- The 90% cap is real, current law for tax years beginning after December 31, 2025 (tax year 2026). The One Big Beautiful Bill Act (P.L. 119-21, §70114) amended IRC §165(d): wagering losses are deductible only up to 90% of losses, only against wagering gains, itemizers only, with no carryforward.
- It only reaches Kalshi or Polymarket trading if that activity is characterized as wagering — a question the IRS has not answered. If the same activity is instead treated as a capital asset or a Section 1256 contract, the 90% cap simply doesn't apply.
- Under a gambling characterization, it can create "phantom income" — tax owed on money a trader never kept — because the deduction equals the lesser of 90% of losses or total winnings, so even a break-even year can leave taxable income on the table.
- There's no carryforward — not for the disallowed 10%, and not for losses beyond that year's winnings. A loss year under gambling treatment typically produces no benefit beyond offsetting that same year's winnings.
- Every situation varies — most traders model their own gross wins and losses and confirm the applicable treatment with a CPA who understands trader and wagering taxation before filing.
The 90% wagering-loss cap itself is settled statutory law — it isn't proposed, it's in force for tax year 2026. What remains unsettled is the implementing detail and the prediction-market application. As of September 2026, the IRS and Treasury have issued no formal guidance addressing whether Kalshi or Polymarket event contracts are wagering transactions for purposes of §165(d) — no revenue ruling, notice, regulation, published private letter ruling, or FAQ that we're aware of. The implementing regulations for the cap's own mechanics (REG-113229-25) remain proposed, not final: the NPRM published April 17, 2026, comments closed June 16, 2026, and a public hearing was held July 17, 2026, where Rep. Dina Titus, the American Gaming Association, CPAs, and professional gamblers all pressed the phantom-income problem described on this page. No final rule had been issued as of this update. This is an actively evolving area; we update this page as guidance lands.
Written by the TraderTax Editorial Team and checked against primary sources rather than other tax blogs: the Internal Revenue Code (§61, §165(d), §1221, §1234A, §1256), the One Big Beautiful Bill Act (P.L. 119-21 §70114) and its still-proposed implementing regulations (REG-113229-25), IRS form and topic pages, and the pending federal bills that have tried to change the cap. The Editorial Team are researchers and writers — not CPAs. TraderTax is a referral platform; filings are handled by independent licensed CPAs in our partner network. Where a question is genuinely unsettled, this page says so instead of picking a side, and the primary sources are linked at the bottom of the page.
The One Big Beautiful Bill Act made a change that got surprisingly little attention when it passed in July 2025, given how much it can cost active bettors and prediction-market traders starting with tax year 2026: it capped the wagering-loss deduction at 90% of losses. Not 100%. Ninety percent, and only up to the amount won, with no carryforward for the rest.
For a casual gambler with a small loss for the year, that mostly doesn't matter. For an active Kalshi or Polymarket trader running six- or seven-figure gross volume through hundreds of positions, it can matter enormously — because the math can leave a trader who broke even, or even lost money in real terms, owing tax on income that was never actually kept. That's the "phantom income" problem this page walks through in full, with a worked numeric example. It's also, as of this writing, the single biggest open question in prediction-market taxation for 2026 — bigger than any individual platform's 1099 practices — because whether it applies to event contracts at all depends on a characterization question nobody, including the IRS, has answered yet.
The 90% cap is real, current law for tax year 2026 — but it only reaches prediction-market trading if that activity is characterized as a wagering transaction under §165(d). If a trader's Kalshi or Polymarket activity is instead treated as a capital asset (Form 8949 / Schedule D) or a Section 1256 contract (Form 6781), the 90% cap simply doesn't apply — losses net against gains in the ordinary way, with no haircut and no phantom income. No IRS guidance says which treatment is correct for event contracts, and reasonable practitioners land in different places. Everything below explains the mechanic and the stakes; it does not resolve the characterization question, because nobody currently can.
What Is the 90% Gambling-Loss Cap?
