Centralized and decentralized exchanges solve the same problem (matching buyers with sellers) through fundamentally different mechanisms, and the liquidity characteristics that result from each approach affect everything from execution quality to what tokens you can actually trade.
CEX liquidity is built on traditional order book models. Professional market makers post limit orders at various price levels, creating a ladder of bids and asks. Binance's BTC/USDT book might have tens of millions of dollars within 1% of the current price. This depth allows large orders to execute with minimal slippage, which is why most institutional volume still routes through centralized venues.
DEX liquidity works through automated market maker (AMM) pools, most commonly using the constant product formula (x * y = k) popularized by Uniswap. Instead of professional market makers, anyone can deposit token pairs into pools and earn trading fees. The price is determined algorithmically based on the ratio of tokens in the pool rather than by individual order placement.
The tradeoffs are significant. CEX liquidity is deeper for major pairs but requires trust in the exchange. FTX demonstrated the risks of that trust model spectacularly. DEX liquidity is permissionless and transparent (you can verify reserves on-chain) but tends to be thinner and subject to impermanent loss for providers.
For long-tail tokens, DEXs have a clear advantage. Any token with a smart contract can have a trading pair on Uniswap or Raydium without any listing process. This means new tokens are tradeable on DEXs hours or days before they might appear on centralized exchanges. The tradeoff is that these early pools are often thin and volatile, with high slippage for any meaningful size.
Concentrated liquidity, introduced by Uniswap V3, changed the DEX landscape significantly. Instead of spreading liquidity across all prices from zero to infinity, providers can concentrate it within specific ranges. This makes DEX liquidity more capital-efficient and can produce tighter spreads for active ranges, sometimes competitive with CEX spreads for popular pairs.
MEV (maximal extractable value) is a DEX-specific consideration that does not exist on centralized exchanges. On-chain transactions are visible in the mempool before confirmation, allowing searchers to frontrun, backrun, or sandwich trades. This hidden cost can add 0.5-2% to the effective cost of a DEX trade, particularly for larger orders or tokens with thinner liquidity.
The flow of liquidity between CEX and DEX tells a story about market structure evolution. During periods of centralized exchange distrust (post-FTX, during regulatory crackdowns), DEX volumes spike as traders move to self-custodial solutions. During calm periods, volume tends to migrate back to CEXs for better execution.
Hybrid models are emerging that try to capture the best of both worlds. Projects building on-chain order books, intent-based trading systems, and DEX aggregators that route across multiple pools are all attempts to improve DEX execution quality. The gap between CEX and DEX liquidity is narrowing, but for most trading sizes and use cases, centralized venues still offer better execution on major pairs.
The practical implication is that where you trade should depend on what you are trading. Major pairs on major CEXs give the best execution. New tokens and DeFi-native assets often require DEX access. And understanding the hidden costs of each venue (exchange risk on CEXs, MEV on DEXs) is essential for accurate cost accounting.