BTCBTC$64,938+1.03%|
ETHETH$1,821.86+1.06%|
USDTUSDT$0.99927+0.00%|
BNBBNB$592.0200+0.82%|
USDCUSDC$0.99961+0.00%|
XRPXRP$0.97396+2.58%|
SOLSOL$73.7700+3.89%|
TRXTRX$0.30634+0.79%|
FIGR_HELOCFIGR_HELOC$1.0390+0.00%|
HYPEHYPE$69.4000+6.13%|
ZECZEC$948.2400+9.96%|
XMRXMR$511.3600+2.06%|
DOGEDOGE$0.05940+1.76%|
LEOLEO$9.3200+1.25%|
USDSUSDS$0.99987-0.01%|
WBTWBT$62.7300+0.79%|
LINKLINK$9.3200+2.47%|
ADAADA$0.15438+2.06%|
XLMXLM$0.14052+2.85%|
USDEUSDE$0.99938+0.02%|
DAIDAI$0.99988-0.01%|
BCHBCH$212.6000+3.57%|
NEARNEAR$3.2400+3.21%|
RAINRAIN$0.00551532-3.74%|
LTCLTC$45.0700+1.91%|
CCCC$0.08258-1.27%|
UNIUNI$5.3900+6.61%|
AVAXAVAX$7.2400+2.03%|
XAUTXAUT$3,836.17-0.44%|
SUISUI$0.73536+3.05%|
BTCBTC$64,938+1.03%|
ETHETH$1,821.86+1.06%|
USDTUSDT$0.99927+0.00%|
BNBBNB$592.0200+0.82%|
USDCUSDC$0.99961+0.00%|
XRPXRP$0.97396+2.58%|
SOLSOL$73.7700+3.89%|
TRXTRX$0.30634+0.79%|
FIGR_HELOCFIGR_HELOC$1.0390+0.00%|
HYPEHYPE$69.4000+6.13%|
ZECZEC$948.2400+9.96%|
XMRXMR$511.3600+2.06%|
DOGEDOGE$0.05940+1.76%|
LEOLEO$9.3200+1.25%|
USDSUSDS$0.99987-0.01%|
WBTWBT$62.7300+0.79%|
LINKLINK$9.3200+2.47%|
ADAADA$0.15438+2.06%|
XLMXLM$0.14052+2.85%|
USDEUSDE$0.99938+0.02%|
DAIDAI$0.99988-0.01%|
BCHBCH$212.6000+3.57%|
NEARNEAR$3.2400+3.21%|
RAINRAIN$0.00551532-3.74%|
LTCLTC$45.0700+1.91%|
CCCC$0.08258-1.27%|
UNIUNI$5.3900+6.61%|
AVAXAVAX$7.2400+2.03%|
XAUTXAUT$3,836.17-0.44%|
SUISUI$0.73536+3.05%|
Drift BET
Prediction MarketTier-2Discontinued on-chain event marketsVerified

Drift BET

Drift's capital-efficient prediction market — closed since the 1 April 2026 exploit and left out of the Velocity DEX rebuild that replaced it.

Visit website

Key Facts

TypeDiscontinued on-chain event markets
Chain / RailSolana — Drift Protocol; the successor venue reuses the same Solana program lineage under a new name
CollateralAround 30 Solana tokens accepted as cross-collateral, including SOL, USDC and mSOL; the successor settles in USDT
JurisdictionGlobal, excluding the US and other restricted territories under Drift's interface terms
Data ConfidenceVerified

Scale Signal

About $3.5M of order book liquidity within 24 hours of its August 2024 alpha launch; the hosting protocol was drained of $295.7M on 1 April 2026 and BET has not restarted since.

Overview

Drift BET was the most capital-efficient prediction market ever built on Solana, and it no longer exists. It launched on 19 August 2024 as "Bullish on Everything", a set of event markets that lived inside Drift's perpetuals exchange rather than beside it, and it stopped serving on 1 April 2026 when the protocol was drained of $295.7M. Drift's interface and documentation now redirect to a successor venue, Velocity DEX, which relaunched on 1 July 2026 as a USDT-settled perpetuals exchange — and whose code, per its own migration documentation, removes the instruction that created prediction markets and renames the contract type to DEPRECATED_PREDICTION. This profile is therefore a record of a closed venue rather than a live one: useful if you held positions or built on the SDK, and useful as a case study in how an event market can fail for reasons that have nothing to do with its market design.