It's a 2026 amendment to IRC §165(d): under Section 70114 of the One Big Beautiful Bill Act (P.L. 119-21), the deduction for wagering losses is now capped at 90% of those losses — and even that 90% is allowed only up to the amount of wagering gains for the year, with no carryforward of whatever's left over. It applies for tax years beginning after December 31, 2025.
Before this change, §165(d) already limited gambling-loss deductions to the amount of gambling winnings (no net loss deduction ever existed), but it allowed the full amount of losses up to that ceiling. OBBBA didn't touch the winnings ceiling — it added a second, independent haircut on top of it: even losses that don't exceed winnings are now only 90% deductible. The practical effect is that gambling treatment went from "no worse than break-even" to "capable of producing a real tax bill on a break-even or even a losing year," which is the entire subject of this page.
- Gross winnings are reported as income (typically Schedule 1, line 8b for casual filers).
- Deductible losses are the lesser of 90% of gross losses, or gross winnings — and only for taxpayers who itemize (Schedule A).
- Standard-deduction filers get no loss offset at all under this characterization — the full amount of winnings is taxable.
- Nothing carries forward. A loss year produces no future-year benefit beyond offsetting that year's own winnings.
Does the 90% Cap Apply to Kalshi and Polymarket Trading?
Only if that trading is characterized as a wagering transaction — a question the IRS has not answered. Practitioners typically weigh capital-gain, Section 1256, and gambling treatment side by side for event contracts; sports-outcome contracts are generally viewed as carrying the highest recharacterization risk, while economic and political contracts present a stronger capital-asset case.
This is genuinely unresolved, not a close call TraderTax is willing to call one way. Kalshi is a CFTC-designated contract market — the same regulatory category as the CME — and its own regulatory filings describe its event contracts as swaps under the Commodity Exchange Act. Whether a swap on a binary event outcome is a "wager" for tax purposes, a capital asset, a Section 1256 contract, or something the tax code simply hasn't defined yet is exactly the open question. Practitioners typically frame it as a spectrum rather than one answer for every contract: a CPI or Fed-funds-rate contract looks a lot like a financial instrument; a Super Bowl winner contract looks a lot like a sports bet. Most Kalshi and Polymarket accounts hold a mix. For the fuller treatment of that four-way characterization debate — including the Nadex precedent and the ongoing state-court split over whether sports event contracts are gambling as a regulatory matter — see our Kalshi taxes guide.
What that means practically for this specific cap: it is not automatic. A trader who takes a capital-gain or Section 1256 position on their event-contract activity — positions many practitioners consider at least as defensible as the gambling position, and in the mainstream-default case (capital gain) more so — is not subject to the 90% cap on that activity at all. The cap is a real risk sitting inside one of several plausible outcomes, not a certainty for every prediction-market trader.
What Is Phantom Income, and How Does It Happen?
Phantom income is tax owed on money a trader never actually kept. Under gambling characterization, the allowed deduction equals the lesser of 90% of losses or total winnings — so even a trader who broke even for the year can end up with taxable income, because the disallowed 10% of losses never offsets the winnings that get reported.
The mechanism is the 90% haircut stacked on top of the pre-existing winnings ceiling. Gross winnings go on the return as income in full. Losses are only 90% deductible, and even that 90% can't exceed the winnings figure. Whatever gap that leaves between the winnings reported and the losses actually allowed is taxable — regardless of what happened economically over the full year.
The walkthrough below uses round numbers to show the mechanic. It assumes the trader's prediction-market activity is characterized as wagering under §165(d) — a characterization the IRS has not confirmed applies to Kalshi or Polymarket trading, and one many practitioners consider the least likely of the plausible positions for non-sports contracts. The numbers illustrate how the formula works, not a claim about how any particular trader's activity would or should be characterized, or what any specific person will owe.
