Its a bit hard without drawing a diagram -
If you draw a line from the spindle to the edge of the record.( call it line A )
Then choose a null point on that line
Now draw a line ( call it line B ) at 90 degrees to line A intersecting the null to the armboard or mounting point - so you now have a T.
Normally I assume you would position the arm mounting along that 90 degree line such that the stylus swings an arc where it touches the null, but both sides of the null are underhung.
What I am suggesting is now move the mount forward so that when the stylus is forward of the null point.
Now when the stylus swings through the arc it should intersect line A twice.
So now the arm starts off underhung, then crosses line A and goes to overhung, then crosses line A again and goes to underhung.
In this set up the maximum tracking error is reduced, but you would still have the advantage of no skating forces due to offset.
Alternative Explanation 2
Alternately if you take your existing Rigid Float move the arm across the record.
The stylus draws an arc.
Draw a line through the spindle that crosses that arc at one point.
Now with the stylus positioned at that point, move the arm mount forward along a line 90 degrees to that point..
Instead of 100% underhung either side of the null you will now have both under and overhung. You should if I have my head right have reduced the maximum tracking error - potentially you could halve it - by having a mix of underhand overhung instead of 100% underhung.
Hopefully you can follow this.
By the way here’s another example of a 0 SideForce arm. Fidelix have been around forever.
https://exclusive-audio.jp/en/products/0-sideforce