OBBBA's 90% Gambling-Loss Cap: The Statute, the Confirmed Formula, and the Repeal Risk

Our first post on the 2026 gambling-loss cap (read it here) covers the mechanism and the math. This one goes one level deeper: the exact statute, what the IRS’s own proposed regulations say about how the formula actually stacks, and a live repeal effort that could take the whole thing off the table before most people ever file under it.

The statute, by name

The cap comes from the One Big Beautiful Bill Act, Public Law 119-21, signed July 4, 2025. Section 70114 of that law amends IRC §165(d), the wagering-loss provision, to limit the deduction to 90% of losses. It takes effect for tax years beginning after December 31, 2025, so the first return it touches is tax year 2026, filed in early 2027. Tax year 2025 is unaffected.

The order of operations, now confirmed

The formula is deductible = min(0.90 × losses, winnings), available only if you itemize. What used to be a genuine open question is which cap applies first: 90% of losses, then capped at winnings, or capped at winnings first, then reduced by 10%. The Treasury’s proposed regulations (REG-113229-25, published in the Federal Register April 17, 2026) settle it. The operative text: “The proposed regulations would amend this sentence to limit the deduction to 90 percent of the amount of wagering losses during a taxable year and only to the extent of gains from wagering transactions during a taxable year.” Losses get the 90% haircut first; the §165(d) winnings cap applies second.

In a breakeven year both orders give the same answer, which is why the question looked academic. It stops being academic once losses run past winnings. Win $201,000 and lose $220,000: deductible = min(198,000, 201,000) = $198,000, so you are taxed on $3,000 in a year you lost $19,000. Reverse the order (cap at winnings first, then take 90%) and you would get $180,900 deductible instead, a meaningfully different, and wrong, answer.

This is the same order WagerMints’s gambling-treatment comparison already uses, not as a judgment call but because it’s now the order the government’s own proposed regulation describes. The regulations aren’t final yet (comments closed June 16, 2026, with a hearing contemplated), so treat this as the strongest available reading rather than settled law until Treasury issues a final rule.

None of this changes the §165(d) gate: losses only count if you itemize. On the standard deduction the cap is beside the point, because your losses were already worth $0.

It might not survive the year

There’s an active push to repeal the 90% cap before it lands on anyone’s return. NATP reported in 2026 that Congress may repeal it, citing pressure from Nevada legislators and the phantom-income problem described above. No repeal bill had passed as of this writing, and nothing has amended, delayed, or sunset the provision. It’s the current law for tax year 2026, but it’s also a live target, and a mid-year repeal would change every number in this post retroactively for the year.

What to actually do with this

If your prediction-market activity gets characterized as gambling, the cap is real for now and it’s worse than most summaries suggest, because the 90% comes off before the winnings cap, not after. Compare it against the other three methods (covered here) before you assume gambling treatment is your only option, and if you’re in a state that also disallows the deduction, read how state rules stack on top of this. Then keep an eye on Washington: this specific number could change before you file.

Want your own numbers? The free 2026 gambling loss calculator shows what you can deduct and how much gambling adds to your taxable income, under the old rules and the new.