The setup: a trader who broke even for the year
A prediction-market trader wins $50,000 across resolved and sold positions during 2026, and loses $50,000 across other positions. Net economic result: exactly $0 — a break-even year. Assume the trader itemizes deductions and sits in the 24% federal bracket.
| Step | Amount |
|---|---|
| Gross wagering winnings for the year | $50,000 |
| Gross wagering losses for the year | $50,000 |
| Net economic result | $0 — break-even |
| 90% of gross losses (0.90 × $50,000) | $45,000 |
| Allowed loss deduction (lesser of 90% of losses, or total winnings) | $45,000 — itemizers only |
| Taxable wagering income (winnings − allowed deduction) | $5,000 — phantom income |
| Approx. federal tax at a 24% bracket, itemizing | ≈ $1,200 — on a trader who broke even |
The itemizing checkbox makes it worse before it makes it better. A trader with the same $50,000/$50,000 result who does not itemize gets zero loss deduction at all under this characterization — the full $50,000 of gross winnings becomes taxable, for a federal bill of roughly $12,000 at a 24% bracket, on a trader who broke even. Itemizing typically also means giving up the standard deduction elsewhere on the return, which can partly offset the benefit of unlocking the $45,000 deduction in the first place.
Compare that to the same $0 net year under capital-gain or Section 1256 treatment: losses net fully against gains, and the tax bill is $0. Same trades, same year, same broker statements — the only variable is which characterization applies, and nobody currently knows which one is correct for event contracts.
| Treatment applied to the same $50,000/$50,000 year | Approx. federal tax |
|---|---|
| Capital gain (Form 8949 / Schedule D) | $0 |
| Section 1256 (Form 6781, 60/40) | $0 |
| Gambling (§165(d)), itemizer | ≈ $1,200 on $5,000 of phantom income |
| Gambling (§165(d)), standard deduction | ≈ $12,000 on the full $50,000 of gross winnings |
A more general way to see the mechanic: because the allowed deduction is the lesser of 90% of losses or total winnings, phantom income shows up whenever gross losses run below roughly 111% of gross winnings — which includes break-even years and modestly profitable years alike. A trader whose losses run well above that — a real loss year, not a break-even one — can see the full loss amount effectively absorbed by the winnings-based ceiling once itemizing, without additional phantom income from the 90% haircut itself, though there's still zero benefit for anything beyond that year's winnings, since nothing carries forward either way.
How Is This Different From Section 1256 or Capital-Gains Treatment?
Under capital or Section 1256 treatment, losses typically net fully against gains in the same year — plus up to $3,000 a year against ordinary income, with the remainder carried forward indefinitely (Section 1256 also allows a 3-year loss carryback election against prior 1256 gains). A break-even year under either treatment typically owes $0. The 90% cap exists only inside gambling treatment under §165(d) — it has no equivalent, and no analog, under capital or Section 1256 treatment.
That's the whole reason the characterization question carries so much weight for 2026 specifically. In earlier tax years, the difference between "my Kalshi activity is a capital asset" and "my Kalshi activity is a wager" mattered for rate treatment and paperwork, but a net loser typically didn't owe federal tax either way. Starting with 2026 returns, gambling characterization can produce an actual tax bill in a year with zero or negative real profit — something none of the other three candidate treatments can do. Our prediction market tax calculator models all three side by side on a trader's own numbers, including this cap and the itemizing toggle.
When Does the 90% Cap Take Effect?
It applies to tax years beginning after December 31, 2025 — meaning tax year 2026 returns, typically filed in early 2027. It does not apply to 2025 or earlier tax years, regardless of when OBBBA itself was signed (July 2025).
Two dates are easy to conflate and worth separating cleanly: the statute — the 90% cap itself — was enacted in July 2025 and is unconditionally in force for tax year 2026 regardless of anything else happening in Washington. The implementing regulations (REG-113229-25), which are expected to settle mechanical details like the exact computation method and how the cap applies on a joint return, remain proposed rather than final as of this update. A trader filing a 2026 return doesn't get to wait for final regulations — the statutory cap applies now, on the prevailing reading of the text (deduction equals the lesser of 90% of losses or total winnings), even while some of the mechanical fine print is still being worked out.
