Pretty much all the literature out there states to run a divergent or parallel angle or radius for the squish. Ive ran all of them, and they all do different things. Most interesting was a convergent angle- by several degrees- and what it did to the power curve- in some cases it was desirable.
In the geometry aspect of squish design: One issue that has to be accounted for is how is the squish cut. A piston crown doesnt have an angle as its actually a radius. Most people calculate the angle of a coincident tangent or secant line to the radius of the piston crown. This is because most cut the squish band on a compound set-up in the lathe. As such there will be a tiny little "pocket" on the outside edge of the squish band (in addition to the ring periphery volume). In highly developed engines this can present enough additional end gasses to promote detonation. Many of the software programs out there that people use to calculate squish velocity calculate a radius type parallel squish band, which will have higher values than a more typical "straight" angle squish band to a crown radius.
In the tense of this discussion, I think its going to wind up a discussion of squish velocity, as that is the factor influenced by the band set-up. There's considerable argument regarding MSV for given fuels. I have my opinion and it differs from some and others hold the same. My always standard disclaimer would have to apply that if we all did it the same you'd just go to Wal-Mart to get your stuff. Speaking of Banshee motors- typically the pipes are not highly developed enough and the stock cylinder porting layout is crappy enough that many have not ran into issues with too high of an MSV creating detonation. Its getting closer for certain and with the CP Ind. cylinder line it moved much closer as evidenced by how many people had detonation issues early on. In addition I hold the opinion that too many people view combustion as this homogeneous event that is the same cycle to cycle and can be calculated. The modeling involved would require astounding computing power to calculate whats going on not only in a mere snap-shot of several cycles, but at something with enough RPM resolution to show whats up from 6000 RPM to 11000 RPM. AND- In an effective enough manner to make well enough sense to try and effect a worthy change. Most of my head designs come from empirical data. I have found the software handy to do the geometry for me, but the old adage- 1 test is worth a thousand expert opinions works for me. I typically always strove for high to very high MSV numbers in my head designs, but as of late, Ive been trying some different set-ups that have a lower MSV. Mainly because what I think was happening in a convergent set-up. In a divergent angle squish band set-up, the squish velocities as the bowl edge will be lower than at the squish OD, but it does expose a larger curtain area of moving gasses to the contents of the bowl. In a convergent set-up, I cant help but think that the band acts narrower than it actually is in that only the inner "half" of the squish band is creating the turbulence. Volume and velocity are inverse of each other, so thats something to think about in how a squish band may be working. Especially when you get into very wide bowls either by design(narrow band) or by configuration(big bore).