How to calculate the MTM value today of a 1-w forward contract sold 2 days ago?

[mizhael]

For example, for a FX forward 1-week for EURUSD,

you entered that position on May 10, 2010,

what's the value of that position on May 11, 2010,
and ...

and then today?

I am trying to use Bloomberg Spot, Spot Next and 1-Week forward data points
to interpolate and get the value of 6-day, 5-day, 4-day, etc. contracts...

Any thoughts?

Thank you!

 

[Martinghoul]

So what's your question? You either do it by simple interpolation or by looking at the two depo rates.

 

[mizhael]

So what's your question? You either do it by simple interpolation or by looking at the two depo rates.

I just don't know how to do it correctly...

 

[Martinghoul]

Well, you're gonna have to be more specific than that. Are you having an issue calculating where the fwd to your date trades in the mkt, based on the mkt data available? Or is your issue more with the calculation of the MTM of the fwd, given you know the mkt level?

 

[mizhael]

Well, you're gonna have to be more specific than that. Are you having an issue calculating where the fwd to your date trades in the mkt, based on the mkt data available? Or is your issue more with the calculation of the MTM of the fwd, given you know the mkt level?

Okay, it was originally a 7-day forward, traded/incepted on 5/10/2010.

What's the value of this contract end-of-day yesterday (5/13/2010)?

What's the value of this contract right now today (5/14/2010)?

What we do have are the data as follows:

1. Spot, Spot-Next, 1-w Forward on 5/10/2010.

To get the EOD value of that 4-day forward EOD yesterday,

we would need

2. Spot, Spot-Next, 1-w Forward on 5/13/2010.

To get intraday realtime value of that 3-day forward right now today,

we would need

3. Spot, Spot-Next, 1-w Forward on 5/14/2010 and they are realtime ticks.

------------------

How do you do the calculation based on these market data?

Thank you!

 

[Martinghoul]

Why not just do it using simple linear interpolation?

 

[mizhael]

How, exactly, in detailed steps?

 

[Martinghoul]

y - unknown fwd; y0 - s/n fwd; x - days to unknown fwd; x0 - days to s/n; y1 - 1w fwd; x1 - days to 1w

 

[mizhael]

y - unknown fwd; y0 - s/n fwd; x - days to unknown fwd; x0 - days to s/n; y1 - 1w fwd; x1 - days to 1w

Thanks yet a few questions:

1. When you do day-counts, do you take holidays and/or weekends into account?

2. You have the following 3 data points:

the settle date for spot,
the settle date for SN,
the settle date for 1W forward,

you can use any two data points to do the linear interpolation,

why do you use SN and 1W?

3. Say now we have more data points:

Spot,
SN,
1W,
2W,
3W,
1M,
2M,
3M,

we would like to interpolate any dates in between,

do we always choose the two neighboring points to do linear interpolation?

For example, if we want to get a 17-day forward, do we use the settle dates of 2W and 3W contracts to do linear interpolation?

4. Is Linear Interpolation arbitrage-free? Is it the best method here?

Thanks again for your help!

 

[Martinghoul]

Thanks yet a few questions:
1. When you do day-counts, do you take holidays and/or weekends into account?

Shouldn't matter a lot for interpolation, as long as you're consistent. Ideally, you should use whatever convention is used for the mkt in question, but I'd suggest you just do calendar days. It's easier that way.

2. You have the following 3 data points:

the settle date for spot,
the settle date for SN,
the settle date for 1W forward,

you can use any two data points to do the linear interpolation,

why do you use SN and 1W?

Because spot isn't, strictly speaking a fwd. You need to interpolate between two fwds.

3. Say now we have more data points:

Spot,
SN,
1W,
2W,
3W,
1M,
2M,
3M,

we would like to interpolate any dates in between,

do we always choose the two neighboring points to do linear interpolation?

For example, if we want to get a 17-day forward, do we use the settle dates of 2W and 3W contracts to do linear interpolation?

You can do whatever you want, but it would make a lot more sense to me to use two neighboring points.

4. Is Linear Interpolation arbitrage-free? Is it the best method here?

Thanks again for your help!

Answer to both questions is "probably isn't". However, it offers the best blend of the practical and the accurate, IMHO.