Market Structure & Microstructure
Why the best displayed price is not always the best executable price. Depth, fees, queue priority and adverse selection change the cost of completing an order.
A quote is an invitation to trade a particular quantity under particular rules. Comparing only the best offer discards most of that information. For an execution desk, the relevant comparison is the cost and uncertainty of completing the intended order, given the liquidity it can actually access.
The spread is a boundary condition
The quoted spread is the difference between the best bid and offer. Crossing it buys immediacy for the quantity available at the touch; it says little about the next unit. Beyond that quantity, the shape of the book matters. A narrow spread beside shallow depth can produce a more expensive sweep than a wider spread supported by size.
Depth is also conditional on access. Two books cannot be treated as one pool if balances, credit, instrument specifications or trading permissions prevent execution on one of them. Fragmentation separates both information and the ability to act on it. The earlier note on liquidity in fragmented markets develops that distinction.
A better touch, a worse completed order
| Input / result | Venue A | Venue B |
|---|---|---|
| Accessible offers | 100 at 100.01; 900 at 100.12 | 1,000 at 100.04 |
| Taker fee | 2 basis points | 1 basis point |
| Average fill price | 100.1090 | 100.0400 |
| Cost per unit, including fee | 100.1290218 | 100.0500040 |
| Cost above midpoint | 12.90 basis points | 5.00 basis points |
For A, the average fill price is (100 × 100.01 + 900 × 100.12) / 1,000 = 100.109. Including the fee gives 100.109 × (1 + 0.0002) = 100.1290218. For B, 100.04 × (1 + 0.0001) = 100.050004. Dividing each difference from 100.00 by 100.00 and multiplying by 10,000 gives the basis-point costs above.
A advertises the cheaper first unit, yet B costs about 79.02 currency units less for the full 1,000-unit order under these assumptions. This compares two complete sweeps, not an optimized split route. It does not predict a saving: cancellation, latency, fees assessed differently or a funding constraint can change the result. Nor does a static book sweep estimate the market's response after the order becomes visible.
The Iran digital-asset liquidity pilot applies size and coverage measurements to a bounded sample of public books. Its static pre-fee estimates differ from the fee-inclusive hypothetical example here and are not realized execution costs.
Passive execution changes the uncertainty
A resting limit order exchanges immediate execution for price control. Its fill depends on subsequent order flow, cancellations, priority and the venue's rules. On a price-time book, joining a large queue does not give immediate access to all incoming flow. Coinbase's matching-engine documentation, for example, specifies price-time priority and execution against the resting order's price. That is a venue-specific mechanism, not a universal description of every market.
A passive order can avoid crossing the spread and still be costly. Adverse selection arises when the circumstances that produce a fill are associated with an unfavorable change in fair value for the liquidity provider. A high fill rate can therefore coexist with poor post-fill outcomes. Inventory pressure can further alter the value of another fill, even at the same displayed price.
Markouts help examine this selection: compare the fill price with a reference midpoint after a stated interval, using a consistent trade-side convention. They are diagnostics, not proof that a counterparty was informed. A price move can reflect common news, the trade's own impact, a stale reference or other flow. Report the horizon and reference venue; a one-second and a one-minute markout answer different questions.
Price discovery is not a routing instruction
Price discovery concerns how information becomes reflected in market prices. A venue may be informative while offering little accessible size. Another may support a useful execution without leading the price adjustment. A rapidly changing touch should therefore prompt questions about information and freshness, rather than automatically dictate where the entire order goes.
The SEC's 2010 concept release examines market quality through several measures, including spreads, execution speed, fill rates and price impact, and discusses public price discovery. It is a historical US-equities review. Its questions are useful analytical starting points; its legal setting and empirical observations should not be transplanted to digital assets as current rules or universal results.
Measure the decision as well as the fill
An execution review needs the intended quantity, decision and arrival timestamps, benchmark, available routes, limits, fills and unfilled remainder. Looking only at successful fills can reward a route that completes easy orders and leaves difficult ones behind. Conversely, a faster route may be appropriate for urgency even when its realized spread cost is higher.
- Before routing: compare fee-adjusted depth for the actual size, then apply funding, venue and risk constraints.
- During execution: track partial fills, remaining quantity and the validity of the market view. A cancel request is not a confirmed cancellation.
- After execution: compare realized costs with the stated benchmark; examine completion, delays and markouts separately, including difficult market periods.
The practical distinction is between a competitive quote and a competitive execution process. A screen can rank prices. A desk must decide which prices are reachable, what can change before completion and which risks remain afterward.
Sources and scope
- SEC (2010), Concept Release on Equity Market Structure — historical US-equities discussion of market quality and price discovery, especially sections III and IV.C.
- Coinbase Exchange, Matching Engine — a specific example of priority, resting-order prices and order states; documentation checked October 18, 2026.
The numerical comparison is constructed for this note. The execution-review framework is editorial analysis, not a description of a particular desk's trades or systems.