Last Updated on April 4, 2026 by Patrick Camuso, CPA
Quick answer (read this first):
The short answer: It depends entirely on how your prediction market activity is characterized, and the rules differ so significantly across frameworks.
Why prediction market loss deductions matter: Under capital or ordinary income treatment, losses net against gains and excess losses carry forward. Under gambling treatment, losses are deductible only to the extent of gross winnings, only if you itemize, and beginning in 2026, only up to 90 percent of those winnings. A trader with $200,000 in losses and $150,000 in gains nets a $50,000 loss under capital treatment. Under gambling treatment, that same trader may owe tax.
What the IRS has not done: Issued any notice, ruling, or guidance that resolves how prediction market contracts should be characterized. Until it does, loss treatment is determined by the framework applied, and that framework is a documented analytical decision, not a default.
The practical posture: Loss treatment is one of the three most consequential differences between characterization frameworks, alongside rate and the 2026 phantom income problem. It should be analyzed before filing, not after losses have already been reported in a way that cannot be defended.
Why Loss Treatment Matters
Most public commentary on prediction market taxes focuses on rates. The gambling versus capital gains discussion typically centers on which rate is lower, which framework is more favorable, and whether Section 1256’s 60/40 blended rate is available.
For most active prediction market traders, positions are short-duration. They open and close within days or weeks. The practical rate difference between short-term capital gain treatment and ordinary income treatment is zero. Both are taxed at ordinary rates. The Section 1256 rate advantage, where applicable, produces a modest benefit for most income levels.
The rules governing what a trader can do with prediction market losses vary dramatically across the available characterization frameworks, and those differences compound materially at scale. A trader generating $500,000 in gross activity across a year, wins and losses roughly offsetting, faces a fundamentally different tax outcome depending on which loss framework applies. Understanding those differences is the starting point for any defensible prediction market reporting strategy.
How Loss Treatment Works Under Each Framework
Under Section 1001, a loss is recognized only when a taxpayer has disposed of the contract through sale, settlement, expiration, or abandonment for a determinable amount. An open position that has declined in value has not produced a deductible loss. The loss is recognized at disposition. This sequencing matters practically because prediction market contracts that expire without value produce a recognized loss equal to basis at the moment of worthless settlement, not at the moment the contract became economically unlikely to pay out. The characterization framework that applies then determines how that recognized loss is treated.
Capital Gain or Loss Treatment
Under capital treatment, prediction market contracts are analyzed as property under the capital asset rules of Section 1221. Gains and losses are netted against each other at year end. Short-term capital losses offset short-term capital gains first, then long-term capital gains. Long-term capital losses offset long-term capital gains first, then short-term capital gains.
Where net capital losses exceed net capital gains for the year, the excess is deductible against ordinary income up to $3,000 per year for individual filers ($1,500 for married filing separately). Losses beyond that limit carry forward indefinitely into future tax years, available to offset future capital gains or up to $3,000 of ordinary income per year until exhausted.
For a prediction market trader with a net losing year, capital treatment produces a real, usable loss benefit. The $3,000 annual deduction against ordinary income is modest, but the carryforward is permanent. A trader with $50,000 in net capital losses in Year 1 retains that loss and applies it forward against gains in subsequent years, with no expiration date.
The capital treatment loss framework also allows prediction market losses to offset capital gains from other investments. A trader who lost money on Kalshi contracts but realized gains from stock positions can offset those gains with prediction market losses, reducing the tax on the equity portfolio. That cross-asset offsetting is not available under gambling treatment.
Ordinary Income Treatment
Under ordinary income treatment, prediction market contracts are analyzed as contingent financial instruments generating ordinary income or loss upon settlement. Gains and losses are netted against each other in the same manner as capital treatment. There is no prohibition on netting under this framework.
The loss treatment distinction between ordinary income and capital treatment is narrow for most traders. Both frameworks allow annual netting. Ordinary losses reduce ordinary income directly, without the $3,000 limitation that applies to net capital losses against ordinary income. In a year with a net loss from prediction market activity under ordinary income treatment, that loss reduces other ordinary income dollar for dollar. That is generally more favorable than the capital loss framework in a net loss year.
There is no carryforward mechanism for ordinary losses in the same technical sense as capital loss carryforwards, but the full loss is generally deductible in the year incurred against other ordinary income, which for most traders produces a better immediate result than the $3,000 annual cap on capital losses against ordinary income.
