Latency Tier

The latency tier of a trader refers to the speed at which market information reaches them and orders are executed. Latency is measured in microseconds (μs), milliseconds (ms), and seconds. High-frequency traders operate in the 10–1,000 microsecond range; institutions typically experience 1–100 millisecond latencies; retail investors may see second-level delays. Reducing latency by even microseconds can mean the difference between profit and loss.

Latency components

Total latency has multiple components:

1. Market latency: Time for the exchange to execute an order (typically 1–10 microseconds for orders that can be matched immediately).
2. Network latency: Time for data to travel from the exchange to the trader’s system (typically 0.1–10 milliseconds).
3. Processing latency: Time for the trader’s algorithm to process data and generate an order (typically 0.1–10 milliseconds).
4. Transmission latency: Time to send the order from the trader’s system back to the exchange (typically 0.1–10 milliseconds).

Total end-to-end latency might be 1–100 milliseconds or more for most traders.

Latency tiers and trading styles

Ultra-low latency (10–100 microseconds): High-frequency traders. Infrastructure at this level is expensive (servers co-located at exchanges, specialized networking, custom code). Profits come from microsecond-level advantages.

Low latency (100–1,000 microseconds = 0.1–1 millisecond): Sophisticated institutions and algorithmic traders. Achievable with quality infrastructure and direct exchange connections.

Medium latency (1–10 milliseconds): Active traders and institutions without ultra-low-latency infrastructure. This is typical for most institutional traders.

High latency (10–1,000 milliseconds = 0.01–1 second): Retail traders using standard brokers. Brokers themselves may route orders quickly, but the retail client’s connection may add delays.

Very high latency (1–15+ seconds): Delayed data users. Free market data often has a 15-minute delay; some reports are even older.

Latency arbitrage

Latency differences create trading opportunities:

Example: A stock trades at $100 on the NYSE and $100.05 on NASDAQ. A high-frequency trader with direct feeds from both venues might:

1. See the price discrepancy (1 millisecond).
2. Buy on NYSE at $100.00 (2 microseconds).
3. Sell on NASDAQ at $100.05 (2 microseconds).
4. Profit $0.05 per share (say, 1,000 shares = $50).

A retail investor with 1-second latency cannot execute this arbitrage; by the time they see the prices, they have moved.

Controversy and fairness

Latency advantages are controversial:

Arguments in favor:

• Speed encourages market-making and liquidity.
• Technology investment is rewarded.
• Information is still ultimately available to all.

Arguments against:

• Creates information asymmetry; fast traders see prices before slow ones.
• Enables predatory strategies (front-running, spoofing).
• Disadvantages retail investors who cannot afford low-latency infrastructure.
• Contributes to flash crashes and volatility.

Infrastructure costs

Achieving lower latency requires expensive infrastructure:

• Servers and software: Custom-built systems optimized for speed. Cost: $100,000–$1,000,000+.
• Co-location: Placing servers at exchange data centers to minimize network latency. Cost: $10,000–$100,000+ per year.
• Direct feeds: Subscribing to exchange market data feeds. Cost: $1,000–$10,000+ per month.
• Network: High-quality, dedicated network connections. Cost: $10,000–$100,000+ per year.
• Total system cost: A full high-frequency trading setup: $1–$10 million or more.

This cost structure means that ultra-low-latency trading is accessible only to well-capitalized firms.

Measures of latency

Absolute latency: Total time from market event to trader decision (e.g., 5 milliseconds).

Relative latency: Latency difference between two traders or systems (e.g., “Firm A has 2ms advantage over Firm B”).

Latency distribution: Typical latencies vary; traders care about the worst-case (tail) latencies as much as average.

Regulation and latency

Regulators have discussed whether to mandate uniform latency or slow down fast traders:

• No direct regulation. The SEC has not mandated specific latency targets.
• Indirect regulation. SIP data is slower than direct feeds, but this is not a formal constraint.
• Proposed solutions. Some propose speed bumps (artificial delays to eliminate sub-microsecond advantages) or faster SIPs.

However, no major regulatory changes have materialized.

Latency and market stability

During flash crashes and volatile episodes, latency differences can exacerbate volatility:

• Fast traders see cascading bids disappearing and hit bids aggressively.
• Slow traders are still seeing outdated prices.
• By the time slow traders’ orders execute, prices have moved sharply.

This can lead to disorderly markets and execution at irrational prices.

See also