What BET actually was

BET was never a separate platform. A BET market was a Drift perpetual market carrying the contract type Prediction, with its price bounded between 0 and 1 and settlement forced to one end or the other. It ran on the same program, the same order book, the same matching engine and the same margin system as every other Drift market, and Drift's documentation said so plainly: prediction markets had the same fee structure as perp markets. That reuse was the product. The launch pitch, and the reason it got attention, was capital efficiency — roughly thirty Solana tokens accepted as collateral rather than USDC alone, lending interest accruing on that collateral while a position stayed open, and a single margin account in which an outcome position and a Bitcoin short could offset each other.

How a market worked

Each market was a bounded perpetual: buy the outcome you expect, sell it when you like, and the price stays between 0 and 1 because the contract settles at one boundary or the other. Because it was a perpetual, it inherited everything Drift's perpetuals had — limit, stop and reduce-only orders, the keeper network that filled and triggered them, cross-collateral in a single account. Fees came from Drift's perpetual tiers: the entry tier taker fee on the schedule published in June 2026 was 0.0350%, with maker rebates above it, and markets were denominated in whatever collateral the trader posted. On an event market that structure is attractive, because a position can be entered and exited like any other perp instead of being locked until resolution.

Who controlled the markets

Nothing about BET was permissionless. Drift's security council — an elected multisig under Realms governance — initialised every market on request, and the same council resolved each one by setting the oracle to a fixed 0 or 1 and an expiry date, after which the market went reduce-only. Permissionless market creation was promised in the launch announcement for "later this year", trailed again for early 2026, and never shipped: in the final published version of the code, initialize_prediction_market was still an admin instruction. For a product in a category where the open question is who decides what happened, BET's answer was a small elected group, resolving by hand.

The flaw in the settlement design

Drift's documentation stated that where a market ran a shortfall its fee pool and insurance could not cover, the shortfall would be socialised into the settlement price. A market resolved to 1 could therefore settle below 1. That is a risk a fully collateralised Yes/No share does not carry, and it separates BET from almost every venue that now competes for the same users: on a fully collateralised contract, the worst case is that the position you hold is worth nothing, not that a winning position pays out short.

What ended it

On 1 April 2026, Drift was drained in roughly twelve minutes across 31 transactions. The protocol's own recovery update put the stolen assets at $295,706,374.93 and timestamps the snapshot of affected positions at 18:31:47 UTC that day; on-chain analysts and TRM Labs tied the operation to North Korean actors, with durable nonce accounts set up in the preceding days and a social-engineering campaign running for months before that. It was the largest DeFi exploit of 2026 and the second-largest in Solana's history. The DRIFT token fell more than 40% on the news and changes hands near two cents today, against roughly $0.42 when BET launched. Deposits and withdrawals were paused, the app stayed offline, and every user position — including anything open on a BET market — has been frozen behind the protocol's recovery process since that afternoon.

The recovery package

Drift's plan was not a buyback. It describes a transferable SPL recovery token worth one dollar per dollar of verified loss, funded out of exchange revenue rather than treasury cash, with $127.5M committed by Tether and $20M by other partners, and two independent audits required before any relaunch. That is a revenue-funded promise paid over time, not an immediate reimbursement — which matters if you are valuing a frozen BET position today, because the recovery token prices your loss rather than settling it.

What replaced it

The protocol relaunched on 1 July 2026 as Velocity DEX, with USDT replacing USDC as the quote asset and a new team, a new protocol lead and former risk specialists in place. Velocity is a perpetuals exchange and a money market in one Solana program: spot trading has been removed from the program entirely and exists only for collateral and lending, and the documentation states plainly that trading is perpetuals only. Fees run 4 bps of filled notional at the base tier, 3 bps above $5M and 2 bps above $80M of trailing 30-day volume, with the maker paid 0.25 bps and individual markets allowed to add up to 10 bps on top; a beta-testers promotion advertised 2 bps fees for a limited period. The beta opened to everyone on 21 September 2026, and on 30 September 2026 Drift co-founder Cindy Leow stepped back from the rebuilt exchange. Prediction markets are not part of it, and the migration documentation records why: the instruction that created BET markets is gone and the contract type is now marked deprecated.

