Harvesting $5,800 of Crypto Losses in December 2022 Knocked My Tax Bill Down 41% — Until April Got Weird

4:50 PM on December 28, 2022, and I am at the kitchen table with a laptop, a legal pad, and a pot of chili I will forget about until it burns onto the bottom. On screen: four red positions totaling negative $5,832 against cost basis. ALGO down 71%. MATIC down 64% from my May entries. Two slight bags bleeding alongside. My phone shows a voicemail from my accountant, two days old, one sentence: "Call me before year end about the stock sale." The chili smell shifts from dinner toward alarm exactly as I understand what he meant.

Stakes: I had sold $12,000 of Vanguard index fund units earlier that fall for a home repair, booking approximately $4,900 of gains. Without offsetting losses, those gains taxed cleanly at 15% federal plus state. With offsetting losses, potentially zero. Between those outcomes sat about $1,100 — worth one evenin of paperwork and accepting some regrettable prices.

The complication making this interesting rather than routine: wash-sale rules, the code section that polices selling losers then rebuying em, by most plain readings excluded property from coverage. "Excluded" meaning different things to different professionals, as I was about to learn twice.

stock logic fails at the border

Under Section 1091, sell a stock at a loss and repurchase within 30 days, deduction disallowed; basis rolls forward instead. Congress wrote that statute listing "stock or securities." Cryptocurrency is property per Notice 2014-21, not security for this purpose — the loophole every tax-Twitter account screamed about all through 2022's wreckage. Sell BTC December 20, buy back Christmas morning, claim loss fully, own identical position hours later. My accountant confirmed the letter of the law permitted exactly this. Then he paused and said something worth framing: everything legal is not everythin safe, because statutes get amended retroactively more frequently than taxpayers expect.

running both tracks simultaneously

We split my positions deliberately. Half executed straightforward loss harvesting with no same-asset rebuy inside 30 days — cleanest treatment under any future reading, losses recognized permanently since my basis honestly reset lower on repurchases made weeks apart in January. The other half played the aggressive version: sold ALGO December 29, repurchased January 3, harvesting $1,890 while maintaining exposure through the bounce nobody predicted would become 2023's entire rally. Documented contemporaneously with memos explaining reasoning, screenshots timestamped, advice file created, cuz contemporaneous documentation is the difference between confident position and expensive apology should examination ever come.

the January 3 buyback nobody designs on purpose

Five days exposure-free felt longer than five weeks emotionally, watching ALGO tick upward without me while cash sat in settlement limbo earning nothing measurable. Rebuy executed January 3 near $0.098, slightly above my exit price of $0.094 — the honest cost of maintaining continuity rather than a round-trip profit parade some forums imply. Total economic effect across the window approximately negative $41 against pure hold, treated transparently as the price of documentation-friendly structure.

HSA-adjacent thinking, applied badly initially

First pass, I modeled harvested losses like flexible spending — use em or lose them annually. Wrong mental model, corrected sharply by reality each April since. Capital losses first offset gains same-type (long versus short), then cross over up to $3,000 against ordinary income, remainder carries forward indefinitely. After full application, my 2022 return showed net capital loss carryforward of $930 entering 2023, which then absorbed part of 2023's rebalance gains, which altered which lots those sales should draw from — abruptly lot selection methodology mattered across years, compounding complexity like compound interest's evil sibling. Specific identification election filed consistently with broker confirmations now sits filed behind three years of returns.

April delivered its surprise anyway

Filing spring 2023 revealed the odd wrinkle my title promises: state treatment diverged. My state conformed partially to federal property classification but applied its own version ordering rules differently, effectively allowing less loss absorption than federal schedules projected. Effective federal-side benefit landed near projections; combined effect still cut my total bill approximately 41% versus no-harvest baseline but modestly short of early estimates owing to that mismatch. Nothing fraudulent, nothin recoverable, just friction between parallel codes written decades apart assuming neither Bitcoin nor home repairs existed.

what 2026 changed retroactively-ish

Current landscape renders my December dance mostly historical: stablecoin-related legislation passed summer 2025 included provisions extending wash-sale treatment to digital assets prospectively for transactions after January 1, 2026. No clawback threatened completed prior-year strategies, which my contemporaneous file makes easy to demonstrate were compliant when executed. New harvesting currently requires honestly varied substitutes instead of instant repurchases — swap into differently-constructed exposure like replacing direct ALGO with somethin uncorrelated, wait out windows properly, or simply accept market risk gaps during blackout periods. The free lunch expired officially. My takeaway stands regardless: harvesting works best paired with actual portfolio changes you'd make anyway, positioned within broader plans discussed with your investment advisor rather than engineered purely around code sections vulnerable to amendment sessions.

the burned pot, revisited

That evening ended around 9:40 PM with orders placed, spreadsheets exported, and a smoke detector announcing dinner's conclusion before I did. Scraped the pot twice that night and once more next morning, ultimately soaking it overnight with dishwasher tablets, which internet strangers swear by and actually work surprisingly well. Order execution confirmations printed beside the ruined chili recipe, filed together cuz absurdity deserves preservation. Opened that same filing cabinet last month pulling records for this year's harvest planning and laughed properly at the pairing — burned dinner and booked losses, both cleaned up eventually, both costing less than leaving em festering would have.