Post by Asymmetry Computing

"$238,000 saved in tax" means nothing without the denominator. So here it is, in full. The setup: a $5,000,000 direct-indexing account. The client also realized $1,000,000 of capital gains elsewhere — selling a position. The job: harvest enough losses inside the sleeve to offset those gains at a 23.8% rate. So $238,000 saved = a full $1M of gains erased = 4.76% of the account. We ran six strategies on the SAME account — same data, constraints, costs, and tax: an index ETF, a legacy hold, rules-based harvesting, an open convex solver, an exact commercial MILP, and our engine — across four real regimes (2018 drawdown, 2022 bear, 2020 COVID, 2019 bull). The receipts: • PRISM captured the full $238k — matching the exact optimizer's tax budget, in a fraction of the runtime. • In the 2020 crash it beat that exact optimizer by ~$381k after tax (+7.6% of the account). • Lowest drawdown and best Sharpe in 3 of 4 regimes. • Still harvesting six-figure tax alpha at 5,000+ names, where exact solvers time out. • The tax ladder tells the story: PRISM ≈ exact MILP at $238k, rules-based harvesting ~$118k, an open convex solver only ~$37k. Scale it to your book: in a harvest-rich year that's ~4.8% of the account in tax alone. Swipe for the receipts. 👇 #DirectIndexing #TaxAlpha #PortfolioManagement #WealthTech #QuantFinance #AssetManagement #TaxLossHarvesting #RIA