One additional distinction applies for traders whose activity rises to the level of a trade or business under Section 162. Where a taxpayer qualifies as a trader in prediction market contracts under the trade or business standard, losses from that activity may be treated as ordinary business losses reportable on Schedule C, deductible against all income without limitation. That determination is fact-intensive and depends on the frequency, regularity, and primary purpose of the trading activity. For most retail participants, ordinary income treatment under a contingent contract framework is the more commonly applicable path.
Ordinary income treatment and gambling treatment are not the same framework. These two categories are analytically distinct. Ordinary income from a financial contract and ordinary income from wagering produce ordinary income at the same rate, but the loss rules are entirely different. A practitioner or software that conflates ordinary income treatment with gambling treatment is making a material error.
Gambling and Wagering Treatment
Where gambling treatment applies under Section 165(d), the loss framework is the most restrictive of any available characterization.
Losses are not netted across sessions. The IRS requires gambling activity to be tracked on a per-session basis, and netting is limited to activity within a defined session. Gross winning sessions produce taxable income. Gross losing sessions produce an itemized deduction, subject to the annual cap. Winning and losing sessions are reported separately.
Losses are deductible only to the extent of gross gambling winnings for the year. Net losses, where gross losses exceed gross winnings, produce no tax benefit. There is no carryforward. There is no ability to offset capital gains or other income. A trader who loses more than they win in a year under gambling treatment receives zero tax benefit from the excess losses, regardless of the amount.
Losses are deductible only for taxpayers who itemize on Schedule A. A trader who takes the standard deduction, which in 2026 is $16,100 for single filers and $32,200 for married filing jointly, receives no benefit from gambling losses under any circumstances. For the majority of U.S. taxpayers whose standard deduction exceeds their itemized deductions, gambling treatment means losses are entirely nondeductible.
The per-session rule creates a compliance problem specific to prediction market traders. For casino gamblers, a session is a defined visit. You sit down, you play, you leave. The IRS has provided guidance on applying the session concept to casino and slot machine activity. It has not provided any equivalent guidance for prediction market contract trading, where dozens of contracts may open, settle, and expire on any given day across multiple event types. What constitutes a session in that context is undefined under current IRS guidance. That ambiguity creates both a documentation burden and an interpretive risk that does not exist under capital or ordinary income treatment.
Practitioners applying the gambling framework must adopt and document a session methodology, whether organized by calendar day, login period, market category, or another defensible division, and apply it consistently across the full year. That choice directly affects the calculation of gross winnings reportable as income and the ceiling on deductible losses. Two practitioners applying different session definitions to identical trading activity will produce different gross income figures and different deductible loss amounts even before reaching the 90 percent cap.
A further risk specific to online prediction market activity: the IRS could apply a one-contract-one-session interpretation, treating each individual contract settlement as its own session. If that position were applied, cross-contract netting would be eliminated entirely for non-professional traders. Every winning contract would produce gross taxable income. Losses from losing contracts would be deductible only against those specific winnings under itemization limits. In any year where profitable and unprofitable positions are distributed unevenly across the trading calendar, this interpretation would produce phantom income far exceeding the 10 percent cap created by the OBBBA. No IRS guidance has adopted this interpretation, but it has not been foreclosed either.
A separate consideration applies to traders who meet the legal standard for professional gambler status under Commissioner v. Groetzinger, 480 U.S. 23 (1987). Where a taxpayer’s gambling activity is pursued full time, in good faith, and with regularity and continuity for the purpose of making a living, that activity may constitute a trade or business under Section 162. A professional gambler reports gambling income and losses on Schedule C, which eliminates the itemization requirement and allows gambling losses to be deducted as ordinary business expenses against other income. For high-volume prediction market traders whose activity is regular, substantial, and profit-motivated, professional gambler status is a facts-and-circumstances question worth analyzing, but it requires documented evidence of the business-like nature of the activity. Under the OBBBA, the 90 percent loss cap applies equally to professional gamblers, and Schedule C losses from gambling activity are also subject to the excess business loss limitation under Section 461(l), which caps the amount of business losses that can offset non-business income in a given year. Practitioners evaluating professional gambler status for prediction market traders should factor both the OBBBA cap and the Section 461(l) limitation into the economic analysis before concluding the Schedule C path produces a better outcome.
The 2026 Loss Limitation: Why Gambling Treatment Got Materially Worse
Beginning January 1, 2026, the One Big Beautiful Bill Act changed the gambling loss deduction in a way that significantly increases the economic stakes of gambling characterization for active traders.