Is There Any Carryforward for the Disallowed Losses?
No. Unlike capital losses, which typically carry forward indefinitely, amended §165(d) allows no carryforward of the disallowed 10% or of any losses beyond that year's winnings. Under gambling characterization, a loss year typically produces no tax benefit beyond offsetting that same year's winnings — there's nothing to bank for a future profitable year.
This is a meaningful structural difference from every other candidate treatment discussed on this site. A trader who has a rough year under capital or Section 1256 treatment at least banks a loss carryforward against a better year later. Under gambling treatment, a rough year with wagering losses simply produces no future value at all — the loss, and the disallowed 10% of whatever losses were used, both disappear for tax purposes once the year closes.
Does the 90% Cap Apply to Sports Betting and Casino Gambling Too?
Yes — §165(d) as amended by OBBBA is the general wagering-loss provision, so it reaches traditional sportsbook bets, casino gambling, and any other activity characterized as wagering, not only prediction markets. The prediction-market-specific question is narrower: whether event-contract trading counts as wagering in the first place.
That distinction is important because it means the 90% cap itself is not in doubt for a straightforward sports bettor placing wagers through a licensed sportsbook — that activity has always been squarely inside §165(d), and the OBBBA amendment applies to it without any characterization debate. The 90% haircut and the phantom-income mechanic described above hit sportsbook bettors exactly the same way they'd hit a prediction-market trader whose activity was found to be wagering. Sports-focused prediction-market contracts sit in an interesting middle position — Congress's own framing of the amendment (and the CFTC's separate, non-tax regulatory fight over whether sports event contracts are "gaming") is part of why practitioners rank sports-style contracts as the highest-risk category for recharacterization on the prediction-market side. See our sports prediction market taxes guide for how this compares to a traditional sportsbook bet specifically.
Where Do Things Stand on Implementation?
The statute is settled; the fine print is still being worked out, and repeal attempts have not gone anywhere so far. The proposed regulations (REG-113229-25) were published April 17, 2026; comments closed June 16, 2026; a public teleconference hearing was held July 17, 2026, where Rep. Dina Titus, the American Gaming Association, CPAs, and professional gamblers all pressed the phantom-income problem described on this page. No final rule had been issued as of this update.
On the legislative side, two bills have tried to undo or soften the cap, and both have stalled so far: the FAIR BET Act (H.R. 4304) was blocked in the House Rules Committee in January 2026, and a Senate unanimous-consent attempt on the FULL HOUSE Act (H.R. 6985) was blocked as well. Neither has had a floor vote in either chamber as of this update. Worth saying plainly rather than leaving readers hoping: as things currently stand, the 90% cap is not going away for tax year 2026, and traders modeling their exposure should plan around the statute as written rather than around a repeal that has not happened.
Frequently Asked Questions
What is the 90% gambling-loss cap?
The One Big Beautiful Bill Act (P.L. 119-21, §70114) amended IRC §165(d) so that, for tax years beginning after December 31, 2025, wagering losses are deductible only up to 90% of losses — and only against wagering gains, with no carryforward. Itemizers only; standard-deduction filers get no offset at all.
Does the 90% cap apply to Kalshi and Polymarket trading?
Only if that trading is characterized as a wagering transaction — a question the IRS has not answered. Practitioners typically weigh capital-gain, Section 1256, and gambling treatment side by side; sports-outcome contracts are generally viewed as carrying the highest recharacterization risk, while economic and political contracts present a stronger capital-asset case.
What is phantom income, and how does it happen under this rule?
Phantom income is tax owed on money a trader never actually kept. Under gambling characterization, the allowed deduction is the lesser of 90% of losses or total winnings — so even a trader who broke even for the year can have taxable income, because the disallowed 10% of losses never offsets the winnings that get reported.
How is this different from Section 1256 or capital-gains treatment?
