The Unrealized Losses Hiding in Your Winners: How Lot-Level Scanning Finds Money Brokerage Apps Hide

A position can be up 40% overall and still contain harvestable losses worth thousands of dollars in tax savings. This sounds contradictory until you understand how individual tax lots work — and why the position-level profit and loss numbers shown by every brokerage app systematically hide losses that exist at the lot level. The unrealized losses hiding inside winning positions are not unusual or rare. They exist in nearly every multi-lot portfolio that has been built up over time through periodic purchases, dividend reinvestments, dollar-cost averaging plans, or employer stock purchase programs. They are invisible at the position level by construction, because the brokerage's display logic blends every lot's gain or loss into a single number. They are fully visible — and fully harvestable — at the lot level. The gap between those two views is where most investors leave money on the table.

The logic is straightforward. A "position" is just a label for the total holding in a particular security. The actual tax-relevant unit is the lot — a specific batch of shares purchased on a specific date at a specific price. If you bought a stock four times over three years, you don't own one position. You own four lots that the brokerage chooses to display as one position for cosmetic reasons. The IRS doesn't care about positions. It cares about lots. And inside any position purchased over multiple dates, the prices paid will vary, which means some lots will have higher gains and some will have lower gains — and some, frequently, will have outright losses, even when the position as a whole is in the green.

Why Brokerage Apps Show You the Wrong Number

Open any brokerage account on any platform — Schwab, Fidelity, Vanguard, Robinhood, E*Trade, Interactive Brokers — and the default position view shows a single P&L number per holding. That number is computed by taking the current market price, multiplying by total shares, and subtracting the average cost basis. The result tells you how much money you've made or lost on the holding overall. It tells you essentially nothing useful about which lots are harvestable.

This default exists because most retail investors don't think in lots. They think in positions. "I own Apple. It's up." That mental model is fine for understanding portfolio performance. It is actively destructive for tax planning, because it conceals the granular information needed to make good harvesting decisions. The position view is to tax loss harvesting what a daily weather summary is to deciding whether to bring an umbrella to a specific meeting at 3pm — technically related, practically useless.

Almost every brokerage will show you the lot-level view if you click through two or three menus. The information is there. It just isn't the default. The result is that investors who casually check their accounts and see "I'm up overall" never investigate further, and the harvestable losses inside their winners stay buried for years until the position eventually gets sold and the embedded gains all come due at once.

Three Patterns That Produce Hidden Losses

In practice, hidden lot-level losses come from three recurring situations. Recognizing the pattern helps explain why the phenomenon is so common — and why a continuous lot-level scan finds opportunities that a once-a-year December review usually misses.

Pattern 1: The peak buyer. An investor builds a position over years, then makes one large purchase near a local top. The position rallies again, the average gets blended, and the peak-buy lot disappears into the average. But that lot — purchased at the peak — is often still underwater even when everything else is up. A common example: an employee who exercises stock options at a stock's all-time high in 2021, then watches the company recover from a 35% decline in 2022 to new highs in 2024. The 2021 lot may still be sitting at a $4,000 to $8,000 unrealized loss even as the position overall is up 60%. The position view shows green. The lot view shows red on that specific tranche.

Pattern 2: The dividend reinvestment trap. Many investors automatically reinvest dividends, which creates a new tax lot every quarter — sometimes every month — at whatever price the stock was on the reinvestment date. Over five or ten years, a single position can accumulate 40 or more lots, each with a different basis. Some of those lots will inevitably have been purchased near short-term peaks. They sit at small unrealized losses — $50, $100, $300 each — that individually look trivial. Aggregated across a portfolio of 15-20 dividend-paying stocks, they easily total $5,000 to $15,000 in harvestable losses on a $500,000 portfolio. The position view shows zero of this.

Pattern 3: The dollar-cost averaging cycle. An investor with a $2,000 monthly contribution into a brokerage account ends up with 12 lots per year per fund — 60 lots over five years on a single fund. During any prolonged sideways market or modest pullback, several of those lots will be underwater. The fund as a whole may be up 8% over five years. Three or four individual monthly purchase lots from peak periods may be down 5% to 12%. The harvestable amount is small per lot but real, and the lots replenish themselves continuously as new contributions go in and markets fluctuate. This is the most common pattern in retirement-account-adjacent taxable brokerage investing, and it's the one that benefits most from continuous (rather than annual) scanning.

A Worked Example: $1,142 From a Position That Looked Like a Winner

Consider an investor with a $48,000 position in a single ETF, held in a taxable brokerage account, accumulated over four years through monthly purchases. The position is up 22% overall — a $48,000 market value against a $39,300 blended cost basis. The position-level P&L: +$8,700.

Drill into the lot view, and the picture is more complicated. Of 48 monthly purchase lots over four years, 38 are up and 10 are down. Six of the down lots are individually small — losses of $80 to $250 each. Four are larger: lots purchased during three brief market peaks now sitting at unrealized losses of $640, $920, $1,400, and $1,840. The total of all 10 down lots: $4,800 in harvestable losses, fully invisible at the position level because the 38 winning lots more than offset them in the blended display.

The investor sells the 10 losing lots, books a $4,800 short-term capital loss, and immediately buys a different but correlated ETF — say, replacing the S&P 500 fund with a Russell 1000 fund — to maintain market exposure without triggering the wash sale rule. The portfolio's market value is unchanged. The portfolio's market exposure is essentially unchanged. But the investor now has $4,800 in realized losses available to offset $4,800 of realized gains elsewhere this year.

