The Legendary Dragon Alephwyr, named Alephwyr, BA

I use probability heuristics to put my opponent on a range of arguments and then game theory to randomly determine how frequently to attribute each argument to them. My own arguments are a binary range of extremely strong and extremely weak arguments, to keep them guessing.

You're forgetting folding equity: a bad argument pressed aggressively enough just becomes an incomprehensible or unfalsifiable argument, therefore obfuscating my actual range.

Not if my argument strategy is actually GTO, no. Optimally, you should be indifferent to engaging or not engaging my arguments if my ranges are properly selected.