"Sell the loser, book the loss, save on taxes" is one sentence and three constraints. Do it well, across a whole book, without tripping a wash-sale window or drifting from the index, and harvesting stops being a screen and becomes the constrained optimization it actually is. That is the problem PRISM was built to solve.
In plain English. If an investment has lost value, selling it lets you use that loss to reduce the tax you owe on gains elsewhere — real money back. But three rules fight you: you have to buy something similar so the client stays invested, tax law won't let you buy back the same thing within 30 days, and the portfolio still has to track its index. Doing that well for one account is fiddly. Doing it for thousands, every night, is the problem this page is about. Unfamiliar words? See the glossary.
In a taxable account, a harvested loss is a genuine, spendable benefit: it offsets gains and, within limits, ordinary income. A fund cannot pass an individual security's loss to you; an account holding the securities directly can. That is the entire economic case for direct indexing. But the naive approach — periodically sell whatever is down and buy a placeholder — leaks the benefit in a dozen quiet ways: it books losses the wash-sale rule will later disallow, ignores which lots to sell, and drifts the portfolio off its benchmark. Each leak is small; across a book and a year, they are the difference between a program that markets well and one that delivers the after-tax dollars it promised.
Doing it properly means treating harvesting as an optimization that maximizes after-tax value subject to tracking error, the wash-sale envelope across the household and across time, lot-level holding-period rules, and turnover and cost limits — all at once, because they are coupled. We wrote about exactly why this is hard →
PRISM is a black box with a clean contract: your positions, lots, constraints, and benchmark in; optimized, wash-sale-aware trades and a reproducible audit log out.
Maximizes after-tax value subject to all the coupled constraints — for every account.
The 61-day envelope is respected across the household and across time — no quietly-disallowed losses.
Which specific shares to sell is treated as the decision it is, with holding-period awareness built in.
Harvesting stays inside your tracking-error budget — you capture the loss without abandoning the index.
PRISM does not claim to beat the exact single-account optimum every time. Its edge is capturing tax alpha that simpler approaches structurally miss, and doing it at a scale exact methods cannot reach. Comparators are labeled generically.
Share of tested regimes where PRISM delivers more after-tax wealth. An open-source baseline captures only ~$3k at scale where PRISM captures six figures.
Share of tested regimes where PRISM beats rules-based harvesting on after-tax wealth.
Names at which PRISM is the only engine measured still producing six-figure tax alpha.
Move the sliders to your book, return, volatility, and tax rate to see the after-tax wealth a tax-loss-harvesting overlay can recover over time.
A matched-workload pilot runs harvesting on your accounts and constraints and shows the after-tax result against your current approach — every losing case included, ROI computed with your numbers.
Request a pilot →