Macro “pattern matching” + online portfolios = Sharpe ~1.3 across major equity indices.
Most pattern-matching strategies only look at prices. This preprint flips it:
Step 1: pick stocks that actually “move with macro” using Δβ (asymmetric macro sensitivity)
Step 2:cluster them (hierarchical + symmetric uncertainty) to avoid redundant bets
Step 3:two-stage analogue search: find similar macro regimes in history, then find similar stock patterns within those regimes
Step 4: weight macro signals by spillover importance, pick the strongest name per cluster, and allocate
Results (1995–2023, daily): the macro-aware clustered model is the top performer, reporting ~110% cumulative (EW) on S&P 500 and strong outperformance on Dow + Nikkei too.