source: https://godzilla.dev/learning/ai_quant_traders_series_11/
See below for godzilla.dev materials about: AI x Quant Trader Series - Day 11
What is Market Data?¶
Reading time: ~15 minutes
Prerequisites: What is High Frequency Trading, What is Market Microstructure, What is an Order Book
Focus: understanding the data flowing through modern electronic trading systems
Part 1: Introduction¶
Every quantitative trading system begins with one thing.
Market Data.
Before a strategy can decide whether to buy or sell, it must first understand the current state of the market.
That information comes from market data.
Whether you are trading:
Stocks
Futures
Options
ETFs
Cryptocurrencies
every trading decision ultimately depends on a continuous stream of market events.
For High Frequency Trading, market data is not just information.
It is the raw material from which every trading opportunity is created.
Part 2: What is Market Data?¶
Market Data is the real-time information published by an exchange describing everything happening in the market.
Typical market data includes:
Best Bid
Best Ask
Trade Price
Trade Size
Order Book Updates
Volume
Market Status
Instrument Information
Every update represents a new event occurring inside the exchange.
Unlike historical datasets, market data never stops arriving.
It is an infinite stream of events.
Part 3: Types of Market Data¶
Modern exchanges usually provide several categories of market data.
Trade Data¶
Trade data records completed transactions.
Example:
Price: 101.20
Quantity: 5 BTC
Time: 09:30:15.123456
Trade data answers one question:
What actually traded?
Quote Data¶
Quote data describes the current market.
Typical information includes:
Best Bid
Bid Size
Best Ask
Ask Size
Example:
Bid
101.18
Size 25
Ask
101.20
Size 40
Most execution algorithms continuously monitor quote updates.
Order Book Data¶
Rather than publishing only the best prices,
many exchanges provide multiple price levels.
Example:
Ask
101.30
101.20
101.10
100.90
100.80
100.70
Bid
This information allows trading systems to reconstruct the entire local order book.
Part 4: Snapshot vs Incremental Updates¶
Exchanges generally publish market data in two formats.
Snapshot¶
A snapshot contains the complete market state.
Example:
Entire Order Book
↓
One Message
Snapshots are simple but expensive to transmit frequently.
Incremental Updates¶
Incremental updates publish only changes.
Example:
Before
101.20
Size 30
↓
Update
Size 18
Only the modified information is transmitted.
Nearly every modern HFT platform relies primarily on incremental updates because they minimize bandwidth and latency.
Part 5: Market Data Feed¶
Exchanges distribute market data through specialized data feeds.
A simplified architecture looks like:
Exchange
↓
Market Data Feed
↓
Decoder
↓
Local Order Book
↓
Trading Strategy
The market data feed is responsible for delivering every market event to participants as quickly as possible.
For High Frequency Trading,
the market data feed is often the most latency-sensitive component of the entire system.
Part 6: Why Latency Matters¶
Imagine two trading firms receive the same market update.
Firm A processes the update in:
8 μs
Firm B processes it in:
120 μs
Both firms observe the same opportunity.
Only one is likely to execute first.
This is why HFT engineers spend enormous effort optimizing:
Message parsing
Memory allocation
Cache locality
Lock-free queues
Network I/O
Every microsecond matters.
Part 7: Market Data Processing¶
Receiving market data is only the beginning.
A production trading system must also:
Decode exchange protocols
Validate messages
Handle sequence numbers
Detect packet loss
Recover missing data
Maintain synchronization
Update the local order book
These operations occur continuously throughout the trading day.
For active markets, this may involve millions of messages every second.
Part 8: Local Market Data¶
Professional trading systems rarely query the exchange whenever market information is needed.
Instead, they maintain an in-memory representation of the market.
Exchange
↓
Market Data Feed
↓
Incremental Updates
↓
Local Memory
↓
Trading Strategy
Strategies then read data directly from memory.
This architecture eliminates unnecessary network latency and dramatically improves performance.
Part 9: Market Data in High Frequency Trading¶
For long-term investors,
market data is simply information.
For HFT systems,
market data is an event stream.
Strategies react to:
New trades
Quote changes
Order book updates
Liquidity changes
Spread changes
Market imbalance
Many HFT strategies process thousands of events before placing a single order.
Understanding event flow is often more important than predicting future prices.
Part 10: Where godzilla.dev Fits¶
Efficient market data processing is one of the foundations of every ultra-low latency trading platform.
A production implementation must:
Decode exchange messages
Process incremental updates
Maintain local market state
Synchronize order books
Distribute events across multiple strategies
Minimize memory copies
Maintain deterministic latency
These requirements define much of the architecture behind godzilla.dev.
Rather than rebuilding market data infrastructure for every project, developers can focus on strategy research while relying on a modular, high-performance framework designed for modern electronic markets.
Part 11: Key Takeaways¶
Market Data is the real-time information published by exchanges describing market activity.
It includes:
Trades
Quotes
Order Book Updates
Market Status
Instrument Information
Professional trading systems transform this continuous stream of events into an in-memory representation of the market, allowing strategies to react with minimal latency.
Understanding market data is the first step toward building production-grade trading infrastructure.
What's Next?¶
The next article explores the component responsible for turning incoming orders into completed trades:
How Matching Engines Work
Top comments (0)