2.1.4 Auctions and Continuous Trading

Basic Prerequisites: 03 - Exchanges & Venues.

1. Intuition & Practical Objective

Equities and futures do not trade continuously all the time: they open and close with a call auction, and many instruments (and the whole of the FX fixings) are set periodically. An auction is a fundamentally different mechanism from the continuous book, and understanding it is required to trade the open/close and to reason about market-design proposals. The twin facts:

A call auction collects orders during a window, then executes them all at one uniform clearing price that maximizes the volume that can trade. Nobody gets price-discriminated against - everyone crosses at the same price.
Continuous trading executes orders on arrival, so early orders get better prices than late ones, and the price history is a path, not a single point.

The practical objective: compute an auction’s clearing price and volume from a supply/demand schedule; contrast it with continuous price–time matching; and understand why openings, closings, and fixings are auctions (they need one price for valuation/settlement) and why designers worry about manipulating the close and about last-instant bidding (Hasbrouck Ch 2).

The one-sentence essence. “A call auction answers ‘what single price clears the most volume?’ while continuous trading answers ‘who arrived first at the best price?’ - the first is a market-clearing equilibrium, the second a price–time queue.”

2. Mathematical Ground Truth & Derivations

Uniform-price double auction. Suppose buy limit orders with limits and sizes ${(p_{i}^{b}, q_{i}^{b})}$ and sell orders ${(p_{j}^{s}, q_{j}^{s})}$ . At a candidate clearing price $p$ , define

D (p) = i \sum q_{i}^{b} 1 {p_{i}^{b} \geq p}, S (p) = j \sum q_{j}^{s} 1 {p_{j}^{s} \leq p}, E (p) = min (D (p), S (p)) .

The executable volume $E (p)$ is maximized at the clearing price

p^{*} = ar g p max E (p), V^{*} = E (p^{*})

( $E$ is unimodal; ties broken by exchange rule, e.g. closest to the previous price). All buy orders with $p_{i}^{b} \geq p^{*}$ and all sell orders with $p_{j}^{s} \leq p^{*}$ execute, all at $p^{*}$ - buyers do not pay their limits and sellers do not receive theirs. The market is a single-price market at the instant of the auction.

Why uniform pricing is the design. Because $p^{*}$ is common to every fill, no participant is penalized for being early, and the price is an un-manipulable clearing price rather than the last of a sequence. This is why fixings (WM/Reuters FX, SOFR, closing benchmarks) are auctions: a benchmark defined as one price is hard to move with a single small trade, whereas a last trade price is trivial to nudge.

Continuous trading for contrast. Orders execute on arrival under price-time priority; the realized prices are a sequence ${p_{1}, p_{2}, \dots}$ , and the day’s volume-weighted price is $\sum_{k} v_{k} p_{k} / \sum_{k} v_{k}$ with $v_{k}$ the trade size. Continuous gives immediacy and a live path but disperses prices; the auction gives one price at the cost of waiting for the window.

Market-design concerns (Hasbrouck Ch 2).

Random stopping times. To defeat last-instant “sniping,” exchanges end the window at a random instant within a band, so a trader cannot know exactly when to inject a price-moving order.
Early deadlines. Orders that could destabilize the clearing (e.g., large market-on-close orders) are subject to earlier cut-offs or price-collar constraints.
“Mark the close” manipulation. Because the closing price settles derivatives and index funds, there is a direct incentive to trade the close to move it - which is why closing auctions are heavily monitored and collared.

Continuous vs batch - the wider design debate. Budish, Cramton & Shim (2015) argue that the continuous-time matching of modern exchanges manufactures a latency arms race, and propose frequent batch auctions (all orders in a tiny interval clear together, tie-broken by size) as a design that removes the speed race - the modern, formal echo of the call-auction mechanism.

3. Computational Implementation - clearing an auction, then trading it continuously

Build a supply/demand schedule, find the volume-maximizing uniform clearing price, then replay the same flow through a price–time continuous book and compare the price dispersion. Standard library only; re-executed and reproduced.

Read the numbers. The executable-volume curve peaks at $p^{*} = 100.10$ with $V^{*} = 2000$ shares - at that single price, 2100 shares of demand and 2000 shares of supply overlap, so 2000 trade and every one of them crosses at $100.10$ , whether their limit was $100.10$ or $100.30$ . Replayed continuously, the same order flow prints 2100 shares at prices ranging from 100.10 to 100.30, with a volume-weighted average of 100.1810 - early buyers got 100.10, late sellers got 100.30, and price was discriminated. The auction delivered one fair price; the continuous book delivered immediacy at the cost of a $\sim 20$ -cent price band (100.10 to 100.30).

4. Failure Modes & First-Principles Breakdowns

Last-instant sniping in the auction. If the closing time is deterministic, a trader can inject a large order in the final microsecond to move $p^{*}$ . Random stopping times and early deadlines exist precisely to defeat this; a deterministic window is a standing invitation.
Marking-the-close manipulation. Because closing prices settle ETFs, index funds, and derivatives, there is real money in nudging $p^{*}$ . Continuous-only traders underestimate this; the manipulation is in the auction, not the tape.
Assuming the auction price equals the continuous price. They are different mechanisms and generally give different prices. An execution algorithm benchmarked to the continuous VWAP can look “cheap” or “expensive” at the open/close purely because the auction cleared somewhere else.
The auction rewards size, the continuous rewards speed. Under continuous price–time priority, speed (arriving first) wins; under a batch auction, ties can be broken by size (largest orders fill first), inverting the advantage. A strategy tuned for one mechanism mis-executes in the other.

5. References

Hasbrouck, Joel - Empirical Market Microstructure (2007)
Budish, Cramton & Shim - “The High-Frequency Trading Arms Race: Frequent Batch Auctions as a Market Design Response,” QJE 130(4), 2015. The modern market-design argument for batch auctions over continuous matching.
Foucault, Pagano & Röell - Market Liquidity (2013)
Harris, Larry - Trading and Exchanges (2003). The practitioner treatment of call auctions, openings, and closings.

6. Connected Graph Bridges

Back: 03 · Exchanges & Venues · Index Hub
Continue: 05 · Failure Modes & Practice · 06 · Advanced Extensions
Matching and priority: Matching & Priority
Execution benchmarks: Scheduling & Volume Profiles

Kwant Atlas

Explorer