Prior to 2026, a taxpayer who itemized could deduct gambling losses up to the full amount of gambling winnings for the year. A trader with $300,000 in gross winnings and $300,000 in gross losses could deduct $300,000 against $300,000, producing zero net taxable gambling income. The economics were bad. No carryforward, no cross-asset offsetting, session tracking required. A breakeven trader who itemized paid no tax.
Beginning in 2026, deductible gambling losses are capped at 90 percent of gambling winnings. The remaining 10 percent of winnings is taxable regardless of how much was lost. The cap applies to the deductible losses, not to the winnings.
A trader with $300,000 in gross winnings and $300,000 in gross losses breaks even economically. Under the 2026 rule, deductible losses are capped at $270,000 (90 percent of $300,000). The remaining $30,000 is phantom income, taxable, unavoidable, and entirely unconnected to any economic gain from the activity. At a 37 percent marginal rate, the federal tax on that phantom income is $11,100. The trader broke even and owes over $11,000.
At higher trading volumes the numbers scale quickly. A trader with $1,000,000 in gross winnings and $1,000,000 in gross losses faces $100,000 in phantom income and a federal tax bill approaching $37,000 on a year with zero net profit.
This phantom income consequence is specific to gambling treatment. Under capital treatment or ordinary income treatment, a trader who breaks even reports no taxable income. The 2026 change makes the economic cost of gambling characterization, relative to capital or ordinary income treatment, larger than it has ever been.
How Platform Reporting Affects Loss Documentation
The documentation required to support prediction market loss deductions varies significantly by framework, and the platforms themselves provide almost none of it.
Under capital or ordinary income treatment, a trader needs contract-level gain and loss records showing acquisition cost, settlement proceeds, and net position by contract. That documentation is the foundation of the Schedule D or ordinary income calculation. It is also the documentation that allows losses to be carried forward. The carryforward amount has to be traceable to specific positions that were closed at a loss.
Under gambling treatment, documentation requirements are more granular. The per-session rule requires a record of each session with the date, the starting and ending position, the gross winnings within the session, and the gross losses within the session. For a prediction market trader, constructing session-level records from platform CSV exports requires making and documenting interpretive decisions about what constitutes a session, which the IRS has not resolved for this asset class.
Kalshi provides limited transaction exports but does not produce gain and loss summaries in a form that maps cleanly to any tax reporting framework. The exports show individual contract activity but do not apply a characterization, do not calculate session-level results, and do not produce a net gain or loss figure that can be directly transferred to a tax form. Polymarket produces no standardized tax documentation at all. Transaction history must be reconstructed from on-chain records.
The documentation burden for loss deductions exists regardless of which framework applies. What changes is what the documentation has to show, how granular it has to be, and what happens when it is incomplete. Under capital treatment, incomplete documentation creates a basis reconstruction problem. Under gambling treatment, incomplete documentation creates both a basis problem and a session characterization problem, with no IRS guidance on how to resolve the latter for prediction market activity specifically.
For a detailed breakdown of what forms cover and what they omit, see our prediction market tax guide.
Loss Deductions and the Section 1256 Question
Section 1256 treatment, where applicable, provides a distinct loss framework that is more favorable than gambling treatment and comparable to capital treatment in most respects.
Under Section 1256, net losses for the year can offset capital gains plus up to $3,000 of ordinary income, the same as general capital loss treatment. There is also a three-year carryback election available for net Section 1256 losses that does not exist under general capital loss rules. A trader with a net Section 1256 loss in the current year can elect to carry that loss back to offset Section 1256 gains from the prior three tax years, which can produce an immediate refund of taxes previously paid on prior-year prediction market gains.
A separate path available to certain traders is the mark-to-market election under Section 475(f)(2) for commodities. Traders who qualify for trader tax status, a facts-and-circumstances determination based on the frequency, continuity, and business-like nature of the trading activity, may be eligible to elect mark-to-market treatment if prediction market contracts qualify as commodities within the meaning of Section 475(e)(2). If the election applies, all gains and losses from those contracts convert to ordinary income and loss, making trading losses fully deductible against ordinary income without the capital loss limitation and without the per-session mechanics of gambling treatment. The trade-off is that any Section 1256 60/40 treatment that might otherwise apply is eliminated, and all gains become fully ordinary. Section 475(f)(2) is untested for prediction market event contracts and the threshold commodity classification question is unsettled. The election carries a five-year lock-in period and cannot be revoked without IRS consent and a user fee. For traders with material recurring losses, it is worth analyzing, but requires careful evaluation of both the qualification threshold and the long-term economic consequences.
