wilmott.com

Currently, I am analyzing a FI ReVal trade. The idea is that the spread between two bonds has widened to a record high. I am trying to implement in DV01-neutral basis with a ratio of 10x5 (notional). However, when I regressed the spread over libor/swap against the spread differential between these bonds I found some directionality in this trade. This is, when spread widens the spread differential between these bonds widens too. Since I would like to remove the directionality, I was thinking in a beta adjusted basis. However, I have some doubt about it. If the beta coefficient from the regression is 0.6647 the ratio would be 10x6.4 per notional????I appreciate any comment! Enjoy!

I´ll really appreciate any help!

Sounds correct, but impossible to say for sure without looking at the trade in more detail...

Thanks for your answer. Regression = 0.66476x – 8.2174. Asset Price Yield Notional 6m Carry DV01DV01 L Bond A 74,50 14,15% 10.000.000 518,0 4.21 S Bond B 82,25 10,13% 5.189.209 146,1 8.12 402 bps 4.810.791 372K 0.51 Beta S Bond A 82,25 10,13% 6.648.000 187K0.66476 402 bps 3.352.000 331K

Sorry I can't read your output above.To step back a bit, when I look at this stuff, I mostly look at yield differentials so suppose a basic 10yr bond yield % over 7yr bond yield%.So, let's say you have a simple regression says that yield spread between the two is too high and the beta is 0.6647. Then, on a DV01 basis, you would put 10,000 (arbitrary, no resemblance to your 10x5 in your post) DV01 in the Y of your regression and 6,647 DV01 in the X of your regression. This regression weighting will eliminate that directionality.This doesn't get you to notional terms (10m vs 5m or 6.4) but you can back out a notional amount from the DV01 and the specific duration of your bonds.