DEX Types Explained: Who You Trade Against and Where Fees Go

Every swap on a decentralized exchange (DEX) has someone or something on the other side. Depending on the DEX type, that counterparty is another trader, a professional market maker, a pool of tokens run by a smart contract, or a pool whose rewards are set by a weekly vote. The fee shown on screen is rarely the whole cost.

That matters to anyone who has ever tapped “swap” in a wallet. A Bullrunners video walks through four main types of DEX, who you are actually trading against on each, and where the fees and risks go. Knowing who sits across the table explains why some trades cost more than they should and why some exchanges pay you in tokens. It also explains why the cheapest-looking option isn’t always the cheapest.

How can a stablecoin swap lose $215,000?

A stablecoin swap can lose most of its value when a bot manipulates the pool around it. According to Cointelegraph, cited in the video, a trader swapped about $220,000 of USDC on March 12, 2025 and got back just over $5,000 of USDT in a Uniswap pool. That is a loss of roughly $215,000. A bot drained the pool’s liquidity in front of the trade and put it back straight after. The whole thing took about 8 seconds.

Uniswap did not break. The pool simply followed its formula. The video adds that one well-known DeFi figure suggested the trade may not have been an accident and could have been a way of moving dirty money. Either way, the episode shows the core risk: a pool doesn’t care who you are.

What is an order book exchange?

An order book exchange matches buyers and sellers who post the prices they want. Most stock markets and centralized crypto exchanges use this model. The video compares it to a notice board: buyers pin up what they’ll pay, sellers pin up what they want, and a trade happens when two notes match.

Your counterparty is another person or a professional firm called a market maker, whose job is to keep orders on the board. Market makers earn the small gap between the buying and selling price, called the spread. That spread is the order book’s hidden cost.

CoinGecko, quoted in the video, lists the trade-offs. Limit orders let you set an exact price and reduce slippage, even on large orders. But order books need many active traders and can perform poorly in thin markets. Because orders are visible to everyone, they are also open to manipulation such as spoofing: placing large orders with no intention of filling them to fake interest.

How does an AMM like Uniswap work?

On an automated market maker (AMM) such as Uniswap, you trade with a pot of two tokens held in a smart contract, not with a person. A formula sets the price based on the balance between the two tokens.

The video gives an example. Imagine a pool holding 10 ether and a matching amount of dollars. Buy one ether and the pool has less ether and more dollars, so the next ether costs a little more. Buy a big chunk in one go and the price climbs with every bit you take. That climb is slippage, and on a small pool it can be painful.

The money in the pool comes from liquidity providers. They deposit both tokens and earn a slice of every swap fee, and in effect they are your opponent. When prices move sharply, traders keep taking whichever token has become the better deal. That leaves providers worse off than if they had just held the tokens. This is called impermanent loss. It only becomes permanent when they withdraw, but it is real, and it is the price they pay for the fees.

There is also a third player: bots. A swap sits briefly in a public waiting area before it is confirmed, and bots can see it. Some jump in front of your trade and push the price up. Your trade then fills at the worse price, and the bot sells immediately. This is a sandwich attack, because your trade is the filling. The profit bots squeeze out by reordering or jumping ahead of pending transactions is known as MEV, short for maximal extractable value.

Two protections help. Most wallets have a slippage tolerance setting, which is the maximum price move you will accept. Set it tight and a trade that would be badly sandwiched simply fails. Some wallets also send trades privately so bots can’t see them waiting. The video suggests checking whether yours does.

How does Hyperliquid run an order book on-chain?

Hyperliquid built its own blockchain to run a full order book on-chain, answering years of argument that order books were too busy for a blockchain. On Hyperliquid you trade against other traders and professional market makers. You also trade against a community vault that anyone can deposit into, which helps keep orders on the board.

Every order, trade and liquidation is public. On a centralized exchange, by contrast, you only see what the company shows you. In a clip shown in the video, founder Jeff Yan describes spot trading as the closest thing to Satoshi’s peer-to-peer vision. In his view, two people are transferring assets on a ledger, and the order book is just a mechanism to intermediate those transfers.