The Section 1256 loss framework only applies where the contracts actually qualify under the statute. As covered in our Section 1256 and prediction markets analysis, most current prediction market contracts do not clearly satisfy the statutory definitions required for Section 1256 treatment. The carryback election is an attractive feature, but it does not justify claiming Section 1256 treatment for contracts that do not satisfy the statutory requirements. Practitioners considering Section 1256 should treat the analysis as requiring affirmative justification, not as a default or presumptive outcome.
Straddle Rules and Loss Deferral Risk
Prediction market traders who hold economically offsetting positions face an additional loss risk that most commentary does not address, the straddle rules under Section 1092.
Where a taxpayer holds substantially offsetting positions, such as paired Yes and No contracts on the same event, or layered positions designed to hedge probability outcomes, the IRS may characterize those positions as a straddle. Under the straddle rules, a loss on one leg of a straddle position is deferred to the extent of unrecognized gain in the offsetting position that remains open. The loss is not permanently disallowed. It is deferred until the offsetting position is closed or otherwise disposed of. The deferral can shift a loss from the current tax year into a future year, which affects both the timing of the tax benefit and the year to which any carryforward applies.
For capital treatment positions, straddle loss deferral is a real risk for traders who run hedged book positions or arbitrage strategies across Yes/No contracts on the same underlying event. The deferral rule applies based on the economic offsetting relationship between positions, not on how the trader labels them.
The straddle analysis requires a portfolio-level review of all open and recently closed positions to identify pairs or groups of contracts that may be economically offsetting. Platform-level transaction exports typically do not flag straddle relationships. Traders with structured or hedged approaches to prediction market activity should factor this into both their loss planning and their documentation.
The wash sale rules under Section 1091 present a related but distinct consideration. Wash sales apply to the sale of stock or securities at a loss where substantially identical securities are acquired within 30 days before or after the sale. Prediction market contracts are not stock or securities, and the wash sale rules do not clearly apply to event-based contract dispositions under current law. In structured trading patterns where a trader repeatedly closes and reopens economically similar positions, for example, closing a losing contract position and immediately opening a new contract on the same event or a closely related one, the IRS could argue for wash sale treatment by analogy or under anti-abuse principles. The risk is not well-established for prediction markets, but it is worth noting for traders with high-frequency or systematically structured loss harvesting activity.
Characterization Is a Facts-Based Determination, Not a Selection
Tax characterization for prediction market contracts is not a choice a trader makes at filing time. It is a legal conclusion that follows from analyzing the specific contracts traded, the platform they were traded on, the regulatory structure governing that platform, and the nature of the underlying events. A trader cannot select capital treatment because it produces a better loss outcome, then switch to ordinary income treatment in a year where that produces a better result. The framework is determined by the facts. It applies to the full year’s activity, consistently, once a defensible position is established.
This matters most in the context of losses because the economic differences across frameworks are largest there. A trader who assumes capital treatment without analyzing whether the facts support it is not making a planning choice. They are taking an undocumented position on an unsettled legal question. If the IRS concludes on examination that gambling treatment applies, the capital loss offsets used against other investment gains are unwound, the per-session limitations apply retroactively, and the 2026 phantom income calculation may produce additional tax on years already filed. The exposure compounds across every year the position was applied.
The analytical work has to happen before the return is prepared. That means working through the sequential framework with realization under Section 1001, the wagering threshold under Section 165(d), the Section 1256 requirements if the wagering threshold is cleared, and the capital versus ordinary income analysis if Section 1256 does not apply. Each step depends on the specific facts of the contracts held. Sports-outcome contracts and macro-economic contracts on the same platform in the same account may not support the same characterization. The analysis is contract-level, not platform-level.
An examiner reviewing prediction market activity will evaluate the underlying contracts on their merits and apply the framework supported by those facts, regardless of what the return asserts. A return supported by a documented, sequential analytical position that explains the reasoning and applies it consistently is materially more defensible than one where characterization was assumed or selected for rate purposes.
Loss treatment positions taken without adequate analytical support carry penalty exposure under Section 6662, where accuracy-related penalties apply when a position lacks reasonable basis. A formal position document that works through the statutory frameworks sequentially and documents why the characterization was reached supports a reasonable basis standard and reduces that exposure even if the IRS ultimately reaches a different conclusion. Taxpayers taking positions under significant uncertainty may also consider disclosure on Form 8275, which can further protect against penalties even if the position does not ultimately prevail.
