High frequency trading, usually shortened to HFT, is a style of automated trading that places, cancels and fills large numbers of orders at very high speed. The aim is to capture small, repeatable profits from the microstructure of electronic markets, often by holding positions for seconds or less.
All HFT is algorithmic, but not all algorithmic trading is high frequency. The difference is time horizon and order churn. HFT focuses on ultra-low latency, rapid order management and tight risk control around small, short-lived edges on an exchange order book.
What makes trading high frequency
There is no single cut-off, but HFT tends to show a few traits:
- Speed as an input. Systems react in microseconds or milliseconds to price changes and order book updates.
- Short holding periods. Inventory is often flat by the end of the day, and many positions last seconds.
- High order-to-trade ratio. Many quotes are updated or cancelled without trading as conditions change.
- Focus on microstructure. Strategies depend on queue position, tick size, fee tiers and matching rules.
HFT usually operates on central limit order books in equities, futures and currencies, and also on crypto venues with fast APIs. It relies on predictable matching engines and consistent market data feeds.
How HFT firms get their speed
- Co-location. Placing servers in the same data centre as the matching engine reduces physical distance and therefore latency.
- Optimised networks. Low-latency fibre, microwave links between cities and tuned network stacks cut transmission time between venues.
- Lean code and hardware. Code paths are minimised, often in C or C++, sometimes with FPGAs for specific tasks.
- Direct market data. Firms subscribe to raw feeds and parse them quickly to update fair values and order queues.
- Smart order routing. Algorithms choose where and how to place or cancel quotes across venues while passing pre-trade risk checks.
Some setups use direct market access through a broker’s infrastructure, while others connect using their own memberships. Exact arrangements vary by venue and provider.
Common HFT strategies and where profit comes from
- Electronic market making. Posting buy and sell quotes around a fair value, earning the bid-ask spread when both sides trade. The job is to update quotes as prices move and to manage adverse selection when informed flow appears.
- Cross-venue arbitrage. Buying where a price is momentarily lower and selling where it is higher. This can be the same instrument on two venues or closely linked products.
- Latency or queue arbitrage. Reacting to new information faster than others, for example, cancelling a stale offer just before it would be lifted, or joining the front of a queue when price is about to move.
- Rebate capture and fee optimisation. Maker-taker venues may pay rebates for providing liquidity. When combined with tight risk control, the economics of fees and rebates can matter as much as price movement.
- Index and futures alignment. Keeping related markets in line, such as a liquid future and its underlying basket or index, by leaning quotes toward the side where hedges look favourable.
| Strategy | Typical holding time | Main risks |
|---|---|---|
| Market making | Milliseconds to minutes | Adverse selection, inventory swings, sudden spread collapse |
| Cross-venue arbitrage | Seconds | Legging risk if one side fills and the other does not, outages |
| Latency/queue tactics | Milliseconds | Race conditions, cancelled orders, fees eating edge |
Per-trade profits are usually tiny, so total return depends on scale, consistency and tight control of costs such as fees, connectivity and slippage.
Order types and microstructure details HFT relies on
Because HFT lives inside the order book, order type choice is central:
- Limit orders provide liquidity and set queue position. First in the queue is usually first to trade, so timing matters.
- Market orders take liquidity when the priority is speed of fill rather than price control.
- Immediate-or-cancel (IOC) and fill-or-kill (FOK) limit partial fills and help manage risk when prices are moving.
- Hidden and iceberg orders reduce information leakage but can change queue dynamics when the hidden part refreshes.
- Pegged orders track a reference, such as the best bid or midpoint, useful for staying competitive without constant manual price changes.
Microstructure details such as tick size, auction rules, opening and closing crosses, and maker-taker fees shape the edge. Execution quality is judged by speed, slippage and how often fills arrive at the intended price, which ties directly to execution quality metrics.
Where you will encounter HFT in practice
Most investors meet HFT indirectly. You see it in tight spreads on liquid shares or futures, and in how quickly quotes update after news. If you place a limit order and get a partial fill followed by a rapid price move, you may be feeling the presence of faster liquidity providers ahead in the queue.
Institutional desks may interact with HFT when using smart order routers, participating in auctions, or deciding which venues to use. Broker algorithms are often designed to reduce signalling to fast traders, for example by slicing orders and randomising placement.
Benefits, criticisms and limits
- Benefits. HFT can narrow spreads, deepen displayed liquidity and keep related markets aligned, which lowers trading costs for many users most of the time.
- Criticisms. Very fast order updates can create noise, increase message traffic and make it harder for slower participants to avoid adverse selection. Some tactics that seek to mislead other traders, such as spoofing, are illegal in many places.
- Limits and safeguards. Feedback loops can amplify short-term volatility, so venues use tools like throttles, kill switches and trading pauses. Rules differ by country and can change, and firms must comply with risk controls set by venues and regulators.
HFT in crypto and other markets
On centralised crypto venues, HFT looks similar to equities and futures: fast market data, API access, fee tiers and maker-taker economics. Liquidity tends to concentrate in the most traded pairs, where spread capture and cross-venue arbitrage are feasible. On-chain trading has different constraints since block times and mempool dynamics limit how fast orders can settle, and transaction ordering introduces its own set of behaviours that are not identical to traditional HFT.
In all markets, the core idea stays the same. Speed is used to manage inventory and pick up small, low-risk edges over and over, assuming the technology, fees and operational risks are under control.