Under capital or Section 1256 treatment, losses typically net fully against gains in the same year (plus up to $3,000 a year against ordinary income, with the rest carried forward) — a break-even year typically owes $0. The 90% cap exists only inside gambling treatment under §165(d); it has no equivalent under capital or 1256 treatment.
When does the 90% cap take effect?
It applies to tax years beginning after December 31, 2025 — meaning tax year 2026 returns, typically filed in early 2027. It does not apply to 2025 or earlier tax years, regardless of when the OBBBA itself was signed.
Does the 90% cap apply to sports betting and casino gambling too?
Yes — §165(d) as amended by OBBBA is the general wagering-loss provision, so it reaches traditional sportsbook bets, casino gambling, and any other activity characterized as wagering, not only prediction markets. The prediction-market-specific question is narrower: whether event-contract trading counts as wagering in the first place.
Is there any carryforward for the disallowed losses?
No. Unlike capital losses, which typically carry forward indefinitely, amended §165(d) allows no carryforward of the disallowed 10% or of any losses beyond that year's winnings. Under gambling characterization, a loss year typically produces no tax benefit beyond offsetting that same year's winnings.
Does the 90% cap apply to full-time or professional gamblers too?
Reportedly yes. Filers who meet the Groetzinger standard for a gambling trade or business typically report on Schedule C rather than as a casual gambler, but the amended statute folds ordinary wagering-related business expenses into "losses" for this purpose — so combined losses and expenses face the same 90%-of-losses-or-winnings cap, and any resulting profit remains subject to self-employment tax.
Has anyone tried to repeal or change the 90% cap?
Yes, so far without success. The FAIR BET Act (H.R. 4304) was blocked in the House Rules Committee in January 2026, and a Senate unanimous-consent attempt on the FULL HOUSE Act (H.R. 6985) was blocked as well. As of September 2026, no repeal or fix has passed, and the cap stands for tax year 2026.
See what the 90% cap could mean for your own numbers — across all the possible treatments
Because the IRS hasn't ruled on event-contract characterization, the honest answer is a range, not a number. Our calculator models the capital-gain, Section 1256, and gambling branches side by side — including this 90% loss rule and the itemizing toggle — using a trader's own gross wins and losses. The tools below give a ballpark; a CPA confirms what actually applies.
Or create a free account to get matched with a CPA →The 90% cap is one piece of a bigger characterization picture. Here's the rest of the prediction-market tax cluster:
Important framing: none of the IRS pages below mentions prediction markets or event contracts — as of this update, no IRS page does. They are the primary material for the wagering-loss framework and forms the 90% cap runs through, so a trader (or a CPA) can read the source directly rather than a summary of it.
- IRS — Topic No. 419, Gambling Income and Losses. The baseline framework the amended §165(d) sits inside. Note it describes pre-2026 rules in places; the OBBBA 90% cap applies to tax years beginning after December 31, 2025.
- IRS — About Schedule A (Form 1040), Itemized Deductions. Where a gambling-loss deduction is claimed — and why standard-deduction filers get none.
- IRS — About Schedule 1 (Form 1040). Where gross wagering winnings are typically reported as other income.
- IRS — About Form 8949 and Schedule D. The capital-gain path, which is not subject to the 90% cap at all.
- IRS — About Form 6781, Gains and Losses From Section 1256 Contracts and Straddles. The 60/40 path, also not subject to the 90% cap.
- IRS — About Form 1040-ES, Estimated Tax for Individuals. Relevant because no prediction-market platform or sportsbook appears to withhold on winnings.
Statutory and regulatory citations used on this page: IRC §61 (gross income), §165(d) as amended (wagering losses, the 90% cap), §1221/§1234A (capital assets and terminations), §1256 (60/40 treatment), P.L. 119-21 §70114 (OBBBA), REG-113229-25 (proposed implementing regulations), H.R. 4304 (FAIR BET Act, unenacted), and H.R. 6985 (FULL HOUSE Act, unenacted).