For an investor in the 24% federal bracket realizing $4,800 of short-term gains elsewhere, the tax savings is $4,800 × 24% = $1,152. For an investor harvesting against long-term gains at the 23.8% combined rate (top bracket plus NIIT), the savings is $4,800 × 23.8% = $1,142. From a position that looked, on the brokerage's main screen, like there was nothing to do.

This calculation is what gets lost in conventional tax loss harvesting advice. The standard framing is "harvest losses when the market drops." The lot-level framing is "harvest losses whenever they exist, regardless of what the position-level number says." Markets don't have to drop for the lot-level view to surface harvestable opportunities. Normal market churn — the everyday up-and-down of asset prices — produces enough lot-level losses inside winning positions to keep a continuous harvesting program running year-round.

Why Once-a-Year Scanning Misses Most of It

The traditional approach to TLH is to review the portfolio once, in December, and harvest any positions sitting at a loss. This catches the obvious losses — full positions that are red. It misses almost everything described above.

Three reasons. First, lot-level losses inside winners are invisible to the position-level scan that most investors and advisors run. Even with the right intent, an annual review using the brokerage's default view will not surface them. Second, lot-level losses are transient. A lot purchased near a peak may be down 8% in February, recover by April, drop to -3% in August, and be back to flat by November. A scan run only in December will catch only whatever happens to be down at that specific moment. The losses that briefly existed earlier in the year — and could have been harvested — are gone. Third, the size of any single lot's loss may be too small to feel worth a transaction. A $200 loss on one lot doesn't seem like much. But $200 here, $400 there, $850 over there — twenty of those across a portfolio is $10,000 to $15,000 in harvestable losses per year, captured only by software that scans continuously and acts on small opportunities as they appear.

A continuous lot-level scan is the only way to capture this systematically. The mechanics are not complicated: every lot in the portfolio is monitored against current market prices in real time; when a lot crosses below its purchase price, it becomes eligible to harvest; the system evaluates wash sale risk, replacement security correlation, and the investor's current-year tax position; if all three pass, the loss is realized and a substitute security is purchased the same day. This happens automatically, multiple times per month for most portfolios, and the captured losses accumulate as a running balance available to offset future realized gains.

What This Looks Like in a Real Portfolio

Consider a $750,000 taxable brokerage portfolio with 18 holdings across stocks, ETFs, and a couple of mutual funds. At the position level, on any given day, the portfolio might show 14 winners and 4 losers. The four losers represent obvious harvesting candidates — say, $22,000 in unrealized losses across them, fully visible.

At the lot level, the same portfolio might contain 312 individual lots — most accumulated through years of contributions, dividend reinvestment, and rebalancing. Of those 312 lots, perhaps 220 are up and 92 are down. The 92 down lots span all 18 holdings, including several positions where the holding overall is up 15-30%. The total of all 92 down lots: roughly $48,000 in harvestable losses — more than double what the position-level scan would have surfaced.

That extra $26,000 in harvested losses is worth $3,900 to $6,200 in federal tax savings depending on the investor's bracket. For a high-income investor at 23.8% combined rate, it's $6,188. For a year. From the same portfolio. From the same market conditions. From the same set of investment decisions. The only difference is the depth of the scan.

This is what Capability #3 of TaxHarvest does continuously: scan every lot in the portfolio, every market day, against current prices, identifying every harvestable loss the moment it appears, before it disappears again. Most of those individual opportunities are too small to be worth a human's time to act on. None of them are too small for software to act on. Aggregated, they produce dramatically more captured loss per year than even the most diligent annual review.

What Investors Should Actually Do

Three things, in order of how easy they are.

First: turn on the lot-level view in your brokerage account. Every major platform offers it. The path varies — Schwab calls it "Cost Basis Lots," Fidelity calls it "Tax Lots," Vanguard calls it "Specific Lots." Once enabled, every position you own should show you the underlying lots with individual gain/loss numbers. Look at the holdings you assume are winners. Inside several of them, you'll find lots with individual losses you didn't know existed.

Second: when you do harvest, harvest at the lot level rather than the position level. The brokerage default is to sell oldest-first (FIFO), which usually sells lots with gains rather than lots with losses. Override that to specifically identify the losing lots for sale. The brokerage will let you do this, but it requires actively choosing — often through a multi-step process at the time of trade — rather than accepting the default.

Third: consider whether a continuous scan is worth automating. A motivated investor with a portfolio of 8-12 holdings can probably manage lot-level scanning manually if they're willing to log in monthly and click through to the lot view of every position. Past that scale — a portfolio of 20+ holdings, hundreds of lots — the manual approach becomes implausible, and continuous software-based scanning is the only practical way to capture the available losses without missing most of them.

For the deeper mechanics of why optimal lot selection produces dramatically different outcomes than default brokerage methods, see our optimal tax lot selection deep dive. For a worked-through case study showing how lot-level harvesting compounds into five-figure savings on a single position, see Sarah's NVDA case study. For investors above the NIIT threshold, every dollar of harvested loss is worth 19% more than a flat capital-gains-rate calculation suggests — see our net investment income tax NIIT 2026 explainer. And for the broader strategic case, tax loss harvesting for retirees walks through how the same lot-level technique helps retirees use the 0% bracket to permanently raise basis.

The unrealized losses hiding in your winners are not exotic. They are not the result of clever timing or sophisticated strategy. They are the natural and inevitable consequence of buying anything more than once. The only question is whether you're set up to find them.