High-frequency trading
Also called HFT
Algorithmic trading that relies on very fast systems, co-location and low latency to send, cancel and execute large numbers of orders, holding positions briefly and ending the day nearly flat.

On this page· 6 sections
By the SEC staff’s account in 2020, most activity in US equity markets had for more than ten years come from professionals whose short-term strategies place many orders and generate many trades every day[1]. That activity is called high-frequency trading, or HFT. To do it, nearly every step of trading has to run through algorithms[1].
How it works
The US has no statutory or regulatory definition of HFT[1]. The SEC’s 2010 Concept Release listed five traits typical of firms that do it[1]:
- high-speed, sophisticated programs that generate, route and execute orders;
- co-location services and individual data feeds that cut network delays;
- brief time frames for opening and closing positions;
- many orders that are cancelled shortly after submission;
- ending the day as close to flat as possible, with no large unhedged positions overnight.
A firm does not need all five to count[1].
Under MiFID II, a high-frequency algorithmic trading technique combines three features[2]:
- infrastructure that minimizes latency, such as co-location, proximity hosting or high-speed direct electronic access;
- systems that start, route or execute orders without human intervention;
- high message intraday rates of orders, quotes or cancellations.
A delegated regulation puts the high rate at an average of at least 2 messages per second in any single instrument on a trading venue, or at least 4 per second across all instruments on it[4]. Firms using the technique must keep time-sequenced records of every order placed, cancelled or executed, and every quote, and hand them to the regulator on request[2].
Example
Say a firm quotes one instrument on one venue and updates its bid and offer each time the market moves. In an 8-hour session of 28,800 seconds, it sends 72,000 messages: new orders, changes and cancellations.
- Average rate: 72,000 ÷ 28,800 = 2.5 messages per second, above the single-instrument line of 2.
- Say 1,440 of those messages end in a trade. Then 1,440 ÷ 72,000 = 2% of the activity trades, and 98% never does.
The regulation looks at averages, not one day: on request, venues give firms monthly estimates of their rate based on the previous 12 months of messages[4].
Watch out
The label is used loosely. Nasdaq’s glossary applies it to computerized trading with proprietary algorithms in general, including execution algorithms that split large orders[3]. The SEC staff and EU law mean a narrower, speed-driven kind[1][2].
Heavy cancelling is a typical trait of HFT[1]. Submitting and cancelling orders with no intention to trade, to manipulate other traders, is spoofing, which the SEC staff list among harmful strategies HFT firms may use[1].
At TRONEXO
On the simulated Challenge accounts and TRONEXO Accounts, high-frequency trading and extremely short-term strategies are prohibited. That includes tick scalping, which the FAQ defines as “repeatedly opening and closing positions within up to 10 seconds in a way that forms a coherent trading pattern”. The rules say TRONEXO evaluates market analysis and judgment, not raw execution speed. See Funding models & rules and the FAQ.
Sources
- RegulatorU.S. Securities and Exchange CommissionStaff Report on Algorithmic Trading in U.S. Capital Markets (August 2020)Checked November 9, 2026
- RegulationEuropean Union (EUR-Lex, Official Journal)Directive 2014/65/EU on markets in financial instruments (MiFID II)Checked November 9, 2026
- ExchangeNasdaqHigh Frequency Trading (HFT) DefinitionChecked November 9, 2026
- RegulationEuropean Union (EUR-Lex, Official Journal)Commission Delegated Regulation (EU) 2017/565 supplementing Directive 2014/65/EUChecked November 9, 2026
