Polymarket vs Manifold: cash risk and play-money practice

The one question that separates these products

Ask whether the unit you hold can leave the platform. On Polymarket, ordinary binary winning shares settle at 1 USD equivalent and losing shares at 0, and the price is commonly read as an implied probability, subject to mechanics and costs; no forecast guarantees gains. On Manifold, mana is a virtual currency that cannot be redeemed for cash, so a 100 mana gain is not 100 USD. That single difference changes what a track record means, not just how it looks.

This is a comparisons guide, so the point is not to rank the 2 sites. It is to name the trade: cash exposure buys transferable settlement and the possibility of real loss; play money removes cash risk but also removes any claim on cash. Everything below follows from that split.

Resolution decides what you actually held

A prediction market contract has outcomes and specified resolution criteria, and the title alone is insufficient to know how a market will settle. On Polymarket, ordinary binary markets resolve through UMA, while Up/Down markets use Chainlink TWAP (time-weighted average price) over the same asset stream at start and end: final at or above start means UP, lower means DOWN. Different exchange candles can differ, so the actual market rules must be verified rather than generalized across every duration or source. Rare unknown or 50-50 resolution can pay 0.50 per share, not universal cancellation or refund.

Manifold questions are community-created and creator-resolved, which is why checking creator resolution practices matters rather than assuming a uniform process. The practical consequence on either site is that the rules text is part of the position. If you cannot state the source, deadline and edge cases in your own words, you do not yet know what you bought.

The 100 mana trap: a worked comparison

Imagine 2 hypothetical records. The Manifold record shows a gain of 100 mana, which remains a virtual unit. For a separate illustrative Polymarket position, buying 200 shares at 0.50 costs 100 USD; an ordinary winning settlement pays 200 USD, giving 100 USD profit before transaction costs. A losing settlement pays 0, losing the 100 USD purchase cost before other costs. A 100-mana gain and a 100-USD gain are different units; the record alone does not make them interchangeable.

The illustrative case below makes the accounting explicit by stating which entry prices are known and which are unspecified. Inputs are invented for exposition and are not observed platform results. The purpose is to show where the comparison breaks, not to estimate anyone's returns.

Read the 2 columns as different arithmetic, not a conversion. On Polymarket, gross result equals shares times (1 minus entry price): 40 shares bought at 0.50 USD pay 40 times (1 minus 0.50) = 20 USD before costs. On Manifold, a mana gain is simply the change in your mana balance. There is no exchange rate between the 2 quantities.

Question creators and participant incentives differ

On Manifold, anyone in the community can create a prediction question, and the creator's resolution practice is part of the market's quality. On Polymarket, ordinary binary markets route through UMA and Up/Down markets through Chainlink TWAP, with rules specifying the source, deadline and edge cases. These are different governance and incentive structures. A play-money participant who risks nothing may still trade carefully, but the cost of error is attention and reputation rather than cash.

This also affects how you should read a crowd price. On one side a price can be interpreted as an implied probability subject to mechanics and costs; on the other, the same-looking price is produced by participants using a nonredeemable currency. Similar charts do not imply similar information content.

Where each product fits a learning routine

If your goal is to practise writing forecasts before risking cash, Manifold offers a place to record them without redemption exposure, provided you treat the mana score as a process log rather than a profit figure. If your goal is to test whether you can hold a position through cash loss, only a cash-settled market does that, and only after you have read the resolution rules.

Do not transfer a play-money record into an expectation of real-money performance, because the currencies are not the same, the participants are not the same, and resolution authority sits with different parties. For the next comparison step, see the related guides on alternatives to a single prediction market platform, calibration of your own forecasts, and the risks that come with cash settlement.