Much of Hyperliquid’s activity is in perpetual futures, contracts that let people bet on prices rising or falling with borrowed money. When prices move fast, borrowed positions are closed automatically and someone has to take them over. On Hyperliquid that job partly falls to the community vault, so depositors share both profits and losses.

What does a DEX aggregator like Jupiter do?

A DEX aggregator like Jupiter doesn’t hold your trade. It searches many exchanges at once, finds the best price and can split your order across several pools in a single transaction. Solana has many exchanges, including Raydium and Orca, and a large share of swaps there go through Jupiter. Raydium launched in February 2021 as a pool exchange linked to an order book, a link it fully removed in July 2026.

A host on the Lightspeed podcast said in December 2023 that Jupiter facilitated over 80% of trades on Solana. Jupiter’s co-founder, who goes by Meow, said the figure referred to retail trading. By September 2025, Jupiter handled about 90% of aggregator activity on Solana, according to investment firm 21Shares. The video says that in October 2025, Jupiter described a system that compares its own routes against outside ones and picks the better price, an aggregator of aggregators.

The video compares an aggregator to a travel comparison site: it finds the flight but doesn’t fly the plane. That is convenient, but you are trusting the routing and still facing the pool and order book risks underneath.

How do vote-based DEXs like Aerodrome work?

Vote-based DEXs direct rewards to pools through token-holder voting. According to the video, the idea grew out of Curve, a stablecoin exchange on Ethereum. Around 2020, Curve began letting people lock tokens for years in exchange for voting power over where rewards went. Projects competed so hard for votes that people called it the Curve Wars. A later design known as ve(3,3) pushed the idea further.

One of the biggest examples is Aerodrome, which the team behind Velodrome launched on Coinbase’s Base network in August 2023. Liquidity providers who stake their deposits don’t keep trading fees. Instead, they earn new AERO tokens. The fees go to people who lock AERO, sometimes for years, and vote each week on which pools get the new tokens. Other projects can pay voters to send rewards to their pool, and these payments are called incentives or bribes. Co-founder Alexander Cutler explained on The Defiant podcast that 100% of staked fees and incentives go to voters. He said voters generally back the pools producing the most rewards.

Supporters say this aligns everyone, since voters want the pools that earn the most fees. Critics say rewards can be bought, and pools can look busy only because someone is paying. Aerodrome and Velodrome announced in November 2025 that they would merge into a single exchange called Aero. In September 2026 they set a launch date of October 21, 2026, across seven networks. The video says Aero plans to replace the weekly vote with rewards pointed at pools continuously, so some details will change.

Comparing the four DEX types

DEX type	Example in the video	Who you trade against	Main hidden cost or risk
Pool (AMM)	Uniswap	A token pool funded by liquidity providers	Slippage, sandwich bots
On-chain order book	Hyperliquid	Traders, market makers, a community vault	Spread, needs deep markets, spoofing
Aggregator	Jupiter	Whatever pools and order books it routes to	An extra layer to trust, same underlying risks
Vote-based pool	Aerodrome	A pool whose rewards are set by voters	New token emissions can dilute holders if fees don’t keep up

The video’s point is that none of these models is free. Each one moves risk around rather than removing it. Add the spread, the slippage and anything lost to bots, and a swap with a small visible fee can turn out expensive.

Decentralized also doesn’t mean safe. Smart contracts can have bugs, pools can be drained, and there is no customer service line on a smart contract. The upside is that you keep your coins until the moment you trade, and there is no exchange account to freeze. Even so, as the video notes, big stablecoins can still be frozen by their issuers.

Three habits before your next swap

The video closes with three habits:

1. Know the exchange type. Pool, order book, router or voting system. The type tells you who you are trading against and how prices are set.
2. Set your own slippage. On pools especially, the video calls a sensible slippage limit your best defense against bots. If a trade fails because of it, the setting did its job.
3. Check the pool size before a big trade. A small pool means a big price move. The video suggests splitting large trades or using an exchange with deeper liquidity.

What to check next

Most DEX apps show the expected price impact before you confirm a swap. If that number looks large, the video’s advice is to stop and rethink. It is also worth checking whether your wallet offers private transaction sending. Aero was scheduled to launch on October 21, 2026. Check whether it launched as planned and how its move from weekly votes to continuous rewards works in practice.