Buying Both Sides on Kalshi and Polymarket: Arbitrage, Hedges and the Straddle Rules

Arbitrage profit is taxable like any other gain, but it can cause a timing problem. If you hold offsetting positions, such as YES on one venue and NO on another, closing the loser in December and leaving the winner open into January may not deliver the loss you expect if the straddle rules apply. Whether Section 1092 reaches event contracts hasn't been decided, so treat year-end hedges carefully.

OddsFile is software, not tax advice. The IRS has not issued guidance on how prediction-market contracts are taxed. This article explains the options and the sources behind them; a tax professional can tell you which position fits your situation.

Plain arbitrage

Say YES is $0.46 on one venue and NO is $0.50 on another for the same event, so a matched pair costs $0.96 and pays $1.00 for sure. Each pair earns $0.04 before fees.

If both legs close at settlement in the same year, tax is simple: one leg wins, one leg loses, and your net is the arbitrage profit.

Under capital treatment or other income, you report the $400 net of gains and losses. Note that the two legs are separate rows under capital treatment.

Under gambling treatment, the same trade is much worse. Winnings of $5,400 are taxable, while only 90% of the $5,000 loss is deductible from 2026, and only if you itemize. That leaves $900 taxed on a $400 profit even if you itemize. Worse, if you take the standard deduction and do nothing else, $5,400 is taxed. High-volume arbitrage under gambling treatment can be a tax-negative strategy. See the 90% limit.

Treating the two venues as one session wouldn't be obviously right either. There's no rule for it.

The straddle rule

Section 1092 limits losses on one leg of a straddle. Its core idea: a loss counts this year only to the extent it exceeds unrecognized gain in an offsetting position.

Under the regulations, actively traded property includes any personal property for which there's an established financial market, and that list includes a US board of trade designated as a contract market by the CFTC, an interdealer market, and comparable foreign exchanges.

Does it reach event contracts?

Nobody knows. Consider the pieces.

• Offsetting. YES and NO on the same event clearly offset. Different venues don't change that economically.
• Personal property. Section 1092(d)(1) asks whether the property is of a type that's actively traded. A Kalshi contract on a CFTC-designated market has an argument for yes. A Polymarket token on an offshore order book is less clear, because the listed "established financial markets" don't obviously include it.
• Position with respect to personal property. The statute generally reaches interests in personal property, including futures, forwards and options. An event contract may fit, but no authority says so.

The IRS hasn't applied this to event contracts. The sensible way to handle it is to assume it could apply if you've engineered year-end losses, and to get advice if the amounts matter.

The year-end problem

The rule matters when legs close in different tax years. Suppose your hedge is:

• Position A has a $5,000 loss on December 28, 2026. You close it.
• Position B has a $5,000 unrealized gain. You leave it open until January.

If Section 1092 applies, your $5,000 loss is deferred to the extent of the unrecognized gain in the offset. You'd get no loss in 2026, and it moves to 2027. If it doesn't apply, you have a $5,000 2026 loss and a $5,000 2027 gain, and that would be a tidy deferral on a position with no real risk. That is exactly what Section 1092 was written to stop. The bona fide loss rule at Treas. Reg. §1.165-1(b) points the same way.

Don't build a year-end trade on the assumption that Section 1092 is off the table.

Practical guidance

1. Close both legs in the same year if you can, especially around December.
2. Record pairs. Note which positions you opened as hedges, with dates.
3. Disclose your method to your preparer. Unrecognized gains at year end are the key figure if you're ever asked.
4. Keep year-end prices for open legs.
5. Prefer simple structures at tax time: a merge on the same venue closes the pair at once. See split, merge and convert.

Hedging an outside exposure

If you hold ETH and buy NO on a "will ETH go above $X" market as a hedge, the position may reduce risk on a property you own. That's another possible offset. If it's a straddle, losses on the contract could be deferred while you still hold the ETH, and it could affect the ETH's holding period. This is a further unsettled area. If you're hedging a big asset, get advice.

How this interacts with other rules

Fees and cross-venue records

Arbitrage margins are thin, so fees can be most of the result. Capture all of them, including bridge costs and gas. Records across venues rarely line up, because Kalshi reports in dollars, on its FIFO basis, while Polymarket's history is on-chain in UTC.

Run the free OddsFile check: paste a Polymarket or Hyperliquid address and see your 2026 P&L under each treatment, no signup. Comparing the gambling row with the capital row shows how much your gross volume costs you if you can't use net reporting.

FAQ

Is prediction market arbitrage taxable?

Yes. Profit is income, whichever venue paid it.

Can I net the two legs into one number?

Your net is the economic result. On Form 8949 each disposal is a row, and under gambling treatment, netting across venues isn't established.

Do the straddle rules apply to Kalshi?

Unclear. A contract traded on a CFTC-designated market has a stronger link to "actively traded" property, but no one has ruled on it.

Does a matched pair have a holding period issue?

Possibly. If a straddle exists, the rules can suspend holding periods. Most of these positions are short-term anyway.