The hang-test procedure serves two purposes. The first, as Jake said, is to center the controls at cruise. The second is to allow the fuselage to ride at its designed stance (neither nose-up nor nose-down) at cruise.
The second goal is deeply affected by prop thrustline position relative to the aircraft's CG. The higher the thrustline is above the CG, the LESS nose-down the ideal hang angle will be. So...
The good old "2.5 deg. nose down on the mast " is a Bensen spec. It assumes (1) that the mast is raked 9 degrees and (2) that the gyro has centerline thrust (CLT - - prop thrustline passes through CG).
If you build a gyro with, e.g., a prop thrustline six inches above CG, but use the Bensen spec, you"ll have a gyro that rides badly nose-down at high airspeeds and throttle settings. I know this because I flew one of these -- a no-stab Air Command lowrider -- for years. At 80, throttle hammered, it was like riding in a wheelbarrow waiting to be dumped forward.
The current CLT Air Command can use the Bensen spec.
The exact spec number depends, of course, on which part of the gyro you're measuring. If your gyro has a 9 degree aft mast rake, but you measure the hang angle on the keel, you'll need to add the mast rake angle to the spec to arrive at a keel-based hang spec. In the case of a Bensen, this brings you to 11.5 deg. nose-down on the keel. Example: the Gyrobee has zero mast rake. Its hang spec, translated from Bensen, would be 11.5 deg., measured either on the mast or the keel.
The stock RAF has a very high thrustline. IIR, its 5 degree spec is to be measured on the keel. This would be the equivalent of backing the Bensen spec from 11.5 deg. on the keel down to only 5 deg. This allows the gyro to ride more level despite the HTL.
If a gyro is set up using Bensen specs, but then flies unexpectedly nose-down at fast cruise, it likely has undiagnosed HTL. That situation should be corrected for safety's sake. Rig an adequate H-stab or move weights up relative to the prop's center.