Tax-loss harvesting, plainly explained

Tax-loss harvesting is real. It's also one of the most over-promoted concepts in retail wealth. The math is the same whether you read about it in a Bloomberg article or a robo-advisor brochure. The brochures tend to leave out the second half.

The mechanics in 1 paragraph

Tax-loss harvesting is real. It's also one of the most over-promoted concepts in retail wealth.

You sell a position at a loss. You harvest the realized loss to offset realized gains elsewhere in the same year, or up to $3,000 of ordinary income if you don't have offsetting gains. Anything left over carries forward indefinitely. To stay invested while you wait out the wash-sale rule (more on that below), you replace the position with a different but similar security. The benefit is a lower tax bill this year, with the trade-off that your cost basis in the new position is lower, so a larger gain later.

That's the entire mechanic. Everything else is implementation detail.

A worked example with real numbers

Take a $1,000,000 taxable account. In a typical year a diversified equity book will surface roughly 2% of unrealized losses available to harvest, even in up years. The math:

• $1,000,000 portfolio. 2% of losses harvested ($20,000 realized loss), 37% marginal federal rate plus 5% state, applied against $20,000 of realized gains elsewhere. Savings: $8,400 this year.
• $5,000,000 portfolio. Same 2% harvest rate, same 42% combined rate. Savings: $42,000 this year.
• $250,000 portfolio. Same 2%, 24% federal rate, no state tax. Savings: $1,200 this year.

The benefit scales with 3 things: how much you have, how much you can harvest in a given year, and how high your marginal rate is. Two of the 3 are out of your control. The third (how much you have) is the lever that determines whether the strategy is worth the friction.

The wash-sale rule, plainly

The IRS won't let you sell a position at a loss and immediately rebuy the same security to keep the tax benefit. Specifically: if you buy a "substantially identical" security within 30 days before or after the sale, the loss is disallowed and added to the basis of the new purchase.

"Substantially identical" is interpreted strictly. The same stock, the same ETF, and most call options on the same name will all trigger the rule. A different fund tracking the same index is the gray zone the industry has settled by convention (the IRS hasn't formally ruled), and the practical workaround most advisors use.

A clean implementation:

• Sell the original position at a loss.
• Within the same day, buy a similar-but-not-identical replacement (different index, different fund family, different vintage).
• Hold the replacement at least 31 days.
• After day 31, you can switch back to the original if you prefer, with the new basis intact.

The rule also applies across accounts. A loss in your taxable account can't be saved by rebuying in an IRA. The IRS sees both.

When tax-loss harvesting actually saves you money

This is the part the brochures skip. The benefit is real, but it isn't free, and there are 3 situations where it's overrated for retail investors.

1. You're in a low marginal bracket

The benefit is the realized loss times your marginal rate. At a 12% federal rate the savings on a $5,000 harvested loss is $600, not $1,850. Add the friction of execution (being out of the original position for 31 days, the spread, the bookkeeping) and the net benefit gets small fast.

If your combined marginal rate is below 25%, the strategy is probably worth doing only at the lazy "harvest what's sitting there" level, not as an active program.

2. You don't have offsetting gains, and you won't have them later

The first $3,000 of net losses offsets ordinary income each year. Anything beyond that carries forward, and the carry-forward only pays off if you have gains in a future year. For an investor in pure index funds with a buy-and-hold posture, the carry-forward may never be used. The loss is a tax asset, but only if you eventually use it.

3. You're trading small positions

Tax-loss harvesting has fixed friction. Identifying the trade. Executing it. The spread on getting in and out. Being out of the original for 31 days. On a $5,000 position the realized loss might be $500, the tax saving $150, and the friction half of that. The math gets worse the smaller the position.

When it's genuinely worth doing

The clear cases:

• You hold individual names or sector ETFs at meaningful position sizes (over $25,000 each).
• Your combined marginal rate is at or above 30%.
• You have realized gains to absorb the loss in the same year, or a credible expectation of realizing gains in the next 1 to 3 years.
• You can stay disciplined about the 30-day window and the substantially-identical rule.

This is also where the savings compound. A $5M book at a 2% harvest rate and a 42% combined rate saves $42,000 a year. Reinvested, that becomes a 6-figure tax asset over a decade. The Wealth tier in Finapolis is built around this case for exactly that reason.

Does tax-loss harvesting change your eventual tax bill?

Strictly, no. The lower basis in the replacement security means a larger taxable gain whenever you eventually sell. The real benefit is in 2 places. First, the time value of the deferral: paying $1 in tax 10 years from now is worth less than paying it today. Second, rate arbitrage, if your future tax rate is genuinely lower than today's. Retirement, lower-income years, the step-up at death for assets left in your estate.

Of those, the step-up at death is the largest and least talked-about benefit for long-term taxable holdings. The lower basis from years of harvesting essentially disappears for your heirs.

Where the workbench fits

The Portfolio module in Finapolis scans for unrealized losses across your taxable holdings, flags wash-sale conflicts before you place the trade, and shows the tax-equivalent benefit of each harvest at your stated marginal rate. Every figure traces back to the underlying lot data, so the harvest is auditable line by line.

Same point as everywhere else in the platform. The math runs faster, the sources are auditable, the decision is yours.

FAQ

Is tax-loss harvesting worth doing in a 401(k) or IRA?

No. Tax-deferred accounts don't realize gains or losses for tax purposes when you sell, so there's nothing to harvest.

Can I rebuy the same ETF the next day?

If it's the same fund or "substantially identical," no. You'll trigger the wash-sale rule and the loss is disallowed. A different fund tracking a different index is the standard workaround.

Can my spouse buy what I just sold?

If you file jointly, no. The wash-sale rule applies across spouses on joint filings.

What if I have no realized gains this year?

You can offset up to $3,000 of ordinary income with net losses, and the rest carries forward indefinitely. The strategy is most valuable when you actually use the carry-forward later.

Should I harvest at year end or throughout the year?

Both. The bigger losses tend to show up during the year, not at year end. A continuous program captures them when they exist; a year-end-only program misses the ones that have recovered by December.