Weighted Benchmark?

[chriscross]

Dear all,

I am currently writing a research paper and have the following data available:

Percentages of a certain asset class within one portfolio.
EXAMPLE: 35% in equity, 65% in bonds
Further more I have the total return of that portfolio.

Now I want to create a synthetic benchmark.
EXAMPLE: 35% x S&P500 ++ 65% x T-Bill90

My question: Will that yield a good benchmark? Since it is artifical.
If I run a regression, what do you I need to consider?

I want to test the abnormal returns (dependent variable) against some independent variables (% invested in equity/bonds)

I hope I expressed myself in a clear manner, if you have any questions please tell me!

Thank you in advance,

Chris