What to check before you chase it

  • There is nothing to trade. BET stopped serving on 1 April 2026 and has not restarted; the successor venue offers perpetuals, not events.
  • Frozen positions are a claim, not a balance. Recovery runs through a revenue-funded token worth $1 per verified loss, paid over time.
  • Legacy links mislead. drift.trade still serves pre-hack marketing copy with dead yield figures and hands off to an app that now redirects to the successor, where you cannot reopen a BET market.
  • Don't port the SDK expecting parity. The fork is a new program deployment with no on-chain state carried over and every program-derived address changed.
  • The lesson travels. BET's failure had little to do with prediction-market design — it was parent-protocol risk, and any event market built inside a leveraged exchange inherits it.

Regulatory Status

No longer operating. Drift's interface refused US persons and other restricted territories, and BET was never licensed or registered anywhere; its markets were created and resolved by Drift's elected security council rather than by a regulated exchange or an independent oracle. The protocol was drained on 1 April 2026, stayed offline, and relaunched on 1 July 2026 as a perpetuals exchange with the prediction-market code marked deprecated.

Features & Market Details

One program, three products

BET's design is worth understanding because nothing in the market now replicates it. Drift was a perpetuals exchange with a dynamic AMM, a decentralized order book and just-in-time liquidity sitting on one margin engine, and BET was a market type inside that engine rather than a product built on top of it. Concretely, that meant a BET position was a perpetual: the outcome token's price was bounded between 0 and 1, settlement pinned it to one boundary, and everything else about it — the order types, the keeper network, the collateral rules, the margin maths — was inherited rather than rebuilt. A separate prediction market has to build all of it; BET got it by being the same thing with a bounded payoff.

The capital-efficiency pitch, in detail

Three claims made BET distinctive. First, collateral: roughly thirty Solana tokens were accepted, so a holder of SOL, an LST or a yield-bearing stable could post what they already owned instead of selling into USDC and taking the spread. Second, yield: because those assets sat in Drift's lending markets as margin, interest accrued on collateral while an event position was open — the position was not idle capital in the way a fully collateralised share on another venue is. Third, netting: one account could hold the outcome position and a hedge on another market, with the two offsetting in the same collateral pool. No venue in the category offers that combination today, and it is the reason BET's name kept appearing in reviews years after it stopped trading.

Fees, orders and settlement mechanics

Because a BET market was a perpetual, it priced and charged like one. The fee schedule published in June 2026 — before the shutdown was fully reflected in third-party write-ups — quoted an entry-tier taker fee of 0.0350% of notional, with maker rebates for resting liquidity and tiers that improved with volume. That is a fundamentally different shape from the venues BET is compared against: Polymarket and Limitless charge a coefficient on share count that peaks at the 50-cent midpoint, Myriad and Futuur settle percentage fees, and BET charged basis points of notional regardless of where the price sat. It also inherited a richer order surface than a pure prediction book: limit, stop and reduce-only orders, keeper-filled triggers, and the ability to hold a position indefinitely rather than watch it expire.

Who could create a market, and who resolved it

Drift's security council initialised every prediction market on request, and the same council resolved every market: it set the oracle to a fixed 0 or 1 and an expiry, after which the market went reduce-only. There was no external oracle vote, no staked dispute window and no permissionless path — the launch post promised permissionless creation for later in 2024, the roadmap trailed it again for early 2026, and the final published code still gated market creation behind an admin instruction. That trade-off is the mirror image of how later venues approached it: Hyperliquid's version moved the referee into a validator vote and a bonded deployer, while BET's stayed a multisig. Neither is automatically safer, but the difference tells you where to look when a resolution is contested.

The socialised-shortfall clause

The single most important line in BET's documentation was the one about shortfalls. Where a market's losses exceeded what its fee pool and insurance could cover, Drift said the shortfall would be socialised into the settlement price, meaning a market that resolved Yes could settle at less than $1. That is the opposite of the guarantee a fully collateralised contract gives, and it existed because BET was a leverage engine underneath: an event market sharing a margin system with 20x and 101x perpetuals shares that system's solvency characteristics. Any review of a prediction market should ask whether a winning position is guaranteed its full payout, and in BET's case the honest answer was no.

