Regime upgraded to risk_on. Tech leadership vs SPY intact across 5/10/20d windows (QQQ +3.3% vs SPY +0.05% 20d). US-China dinner optics and AI capex momentum supportive, though rising yields and AI-debt warnings (Burry, Goldman, Oracle force majeure headlines) argue against chasing the parabolic names. Portfolio at 61% invested vs 74% target — close the gap via non-extended tech.
To beat SPY by 5pp I need overweight tech beta, which has been the working thesis (all positions green). But the drawdown clause forbids piling into AMD/META at RSI 76-80 after +31-32% 20d runs — a single air-pocket there would breach the protect side of the mandate. Instead, add to NVDA (RSI 44.5, dipped 2%, laggard with catch-up potential in a risk_on tape), MSFT (strongest 5d among holdings, RSI only 59.6, unified Copilot enterprise narrative), and QQQ (diversified tech beta). This lifts invested to ~74% with ~26% cash cushion protecting the rolling drawdown vs SPY. Skip AMZN despite RSI 41 — negative rel strength and AI-capex payback questions mean it drags relative return.