Trader Portfolio

This wraps neural-trader's mean-variance optimization engine into a Claude workflow that actually remembers your holdings and past allocations. It pulls current portfolio from memory, runs optimization with optional risk targets, spits out rebalancing trades with cost estimates, and stores results for later comparison. The interesting bit is the SONA neural prediction for expected returns and the pattern search through historical allocations by Sharpe ratio. It's basically portfolio theory automation with memory, so you're not starting from scratch every time you want to rebalance. Useful if you're managing multiple portfolios or revisiting allocations frequently enough that context matters.