The exploit

On 1 April 2026 an attacker drained roughly $285M from Drift in about twelve minutes across 31 transactions, with the protocol's own recovery update valuing the loss at $295,706,374.93 and timestamping the position snapshot at 18:31:47 UTC. Investigators described a structured operation: durable nonce accounts created in the days beforehand so pre-signed transactions could be executed later without expiring, a social-engineering campaign against contributors running for months, and privileged access used to take control of the protocol's admin path. Most of the stolen assets were bridged to Ethereum within hours. Attributed to North Korean actors, it was the largest DeFi hack of 2026 and the second-largest in Solana's history behind Wormhole. The DRIFT token fell more than 40% immediately and trades near two cents today.

The recovery token, and what it means

Rather than a cash reimbursement, Drift described a transferable SPL recovery token worth one dollar per dollar of verified loss, funded from exchange revenue, with $127.5M committed by Tether and $20M by other partners and two independent audits required before relaunch. Read that carefully if you held a position: the claim is denominated against future revenue and paid over time, it is transferable rather than settled, and its value depends on the venue that replaced Drift actually earning fees. For BET users it also means the distinction between "my position was going to lose anyway" and "my position was destroyed" was settled in dollars of loss, not by whether the event went your way.

The velocity rebuild: what came back and what did not

Drift's interface and documentation now redirect to Velocity DEX, which relaunched on 1 July 2026 with USDT replacing USDC as the quote asset and a rebuilt program — a fork of Drift v2 at SDK v2.163.0-beta.0, deployed fresh with no on-chain state carried over and every program-derived address different. Velocity is a perpetual futures exchange and a money market in a single Solana program; spot trading has been removed from the program and now exists only for collateral and lending, with swaps as the only way to exchange spot assets. Fees are 4 bps of filled notional at the base tier, 3 bps above $5M and 2 bps above $80M of trailing 30-day volume, with makers paid 0.25 bps and individual markets permitted to add up to 10 bps on top; the beta period advertised 2 bps fees as a tester perk. OtterSec audited the rebuilt program after the fork, reporting 154 findings with no criticals and all highs fixed in the deployed code, while the source itself remains closed until the post-fork report is final. The beta opened to everyone on 21 September 2026, and on 30 September 2026 co-founder Cindy Leow handed the venue to a new team.

The evidence that prediction markets are gone

The successor's documentation runs to dozens of pages and mentions prediction markets exactly once: on the developer migration page, which records that initialize_prediction_market has been removed and that ContractType.PREDICTION is now DEPRECATED_PREDICTION. The protocol repository was renamed and archived read-only in July 2026 under Apache-2.0, so the BET code is readable and unrunnable at the same time. The old docs URL for prediction markets redirects to a page that errors, and the original launch announcement now returns 404. Nothing was ever announced — BET simply stopped existing as a product, which is why third-party reviews still confidently describe its fee tiers as if you could trade them.

What a successor would have to offer to replace it

  • Margin-integrated event risk: no current venue lets an event position share collateral and offset with a perpetual exposure in one account.
  • Multi-token collateral: Polymarket and Limitless take USDC, Myriad takes USDC, most venues take one stablecoin; posting SOL or an LST directly is not an option anywhere.
  • Perpetual-style order handling: stops, reduce-only and keeper-filled triggers came free because BET was a perp; on a prediction order book you build them yourself.
  • Yield while you wait: collateral earning lending interest while backing an event position has no equivalent on a fully collateralised book.

Limitations and risks

  • The venue is closed. BET stopped serving on 1 April 2026 and the successor offers perpetuals only.
  • Open positions and balances have been frozen since the exploit, recoverable only through the protocol's revenue-funded recovery token.
  • A winning position on a BET market was never guaranteed a full payout: shortfalls could be socialised into the settlement price.
  • Market creation and resolution were both controlled by Drift's elected security council, with no external oracle or dispute window.
  • Permissionless market creation was promised twice and never delivered; the admin instruction was still in the last published code.
  • The collapse came from parent-protocol security, not from market design — an event market built inside a leveraged exchange inherits that exchange's failure modes.
  • Drift's own terms excluded US persons and other restricted territories, and no venue in this category is licensed or registered anywhere.
  • Legacy pages for BET still advertise working fee tiers and liquidity, so any review older than April 2026 should be treated as describing a product that no longer trades.

Bitnxt uses cookies to run the site, understand how it is used, and measure ads. You can accept or decline below, and change your choice any time in our Privacy Policy.