Sezonov v. Commissioner: why a rebuilt hours log lost

By Max Medvedev · 6 Aug 2026

A log built at filing time is a story, not a record

Sezonov v. Commissioner, T.C. Memo 2022-40, is the short-term rental case that turned on recordkeeping alone. The taxpayers kept no log as the work happened and put the hours together afterward. The Tax Court dismissed those reconstructed hours as "ballpark guesstimates," the material participation claim failed, and the loss did not reach other income.

What happened

Rental losses are passive by default — §469(c)(2) makes a rental activity passive per se, and passive losses cannot offset W-2 income. Short-term rentals get out through Reg. §1.469-1T(e)(3)(ii)(A): when the average period of customer use is seven days or less, the activity is not treated as a rental activity at all.

That is one gate, not the whole test. Once the automatic passive label is gone, the loss only becomes non-passive if the owner materially participates — one of the seven tests in Reg. §1.469-5T(a), and for a single-property owner usually test 3: more than 100 hours, and more than any other single individual.

Sezonov arrived at that second gate with nothing kept as the work happened. The hours behind the participation claim were assembled later, after the year had closed.

What the court held

The reconstructed hours were dismissed as "ballpark guesstimates." The phrase is borrowed from Moss v. Commissioner, 135 T.C. 365 (2010), where the court held that the regulations "do not allow a post-event ballpark guesstimate." With no credible hour count, material participation was not established.

The consequence is mechanical. A short-term rental that clears the average-stay test but fails material participation is passive again: the loss routes through Form 8582 and sits suspended, waiting for future passive income, instead of running from Schedule E line 21 to line 22 in full and landing on the W-2.

The tell

The tell in Sezonov is not a number in the log. It is the order of events — work first, record second — and it is visible before anyone reads a single entry.

This is where the regulation and the bench part company, and the gap is the whole game. Reg. §1.469-5T(f)(4) says participation may be established "by any reasonable means," that "contemporaneous daily time reports … are not required," and that appointment books, calendars or narrative summaries can suffice. Read on its own, that sentence invites reconstruction. Then the cases arrive: Moss, Mirch, Sezonov — each one rejecting a record that did not exist when the work did.

A rebuilt log also carries a signature the examiner is trained to read, because memory produces regular numbers and real work does not:

Mirch v. Commissioner, T.C. Memo 2025-128, is the modern catalogue of exactly that list: roughly 920 hours claimed, 744.5 of them site management and on-call, 168 cleaning hours claimed against a deduction for professional cleaners — and credible hours under 100.

What this means for your records

The rule that falls out of Sezonov is short: the record has to exist before the return does. Everything else is detail on top of that.

In practice a defensible participation file is a log plus corroboration, not a log alone:

• Dated when it happened, not batched at year end.
• Sized to the task — start and stop times, unequal durations, because real turnovers are unequal.
• Attached to something independent — the guest message thread, the cleaner invoice, the supply receipt, the calendar entry, the bank record, the timestamped photo.
• Including the other runner's hours. Test 3 is a race against the single highest-hour other individual, so the manager's or cleaner's time has to be documented too — the lesson from Pohoski v. Commissioner, T.C. Memo 1998-17.
• Kept for seven years, which is the retention window for participation logs.

That is what a real-time log with evidence attached is for. Each entry is written the day the work is done and cites a record that already existed for its own reasons — a message, an invoice, a receipt. A file built that way is not a reconstruction to defend later; it is the same evidence an examiner would ask for, already in order. It also lets the tells above be caught during the year, when a thin month can still be worked on, rather than in an exam when it cannot.

See where your hours land

The estimator runs the same sequence a case like this turns on: the average-stay math first, then all seven participation tests against your hours and the highest other individual's. Run your year through the estimator and it shows which test you pass or miss, with the citation behind each number.

For the full evidence stack — what counts, what never counts, and how the hour race is scored — start at how to prove your short-term rental hours.