On-Chain Microstructure

Block latency, mempool dynamics, MEV exposure, and the gap between transaction inclusion and economic finality.

35 min · expert · part of Trading Algorithms & Market Microstructure

Why Microstructure Is Different On-Chain

In traditional equities, market microstructure is the study of how orders become trades — the rules of the limit order book, the role of market makers, the latency arms race between exchanges and colocation providers. On-chain markets share the same underlying questions but answer them in radically different ways. There is no central matching engine. There is no NBBO. There is no SEC rule that protects you from being front-run by the entity processing your order. Instead, every order is a transaction broadcast into a public network, and the block producer who builds the next block has wide latitude to choose which transactions to include, in what order, and on what terms. For an algorithmic trader, this changes almost everything. Your fill price is not just a function of the order book at the moment you submit — it is a function of who sees your transaction before it lands, what they can do to it, and how long the network takes to confirm it. A strategy that backtests beautifully against historical mid-market prices can lose money systematically in production because the difference between simulated and realized fills is dominated by latency, gas markets, and adversarial reordering rather than spread. This lesson grounds the rest of the module. You cannot reason about backtesting honesty, strategy selection, or production infrastructure without first understanding what actually happens between "I press send" and "my balance changes." We start with block time across major venues, move through mempool dynamics, then through the MEV exposures every order carries, and end with the distinction between inclusion latency and economic finality. Each of these is a separate cost line in your real PnL, and ignoring any of them is the surest way to publish a strategy with a beautiful Sharpe ratio that loses money the moment it touches mainnet.

Also in this lesson

Block Time: The Heartbeat of Every Venue
Mempool Dynamics: Public vs Private
MEV Exposures: Sandwich, Frontrun, Backrun
Inclusion Latency vs Settlement Finality
Putting It Together: A Realistic Execution Cost Model

Key terms

Block time: The average interval between consecutive blocks on a blockchain. Sets a floor on transaction confirmation latency. Ethereum 12s, Solana ~400ms, Base 2s, Arbitrum ~250ms batched.
Mempool: The staging area where pending transactions wait to be included in a block. On EVM chains the default mempool is public and any participant can observe pending orders.
Private RPC: An endpoint that routes transactions directly to block builders without first broadcasting to the public mempool. Examples: Flashbots Protect, MEV-Blocker. Removes most sandwich exposure.
MEV (Maximal Extractable Value): Value extractable by block producers and searchers through reordering, insertion, or censorship of transactions. Shows up to traders as sandwich attacks, frontrunning, and backrunning.
Sandwich attack: A common MEV extraction where a searcher buys before your AMM swap, lets your trade move price, and sells immediately after. Profits from your slippage.
Inclusion latency: The time between transaction submission and inclusion in a block. The relevant metric for execution speed.
Settlement finality: The point at which a transaction is irreversible. Differs from inclusion: Ethereum L1 finalizes in ~12-15 min, optimistic rollups have a 7-day L1 challenge window.
Sequencer: The component (typically centralized at launch) on an L2 that orders transactions before posting them to L1. Sub-second soft confirmations from the sequencer are usually treated as final for trading.
Commitment level (Solana): Solana commitment tiers: processed (next slot), confirmed (typical trading), finalized (~12-13s, 32 slots, used for high-value transfers).
Reorg risk: The probability that an included transaction is reverted via chain reorganization before achieving finality. Real on probabilistic-finality chains and during sequencer disruptions on L2s.

Continue this lesson — 5 more sections in the CryptoBipto app.

Open lesson

Educational only — not financial advice.