Skip to content

This is general knowledge or an external-engine reference, not an Apollo architecture or implementation claim. See Apollo's architecture status for target terminology and verified implementation boundaries.

Q46: How do you design an auction house? ​

Core Conclusion ​

An auction house is not "an offline version of the trading system"; it is "an asynchronous market plus a search index plus a settlement and clearing system".

The real difficulty is usually not the listing screen, but:

  • Escrow and locking of listed items
  • Search indexing and sorting
  • Deal settlement and fees
  • Expiration, cancellation, and bidding conflicts

Once the player base grows, the auction house quickly becomes a comprehensive system spanning data, caching, search, and asset safety.

One: First Distinguish the Auction House Business Models ​

There are two common kinds:

  • Fixed-price listing
  • Bid-based auction

The two differ greatly at the system level:

  • Fixed-price is closer to a commodity market
  • The bidding model emphasizes deadlines, bid ordering, and final settlement

Many projects run both, but they should not be mixed into one ambiguous flow.

Two: The Fundamental Difference Between an Auction House and Face-to-Face Trading ​

An auction house is asynchronous trading, and usually has these characteristics:

  • Both sides do not need to be online simultaneously
  • Traded items are held in escrow by the system
  • The deal happens at an independent point in time
  • Query and filtering pressure is heavy

So it is more like a "market service" than simple two-person interaction.

Three: The Most Critical Part of Listing Is Escrow ​

Once an item enters the auction house, it must leave the player's free inventory.

That is, the system must either:

  • Directly transfer item ownership into an auction escrow state
  • Or at least lock the item and forbid all other operations

Otherwise typical problems appear:

  • The item can still be used or traded after being listed
  • Conflicts between delisting and deals
  • Duplicate listing

Four: Search and Filtering Are the Performance Core of an Auction House ​

Players usually search by these dimensions:

  • Type
  • Quality
  • Profession
  • Price range
  • Remaining time

So the auction house system cannot only think about "having a listing table"; it must also consider:

  • Search indexes
  • Sort fields
  • Pagination performance
  • Caching of popular filter conditions

This is also why auction houses often need an additional cache or search layer.

Five: Deals and Clearing Must Be Designed Independently ​

After a deal, the system usually still has to handle:

  • Deduction
  • Ownership change of the item
  • Fee
  • Crediting the seller's proceeds
  • Notifying buyer and seller

This is a complete clearing chain; it cannot be treated as just "updating one status".

Six: The Bidding Model Especially Needs Attention to Timing ​

If bidding is supported, you must first define:

  • Minimum bid increment
  • Deadline
  • Rules for bids at the very last moment
  • How funds are handled after being outbid

Otherwise concurrent bids near the end easily cause problems.

Seven: A More Solid Organizational Approach in Engineering ​

A common approach is:

  • The main listing table holds the trade's main state
  • A search index layer carries queries and sorting
  • The asset system handles escrow and deal clearing
  • Scheduled tasks handle expiration, unsold items, and settlement

That way the auction house does not cram search, assets, and notifications into a single table.

Eight: Risk Control and Auditing Are Very Important ​

An auction house is inherently a high-risk system, because it is easily used for:

  • Money laundering
  • Studios transferring assets
  • Exploiting abnormal pricing for arbitrage

So you usually also need:

  • Listing and deal ledgers
  • Abnormal price monitoring
  • Account correlation analysis
  • Restrictions on high-risk items

Nine: Common Misconceptions ​

1. An Auction House Is Just a More Complex Trade ​

Not accurate. It also includes search, escrow, settlement, and risk control.

Once the scale grows, sorting and filtering quickly become performance bottlenecks.

3. After a Deal, Just Swap the Money and Item Directly ​

In practice you still have to handle fees, expiration, notifications, retries, and idempotency.

References ​

  • Various online game auction house, market index, and asset clearing practice materials

Released under the MIT License