What to use instead of Trepa

The product

Trepa asked a different question from everything around it. Instead of yes or no, you named a number: the exact BTC price at the end of the round. Your payout was not a fixed price times a share, it was a share of the pool, and it scaled with how close you landed.

The mechanics, from the company's own reintroduction post of 27 July 2026: each round ran one minute — thirty seconds to place a prediction, thirty seconds to watch the live price come toward your target or away from it. Payout was a convex accuracy score taken relative to everybody else in the pool, distributed pro rata, and the post states you could never lose more than half your stake. It ran on Solana, funded with USDC, in a separate app at its own domain — one communal hour a day, 1 to 2pm UTC, because a pari-mutuel pool is only worth entering when a crowd is in it at the same moment. The plan named in that post was more assets: SOL, HYPE, equities, commodities, FX.

The argument behind it was about information, not novelty. The same post named the thing it was competing with — the 5-minute BTC up-or-down market, where a continuous outcome gets sliced into a yes and a no — and called the loss discretization. If you think the price ends at 64,180, a yes-or-no contract cannot hold that sentence.

How it ended. On 10 August 2026 the two founders published a shutdown post on the company's own blog, and it is unusually specific. The last playable round was 12 August 2026, with no rounds after it. Referral and streak rewards were to be sent to Trepa wallets on 13 August 2026. The app stays online until 30 September 2026 so balances can be withdrawn, and after that "it will not be reachable". Anyone who misses that date is pointed at Privy's own account portal, which works only for accounts opened with an email address.

The post is also candid about why, which is rarer. They could not find product-market fit. The pari-mutuel design meant you could not know your payout when you committed and could not exit a position once entered, and they had accepted both knowingly. Choosing a number to decimal precision asked more of a person than choosing a side. The one-hour window concentrated the crowd the mechanism needed, but required users to build their day around it, and retention fell under that ask — while every plan to scale, more assets and more windows, diluted the very concurrency the mechanism ran on. They also say geographic restrictions removed a large share of the addressable audience before they started. Which restrictions, and where, they do not say, and we could not establish it from anything they published.

What to use instead

Nothing replaces Trepa whole, and the reason is structural rather than commercial: no venue in this catalogue pays you for proximity to a number. They all pay you for being on the right side of one. Split what you were there for.

If you were there for a fast market on the BTC price, Limitless is the closest. Its listing is dominated by short-cadence crypto — BTC and ETH up-or-down over 5 minutes, 15 minutes and an hour — collateralised in USDC on Base, traded on an order book at 1.5 percent taker and zero maker, resolved automatically against a Chainlink TWAP with no identity check. Polymarket runs the 5-minute market Trepa named, on a far deeper book, at 1.75 percent on crypto. Both are yes-or-no, which is exactly the loss Trepa existed to argue about — but both are open now, at any hour, with a price on the screen.

If what you actually valued was the hour and the contest, that part is gone and the nearest substitute is a venue that always has a quote. Myriad prices with an AMM across four chains at 0.5 percent taker and settles in stablecoins, so there is no pool to fill and nobody to wait for. You lose the thing that made the 1pm hour fun and you gain the ability to arrive at 3am.

If you were there to be measured, stop looking at venues. Metaculus is built on the log score and publishes Baseline and Peer scores against a public track record; nobody takes a position and nothing pays out, which is the point. Manifold has a numeric market type among binary, multiple choice, polls and bounties, runs on play money and MIT-licensed code, and lets you write the question at the precision you want rather than waiting for someone to list it.

If you are in the United States, Limitless, Polymarket and Myriad all bar you by their own terms. Kalshi lists crypto contracts alongside politics, macro and sports, settles in dollars on a CFTC-designated exchange at 1.75 percent taker, and requires identity verification — a different product legally and practically, and the available one.

What you give up

Being paid for precision. This is the whole loss and it is worth stating plainly. On every replacement here, a forecast of 64,180 and a forecast of 67,000 are the same trade if the market is "above 64,000" and both are right. Trepa was the only place where the difference between them was money. Metaculus keeps the distinction and removes the money; Limitless keeps the money and removes the distinction. Nothing in this catalogue keeps both.

A bounded downside. Trepa's own post says you could never lose more than half your stake. On an order-book venue a contract you bought at 62 cents and that resolves the wrong way is worth zero, and on an AMM the same is true — the full stake is the downside, every round. A reader treating a Limitless position as "like a Trepa round" is sizing it wrong by a factor of two.

A game you could finish. A round lasted sixty seconds and ended. A position in a 15-minute up-or-down market also ends, but a position in anything longer is a thing you now carry, watch and think about, and the catalogue's venues are mostly built for carrying. Some people liked Trepa precisely because there was nothing to hold.

The crowd being there at the same time. A thousand people predicting in the same minute is what made a pari-mutuel pool pay. It is also what killed the product, by the founders' own account. What replaces it is liquidity that does not care whether you showed up — which is more useful and much less of an occasion.

Migration notes

Withdraw before 30 September 2026, and do it from inside the app. The instruction in Trepa's own post is three steps: open the app, go to your wallet, send the balance to an address you control. There is no sweep, no automatic return and no stated grace period.

Your Trepa wallet was an embedded wallet, not one you made. That is why the fallback in the shutdown post is Privy's account portal rather than a seed phrase, and why it is conditional on having signed up with an email address. If you used a different sign-in method, the app itself is the only route and the date is hard.

The balance includes the rewards paid on 13 August. Referral and streak rewards landed in the Trepa wallet, not in yours. Somebody who stopped playing after the final round on 12 August and never reopened the app has a balance sitting there that they have no reason to expect.

USDC on Solana is not USDC on Base or Polygon. Every money pick here lives somewhere else: Limitless on Base, Polymarket and its clients on Polygon, Myriad across BNB Chain, Abstract, Linea and Celo, Kalshi in dollars in a bank. Moving the balance out of Trepa is one transaction; moving it to the chain your next venue uses is a bridge, a fee and a second chance to send money to an address that cannot return it. Withdraw first to a wallet you control, then bridge deliberately.

Do not carry the position size across. Half-stake downside was a property of one mechanism and it does not travel. Whatever you were putting into a round, size the first ten trades somewhere else as if the whole amount can go to zero, because it can.