Reporting mechanics follow characterization and are not interchangeable. Under capital treatment, dispositions are reported on Form 8949, capturing acquisition date, disposition date, proceeds, and adjusted basis, with net amounts flowing to Schedule D. Under ordinary income treatment, net gain is reported on Schedule 1, Line 8, as other income with a descriptive label. Under gambling treatment, gross winnings are reported as income and losses flow to Schedule A as an itemized deduction subject to the cap. Assigning income to the wrong schedule is a separate reporting error from the characterization decision itself.
State tax treatment adds a layer that federal analysis alone does not capture. Several states, including North Carolina, do not allow an itemized deduction for gambling losses at the state level. Gross gambling winnings are fully taxable at the state rate with no offset for losses, regardless of what is permitted federally. For traders in states with that treatment, the combined federal and state cost of gambling characterization is significantly higher than the federal calculation alone suggests. State conformity to federal characterization rules varies, and traders in high-income states should confirm their state’s treatment as part of any complete characterization analysis.
One compliance issue that surfaces after the return is filed but originates in characterization decisions made during the year: estimated tax and underpayment penalties. Traders with material prediction market gains recognized mid-year who assumed annual loss netting would offset their liability may be significantly underwithheld. The OBBBA phantom income problem compounds this for 2026. A trader who broke even economically may owe tax on 10 percent of gross winnings that no withholding or estimated payment covered, because the liability did not exist under prior law and would not have been apparent when estimated payments were due. Practitioners advising active traders should flag estimated tax obligations during the year, not only at filing time.
Our prediction market tax reporting services are built around exactly this problem, establishing a defensible characterization framework, applying it consistently, and documenting it in a way that holds under scrutiny.
Frequently Asked Questions: Prediction Market Loss Deductions
Can I deduct prediction market losses on my tax return?
Yes, in most cases, but the mechanics depend entirely on how your prediction market activity is characterized. Under capital treatment, losses net against gains and excess losses carry forward indefinitely. Under ordinary income treatment, losses reduce other ordinary income in the year incurred. Under gambling treatment, losses are deductible only to the extent of gross winnings for the year, only if you itemize, and beginning in 2026, only up to 90 percent of those winnings. The framework that applies determines whether your losses are fully deductible, partially deductible, or, in the case of standard deduction filers under gambling treatment, entirely nondeductible.
What is the 2026 gambling loss cap and how does it affect prediction market traders?
The One Big Beautiful Bill Act, effective January 1, 2026, limits deductible gambling losses to 90 percent of gambling winnings for the year. A trader with $200,000 in gross winnings and $200,000 in gross losses can only deduct $180,000 of losses against $200,000 of income, producing $20,000 of taxable phantom income despite no net economic gain. At a 37 percent marginal rate, the tax on that phantom income exceeds $7,000. This consequence is specific to gambling treatment and does not arise under capital or ordinary income frameworks.
Can I net prediction market gains and losses before reporting them?
Under capital and ordinary income frameworks, gains and losses are netted at the contract level before reporting the net result. Under gambling treatment, the IRS requires activity to be tracked on a per-session basis. Winning sessions and losing sessions are not netted against each other. Gross winning sessions are reported as income and gross losing sessions are taken as an itemized deduction subject to the annual cap. For prediction market traders, the concept of a session has no established definition under current IRS guidance, which creates an additional documentation and interpretation challenge.
Does it matter which platform I trade on for loss deduction purposes?
The platform itself does not determine the loss framework. What matters is the characterization analysis which includes the nature of the underlying contracts, the regulatory status of the exchange, and the consistency of the position taken. Kalshi’s CFTC-regulated status is a relevant factor in the characterization analysis but does not automatically resolve it. Polymarket’s settlement in USDC introduces a separate layer of analysis. The type of contract also matters within a single platform. Sports-outcome contracts carry materially higher gambling characterization risk than macro-economic or political contracts, which means the loss framework most likely to apply may differ across contract types within the same account. A trader running both sports and macro contracts on Kalshi may face different characterization risk profiles for different portions of their activity. The loss rules that apply flow from the characterization conclusion, not from which platform was used.
What documentation do I need to support prediction market loss deductions?
Under capital or ordinary income treatment, contract-level records showing acquisition cost, settlement proceeds, and net gain or loss per contract are required to support loss positions. Carryforward losses require continuity documentation connecting current-year losses to prior-year records. Under gambling treatment, session-level records are required: the date, the gross winnings and gross losses within each session, and supporting transaction detail. Platform exports are typically the starting point for this documentation but are rarely sufficient on their own. For Polymarket, on-chain reconstruction is required.
