How to Short on Prediction Markets: Buy NO, Sell YES, and Perps (2026)
There is no “short” button on Polymarket or Kalshi. Every negative position is constructed from long positions — you buy NO, you sell YES you already own, or you use perpetual futures. Each route has different capital requirements, different costs, and different failure modes. Here is exactly how each works.
Route 1: Buying NO shares
The simplest short. If a market trades at YES 63¢, buying NO costs 37¢ per share. You profit when the event doesn’t happen:
| Outcome | Your NO position (100 shares @ 37¢) |
|---|---|
| Event resolves NO | +$63.00 (shares pay $1.00, cost $37.00) |
| Event resolves YES | −$37.00 |
When it’s the right route: you have no position, the NO side is underpriced relative to your estimate, and you’re comfortable with the capital being locked until resolution.
Watch: the YES + NO prices must sum to $1.00 minus the spread. On thin markets, buying NO at the ask can imply you’re paying more than the true NO probability — check the mid, not just the ask. The order book mechanics page covers why the visible quote lies on thin books.
Route 2: Selling YES you already own
If you bought YES earlier at 40¢ and it now trades at 63¢, selling realizes the profit — and if your view turned negative, selling IS the short. This is capital-neutral going forward (you’ve exited), but it only works if you already hold shares.
The advanced version: selling YES you don’t own requires borrowing or minting. On Polymarket, a complete set (YES + NO) always mints for $1.00. So you can mint a set for $1.00, sell the YES at 63¢, and hold NO bought effectively at 37¢ — economically identical to Route 1, useful when the NO book is thin but the YES book is deep. You sell into the deeper side.
Route 3: Perpetual futures (Polymarket Perps)
Since April 2026, Polymarket Perps give true short exposure with leverage: open a short position against the perp price with up to 20× leverage on supported markets. Unlike share-based shorts:
- Capital efficiency: margin, not full notional — see the leverage guide
- No resolution lock: close any time; you’re not waiting for the event
- Funding costs: shorts pay or receive funding hourly depending on the perp’s premium — a real cost that compounds; see funding rate mechanics
- Liquidation risk: adverse moves can liquidate the position at marked levels — a risk share-based shorts never have
| Buy NO | Sell owned YES | Mint + sell YES | Perps short | |
|---|---|---|---|---|
| Capital required | Full cost of NO | None (already held) | $1.00 per set | Margin only |
| Locked until resolution | Yes | — | Yes | No |
| Leverage | No | No | No | Up to 20× |
| Funding cost | No | No | No | Hourly |
| Liquidation risk | No | No | No | Yes |
Which route should you use?
- Long-horizon conviction short, no leverage needed: buy NO. Simple, liquidation-free, capital locked but predictable.
- Flipping an existing long: just sell. Don’t complicate it.
- NO book thin, YES book deep: mint the set, sell YES, hold the NO leg.
- Tactical short with leverage or a defined exit window: perps — and respect funding and liquidation, which is where most perp shorts die, not the directional call.
FAQ
Can you short on Polymarket? Yes, three ways: buy NO shares, sell YES shares (yours or minted), or short perpetual futures on supported markets.
Is buying NO the same as shorting? Economically yes for binary outcomes — NO pays $1.00 when the event fails. The difference from a true short is capital structure: you pay the full NO price upfront rather than posting margin.
What does shorting cost on Polymarket? Share-based shorts cost the spread and (on mint-and-sell) the $1.00 per set tie-up. Perps shorts cost funding hourly plus fees; funding on a 20× position is a real line item — check the current rate before opening.
Do you need margin to short via NO shares? No. Buying NO is fully collateralized by the purchase price. Margin only enters the picture with perpetual futures.
Educational content, not financial advice. Not affiliated with Polymarket or Kalshi. Last updated October 2026.