GammaSwap V2 Architecture

GammaSwap V2 is launching a binary prediction market CLOB for crypto prices. For example, will ETH be up or down in 15 minutes? Or above $2,200 in an hour?

Incumbent prediction markets already exist so how will GammaSwap V2 differentiate itself? Let's dive in.

V2 utilizes an architecture similar to the trusted optimistic L2 rollup framework. The protocol matches orders offchain and handles settlement onchain. When a fill happens, the asset transfer is recorded in this ledger and later submitted to the blockchain to finalize, similar to how an L2 settles transactions on the L1.

There is no gas for transactions and the exchange is permissionless. Orders are filled based on time-price priority. Given V2 is a margin based exchange, a user can trade against themselves to close or flip an order's direction. This architecture also opens up cross margining opportunities for higher capital efficiency in the future.

Current latency expectations are around 100-200ms round trip, which if maintained in production, would make GammaSwap V2 the fastest onchain prediction market. The latency advantage is from the lightweight architecture.

V2 vs Polymarket

Polymarket uses the Conditional Token Framework (CTF), a decade old architecture developed by Gnosis for prediction markets where every order is an onchain token.

The original idea behind prediction markets was to coordinate market participants on unique events: who will win the presidential election in 2020 as an example? Will inflation increase to 4% next year? Etc. A positive externality would be: if people with different information can profit by being right, can market prices become good public forecasts?

The reality is that 75-90% of the volume on PMs is sport and short duration crypto markets, not what many onchain PMs were originally designed for.

At a high level CTF is elegant because it generalizes prediction-market positions into composable ERC-1155 conditional tokens, which helped support initial implementations. As an example, for the 2020 election, Catnip Exchange facilitated $11M on election day by combining two Augur outcome tokens: yTrump, nTrump and DAI into a Balancer pool only possible by using CTF structure. https://catnip1.netlify.app/

Polymarket uses similar CTF structure as Augur but with a different oracle structure, UMA. Polymarket originally started with AMM pools to create liquidity but due to high slippage and Impermanent Loss moved to a CLOB structure.

Legacy CTF structure when implemented on a CLOB creates some structural inefficiencies, especially for high-frequency markets which should operate like a derivative platform.

• Minting, splitting and merging are extra transactions that must happen each epoch, increasing latency. It is also operationally heavier for market makers unlike a margin exchange where positions offset instantly.
• Each "Yes" and "No" token is separately minted onchain with their own unique token id. Those two outcome token IDs require their own orderbook which fragments liquidity. Instead of one unified book, there are two separate instruments.
• After a market resolves, winning shares need to be redeemed to recycle capital. Since these are conditional tokens, each tokens has to be redeemed and then reminted onchain, often taking minutes. For market makers and traders with small bankrolls, this impedes them from rolling over their capital. In short expirations, redeeming must be immediate to maximize capital efficiency.

There have also been recurring issues like ghost fills, CLOB uptime, etc that impact the user experience significantly.

CTF is battle proven. It is composable. It is not as efficient on a CLOB as margin trading, especially with short duration markets where fractions of a second are significant.

V2 vs Kalshi

Kalshi has surpassed Polymarket consistently in notional volume since September 2025, largely due to their faster and more reliable infrastructure.

Kalshi operates a margin exchange built around fully collateralized binary contracts, with margin based on the maximum possible loss of a position or portfolio, similar to GammaSwap V2.

Unlike a typical derivative platform, users are not positing margin for leverage. Instead they post enough margin to cover the worst case loss which means no risk of Automatic Deleveraging (ADL) for users or bad debt to the protocol.

Since Kalshi is not using the CTF framework, there is less liquidity fragmentation and lower latency. The uptime is also significantly better than Polymarket, which is trying to marry an offchain order book with collateralized onchain tokens.

Now onto the differences between GammaSwap V2 and Kalshi.

Kalshi is a centralized CFTC-regulated Designated Contract Market, or DCM. They are permissioned, require KYC and do not settle onchain. Regulation can help build trust but it is even better if any third party can independently verify exchange operations, which is impossible when settlement is off chain.

Unlike Kalshi, GammaSwap V2 settles all transactions onchain which provides benefits like self-custody, faster onboarding and transparency for user deposits. Many users and institutions were bankrupted by FTX since there was no way to verify their funds were misappropriated, until it was too late.

V2 vs Hyperliquid HIP-4

It is more difficult to compare GammaSwap V2 to Hyperliquid HIP-4 as the exchange has only recently launched, however, we can still provide the cursory differences from their documentation.

Outcome contracts run directly inside HyperCore, the core execution layer responsible for perpetual swap and spot exchange markets. It is a collateralized token framework that shares many implementation details with spot trading, with different tokens representing different outcomes.

Although they use a conditional token framework, HIP-4 has an advantage by using the battle tested HyperCore to power their order book and matching engine, which is extremely performant supporting over 200,000 orders per second.

However, using tokens does add more operational complexity for market makers. They have to mint, burn and merge tokens every market epoch which makes inventory management difficult. Also, each outcome token has its own order book fragmenting liquidity.

For a margin exchange like GammaSwap V2, opposing positions automatically net and settle each epoch cleanly.

Hyperliquid supports a large mandate of potential market categories like politics and sports with many potential outcomes. A token framework may make sense here since merging could be useful for market makers across outcomes. Also, the tokens can be reused by DeFi builders on HyperEVM for lending or looping use cases to add leverage.

Since GammaSwap V2 is focused on crypto binary markets, we can build the architecture to be maximally performant with minimal complexity for takers and makers.

Closing

To sum it up, Polymarket and Hyperliquid share the conditional token framework which allow positions to be more composable in DeFi, however, for short crypto binary markets it does add latency and operational complexity. HIP-4 leverages the already operationally sound HyperCore functionality which gives them an advantage.

Kalshi has the most similar architecture to GammaSwap V2. Both operate as a margin exchange but GammaSwap settles markets directly onchain, while Kalshi is centralized.

In the next article, we will be discussing capital efficiency of binary trading vs other derivative platforms and announcing our trading competition(s).