Larry sent me a message indicating that my troubles with inverted flight may be do to a forward CG. I have now moved the CG back 1/2 inch from the location specified on the plans with little degradation in upright stability and an ability to fly and turn inverted. Will continue to experiment.
Finished CG experiments and have concluded that 3/8" behind stock CG gives a good blend of stability and aerobatic performance. This assumes that you have a Polaris with a 13" wingspan. I also moved the step back with a Depron shim. If you have no interest in inverted flight, CG shown on plans is great. Next, water testing.
Can you provide some close up shots of the bottom of the fuse at and just ahead of the step and of the bottom, rear and sides of the tip floats?
I suspect the root cause of the problem is due to prop torque forcing one side down into the water as you accelerate. Then, if you over correct, you get the reverse turn.
However, the way you do the step and the lower side edges on the fuse and the lower sides and rear edges of the tip floats can contribute a lot to making the problem better or worse.
The step itself as well as the bottom side corners of the fuse need to be almost razor sharp corners. The same applies to the tip float lower side corners and rear step.
In the air, you want nice rounded corners to reduce drag. In the water, its just the opposite. Those nice rounded corners will screw up the model badly.
If those corners are not really sharp 90 degree corners, then you get a huge increase in water drag. Any rounding of the corners can actually suck the model down into the water as you are trying to get up on step - just exactly the opposite of what you want to happen. In addition, it can make for directional instability at speed.
You might also try making the tip floats a tad shorter. You want them well out of the water when your on step.
By the way, this kind of thing - sharp corners on the step and sides of the fuse - is important on the larger models all the way up to full scale.
However, its even more critical at very small scales because of the meniscus effect. At small scales water tends to climb up the sides of things and 'clings' more than at large scales. So those sharp corners need to get sharper the smaller you get.