It's another one of those things for which there is no perfect size.
First plan A, external diffusion.
Assume 2700 RPM, 540 cubic inches, a VE of 0.95, and zero forward airspeed. Velocity through the intake ring would be approximately...
3" = 81 knots
4" = 45 knots
We really don't care about zero airspeed values, but it does illustrate the effect of intake area given constant flow. Less area means higher velocity, and that detail comes into play given forward airspeed, because for external diffusion, manifold pressure rise due to dynamic pressure is based on TAS in excess of that static value.
So, assume a 540 at 2600 RPM, same 0.95 VE, 230 TAS, 8500 ft, standard day. Further assuming no loss due to separation or other errors, available pressure before the filter or throttle body would be...
3" = 22.66" Hg
4" = 23.09" Hg
5" = 23.32" Hg
Recall I said "no perfect size"? Larger and larger simply offers diminishing returns. The difference between 3" and 4" rings is 0.43" Hg, while the difference between 4" and 5" is 0.23", and so on. You can push intake area up to get the last little bit of MP, but you have to fair the larger ring with an even larger external shape. You'll likely wind up balancing some additional power against some additional drag.
The engine is an air pump, so assuming same compression, timing, etc, the difference between 260 and 300 HP would be VE, volumetric efficiency. I used 0.95 above. Let's assume a hotrod motor does 100%. The result is a tiny loss of calculated manifold pressure (0.03"), as the static intake velocity is about 2 knots higher.
Plan B is internal diffusion, which requires a well shaped diverging duct inside the cowl. Velocity through the intake ring can be as high as freestream velocity, i.e. TAS. The slowing and expansion resulting in conversion of available dynamic pressure to increased static pressure takes place inside the duct, not out in front of it. Again, no perfect size. At 230 knots, theory says the intake area can be down around 2.54 sq inches, a ring diameter of 1.8". However, given the same RPM it will cost HP at a lower airspeed.
You can see those tiny little pitot inlets on some F1 racers at Reno. Plenty of internal cowl length when you have an ultra long propshaft extension. I used external diffusion on my -8 because it's a 4-cyl with no spool...barely enough length for a decent filtered airbox.