014, as they are not affected. Is it right?What can be used to cover from Mon 25-Aug-2014 to Wed 27-Aug-2014?Libor ON fixing will not be available because of the London holiday, and I would like to fix the [t+0, t+2] part of my overnight USD curve once forever.What are the best practice and conventions?thanks

Why do you use O/N LIBOR as an input into your OIS curve construction? Clearly, if you do such a thing, you're going to run into issues such as you're describing.

I don't really want to use ON Libor, I'm just requested to have the ON curve to start at t+0, while OISwaps start at t+2.I am considering to bootstrap the OIS curve starting at t+2 and then do some left-extrapolation to feed the systems requiring a curve starting at t+0, but wanted to check if there were homogeneous instruments that could be used to cover [t+0, t+1] and [t+2, t+2]

Sure, I understand and, ironically, there's another thread on this subject in another forum. I think you have the right idea to use extrapolation.I just don't really understand why you'd wanna use O/N LIBOR, rather than the FF effective fixing.

This is the thread Martinghoul is referring to. I would use FF Fixing from t to t+1, TomNext from t+1 to t+2 and ois swaps thereafter. Or does this cause problems as FF Fixing is not liquid enough for trading? I always think about this stuff from a theorists point of view ...

However liquid the FF mkt might be (and it's reasonably liquid, last I checked), it's certainly a lot better to use the FF effective fixing than O/N LIBOR. In fact, using O/N LIBOR in the context of a FF OIS curve is downright strange.

QuoteOriginally posted by: BerndSchmitzThis is the thread Martinghoul is referring to. If I read correctly the referenced thread is about EONIA, not Fed Fund effective rates.From my understanding, the EONIA rate is published at the end of the trading day on the start date of the period while the Fed Fund effective is published at the start of the trading on the end date of the period (i.e. t+1). For example on the Fed page Federal Funds Data, you can see today (Sunday 24 August) the rate for Thursday 21 Aug to Friday 22 Aug, bot not yet the rate from Friday 22 Aug to Monday 25 Aug, even if the relevant rates have been collected 2 days ago. While the EONIA fixing is available for you EOD revaluation curve, the Fed Effective rate is not. I don't see any way to incorporate that information into a live or EOD curve. But maybe someone has a more direct access to the Fed statistic that I do and can see the numbers already. That would be called insider information probably.

Interesting point. One just never stops learning. Obviously if the FF-rate from t to t+1 is not known in t, I cannot incorporate it in the respective curve in t.But what puzzles me is the example I already brought up in the other thread:Suppose you receive 1USD tomorrow from a FF-collateralised deals (e.g. remaining payment from a swap). Then I thought the collateral from today accrued until tomorrow with the appropriate rate is supposed to exactly set-off the flow from the deal. However, if the collateral accrues from t to t+1 with the corresponding FF-rate (fixed in t+1 (in arrears)) this is not possible as one would have to post [$]C_t = PV_t = \frac{1}{1+r^{FF}_t}[$] (in t), which is [$]F_{t+1}[$]-adapted ...

The Fed Fund effective fixing is released the same day at arnd 6PM EST, so there's no real difference in that regard to the EONIA rate.

@Martinghoul: Do you know this for sure?In Marc's (Henrard) document "Interest Rate Instruments and Market Conventions Guide" it says: "The rate is published in the morning (between 7:00 and 8:30) of the period end date" ...

Well, it's what I see in BBG...

So is there a market standard for this issue now? What do we use to bootstrap the USD overnight curve from t to t+1, and from t+1 to t+2?Is it acceptable to use the FF rate from t-1 to t for these two days as that rate is the latest known FF rate as of t?