2026-1039
Tax-loss harvest
Harvesting $2,906.15 in two lots
A smaller loss usually costs more in trading, tracking error and paperwork than it saves in tax. How we chose the threshold, and when we ignore it.
Date
Time
09:34 ET
Account
Brokerage
Amount
−$2,906.15
Status
Signed
Checks passed
Wash-sale clear · Band rule · Liquidity floor · Tax bracket · Position cap · Trading window · Cost vs benefit
Signed by
Priya Raman
EA
The morning pass could harvest every lot that dips below its cost. We do not let it. A loss has to reach $500 on a single lot before the AI drafts a harvest, and this note explains why.
Every harvest has costs that do not appear on a trade confirmation. The replacement fund tracks a slightly different index, so the household’s return drifts from its plan. The lower cost basis means more tax later, when the replacement is sold. And each harvest adds a line to the household’s tax forms that someone has to read. Below a few hundred dollars, those costs tend to eat most of the saving.
We tested thresholds from $100 to $5,000 on three years of our own drafts. Below about $400 the number of harvests roughly doubled while the tax saved grew by less than a tenth. Above $1,000 we started missing losses that were worth taking. We picked $500 and review it every January.
The threshold is a default, not a rule. A household with a large gain to offset this year can ask for a lower one, and a household that expects to give appreciated shares to charity may not want harvests at all. Your advisor sets that with you. Whatever the threshold, every harvest still goes through the same wash-sale check across all linked accounts before it can be signed.