Market Microstructure

The price on a screen is often not the price you can actually get. A stock may show $100.00, but the real market is a queue of bids, offers, hidden orders, cancelled quotes, broker routing choices, and machines reacting in microseconds. Market microstructure studies that plumbing: how intentions become trades, and how the act of trading changes the price itself. If efficient market hypothesis asks whether prices reflect information, microstructure asks who got to trade before the price moved.

How prices get made

Most electronic exchanges use a limit order book. Buyers post bids. Sellers post asks. The exchange matches orders by rules such as price-time priority: the better price wins, and ties go to the order that arrived first.

The bid-ask spread is the toll for immediacy. If the best bid is $99.98 and the best ask is $100.02, a buyer who needs certainty now pays 4 cents more than a patient buyer placing a limit order. That sounds tiny until it touches a million shares, a thin bond, or a strategy trading thousands of times per day.

Mechanism	What it changes	Why it matters
Market order	Trades now	Pays the spread and may walk the book
Limit order	Waits at a chosen price	May never fill
Hidden order	Conceals size	Reduces signaling, may lose queue priority
Iceberg order	Shows only part of size	Hides intent while still participating
Maker-taker fee	Pays or charges for liquidity	Can change broker routing incentives

A market price is not a single fact. It is a conditional offer for a certain size, at a certain venue, at a certain instant.

Liquidity is not one thing

Liquidity has at least three moving parts: spread, depth, and resilience. Spread is the gap between buy and sell quotes. Depth is how much size is available near the current price. Resilience is how fast the book refills after trades consume it.

This is why small orders and large orders live in different markets. A retail order for 10 shares of Apple can usually execute near the quoted price. A fund trying to buy 5 million shares has to worry about signaling, impact, routing, and time. The trade itself becomes information.

The 6 May 2010 Flash Crash showed how fast the machinery can thin out. The Dow Jones Industrial Average fell about 9% intraday and recovered much of it within minutes. The event did not require a single villain; it exposed a market where automated liquidity could retreat faster than humans could understand the tape.

Information, speed, and adverse selection

Market makers quote both sides, but they are not charities. They earn spreads while carrying the risk that the person trading against them knows more. That risk is adverse selection.

If informed buyers hit offers before quotes update, the market maker sells too cheaply. If informed sellers hit bids first, the market maker buys too expensively. Spreads are partly payment for standing in that danger zone.

Electronic trading made time part of the market object. Some firms measure advantage in microseconds. Colocation, microwave links, and direct data feeds matter because a stale quote can be picked off. Michael Lewis made this visible in Flash Boys in 2014, but the academic argument predates the book by decades.

What's contested

The hardest fight is not whether speed matters. It does. The fight is whether speed competition improves markets enough to justify its private arms race.

Defenders of high-frequency trading point to narrower spreads in many liquid equities since decimalization and electronic trading. Critics point to latency arbitrage, queue-jumping tactics, exchange fee conflicts, and fragile liquidity during stress. Both can be true: markets can be cheaper in calm weather and more brittle when everyone cancels at once.

Payment for order flow is another live dispute. Retail investors often receive price improvement relative to the public quote, but their orders may never interact with the open exchange book. The open question is whether that improves execution for individuals while weakening public price discovery.

Why this has to do with other realms

Microstructure is a design problem disguised as finance. It resembles distributed systems because exchanges must order events, handle latency, prevent inconsistent states, and survive bursts of traffic. A matching engine is a social contract compiled into code.

It also connects to human attention. The old market maker watched faces, phones, and order flow. The new market maker watches packets, queues, and cancellation rates. The battlefield moved from shouting in a pit to interpreting traces left by other machines.

An open question

If a market is cheaper for small trades but less trustworthy for large trades during stress, did the design improve liquidity or only make fragility harder to see? The next page to write is payment for order flow.

Key Sources

Market Microstructure Theory by Maureen O'Hara (1995) — core text on spreads, dealers, inventory, and information.
Glosten and Milgrom, “Bid, Ask and Transaction Prices in a Specialist Market with Heterogeneously Informed Traders” (1985) — canonical adverse-selection model.
Kyle, “Continuous Auctions and Insider Trading” (1985) — foundational model of market impact and informed trading.
U.S. SEC and CFTC, “Findings Regarding the Market Events of May 6, 2010” (2010) — official Flash Crash report.
Trading and Exchanges by Larry Harris (2003) — practical map of orders, venues, spreads, and trader motives.

Abhishek