Polymarket Split, Merge and Convert: How They Affect Your Taxes

No tax rule covers Polymarket splits, merges or neg-risk conversions, so you have to apply general principles and stay consistent. The most common approach treats a split as buying both outcomes for a total of $1 a pair, and a merge as selling both for $1. The hard part is how to divide cost between YES and NO. Getting that wrong moves gains between years, and many tools treat these events badly or skip them altogether.

OddsFile is software, not tax advice. The IRS has not issued guidance on how prediction-market contracts are taxed, and no guidance addresses these token operations. This article explains the options and the sources behind them; a tax professional can tell you which position fits your situation.

What each operation does

Per Polymarket's documentation:

Don't assume you never did one of these. Check your activity feed and look for split, merge or conversion entries.

Why a split is hard

You give up one token and receive two different ones. Under Treas. Reg. §1.1001-1(a), an exchange for materially different property can be a realization event, and the IRS says exchanging one virtual currency for another is one. If so, the pUSD leg is a disposal at about no gain. The two new tokens together cost $1.

The issue is dividing $1 between them.

Tax law often allocates cost by relative value when a single purchase brings two assets, so we'd start with market price. No authority applies that rule to outcome tokens.

Worked example

You split 1,000 pUSD into 1,000 YES and 1,000 NO when YES is at $0.70. You sell the 1,000 NO at $0.30, receiving $300, and hold YES until it resolves at $1.

The total matches, and it's right: you put in $1,000, took out $300 and got $1,000 back. But equal basis puts a $200 loss in the year of the sale and pushes a $500 gain to the year of resolution. If those are different years, equal basis looks like manufacturing a timing benefit. It also inflates gross figures, which matters under gambling treatment.

Merge

A merge is the reverse. You give up 1 YES and 1 NO, each with its own basis, and get 1 pUSD. Treat it as selling both for a combined $1.

Say you bought YES at $0.40 and later NO at $0.55. Your total cost is $0.95, and a merge returns $1.00, for a $0.05 gain per pair, whichever leg you assign it to. Allocate the $1 across the legs on the same method you use for splits, if your form needs one row each.

Under gambling treatment, that's one market closing with one net result, which is a simple case under a per-market session.

Neg-risk conversion

Here 1 NO on outcome A turns into 1 YES on each other outcome. It's a harder case. Two views:

• Exchange. You disposed of NO and acquired several YES shares. Gain or loss is value received minus the basis of NO. New basis is value at conversion.
• Continuation. Economically, a NO on A already equals holding every other outcome, so the position hasn't changed. Basis moves over, split by value.

The second view is closer to the economics, and it avoids reporting a gain you haven't really made. The first is more cautious under §1.1001-1. The difference is timing only, unless a holding period is at stake. A preparer may prefer either. Just don't mix them across trades.

Holding periods

If a split is an exchange, a new holding period starts. A merge is a disposal of each leg on that date. For a long-term position, a conversion could matter, which is one reason to prefer not to convert a position nearing a year unless you've thought it through.

What can go wrong in software

• A split read as a deposit of two free tokens, with zero basis, then shown as large gains on sale
• A merge read as a withdrawal, so both legs stay "open" forever
• Conversions dropped, leaving an orphan NO that looks unredeemed
• Double counting of pUSD when a mint is mistaken for a trade

Compare your total against a cash test: all money in, all money out, plus the value of open positions. If your reported net differs a lot, look for these events.

A consistent approach

1. List every split, merge and conversion.
2. Allocate by market price at the time, unless your preparer says otherwise.
3. Note the method in your tax file.
4. Apply it to every event in the year and in later years.
5. Reconcile to the cash test.

Run the free OddsFile check: paste a Polymarket address and see your 2026 P&L under each treatment, no signup. If the check shows a result that doesn't fit your own cash flow, compare it with these events. For how positions flow to the form, see Form 8949, and for the overview, see Polymarket taxes: the complete guide.

FAQ

Is a split taxable?

It's unclear. If treated as an exchange, the gain is about zero, since you exchange $1 of collateral for $1 of tokens. The question that matters is basis, not tax on the split.

Do I have to report every split on Form 8949?

Under capital treatment, you report the disposals that close positions. Whether the collateral leg of a split needs its own row depends on your position. See pUSD taxes for the same reasoning.

Which is better, market price or equal basis?

We'd start with market price. Equal basis produces paper gains and losses that don't exist.

Does this apply to Kalshi?

Kalshi works differently, and you don't hold transferable tokens. Trades and settlements appear in its own history.