Systems / HFT
High-frequency trading systems: what actually separates HFT from an ordinary EA
Almost every retail product that claims to be high-frequency is an expert advisor trading a chart. Real high-frequency trading is a different discipline: the decision is cheap and the execution is hard. Here is what the label should mean, how it connects to the FIX API and DMA, and how to tell whether a system earns it.
What HFT is, and what it is not
High-frequency trading means holding positions for very short horizons and relying on speed, market structure and execution quality rather than on forecasting. It is not “a strategy with a fast setting”. The signal can be trivial; what is difficult is acting on it before the opportunity is gone.
Not scalping an indicator
An EA that crosses two moving averages on an M1 chart is a fast discretionary strategy. It has no latency budget and no market-structure edge.
Not guaranteed profit
Speed gets you a fill; it does not tell you where the price goes. Risk control is what turns a fast fill into a survivable book.
An execution problem
In practice, most of the engineering is the path from decision to acknowledged fill: protocol, routing and placement.
The latency budget
Every HFT system has a budget: the total time from the moment a price is observed to the moment an order is acknowledged. If the opportunity lasts 50 milliseconds and your round trip is 80, the strategy is a spectator. The budget is spent in four places, and one of them is usually ignored.
Feed
How the price reaches you — the broker’s market data feed, its depth, and how often it is refreshed.
Decision
The time your logic takes to turn a tick into an order. Usually the cheapest part to optimise, and the part most products obsess over.
Transport
Distance to the execution venue. This is why colocation — LD4/LD5 in London, Equinix in New York — matters more than a faster CPU.
Venue
Queueing, matching and the broker’s own internal path. You rarely control it, but you can measure it and route around the bad paths.
Why HFT runs on the FIX API
A high-frequency system needs three things a retail platform cannot reliably give it: a deterministic order type, a numbered acknowledgement, and no dealer layer between the decision and the market. That is exactly what the FIX API provides. FIX 4.4 is a session protocol — logon, heartbeats, sequence numbers and resend — so a lost message cannot silently corrupt your order state, and fill-or-kill lets the strategy refuse a stale price rather than accept it.
The practical consequence is that an HFT signal which looks brilliant in a backtest can be worthless without this stack. On a bridged retail account, the same rule set produces fills that are late, partial, or priced away from the quote. Routing the order through a FIX API connector over direct market access is what makes the modelled edge survive contact with a live account.
Session integrity
Sequence numbers and heartbeats mean a broken connection is known in seconds, not discovered at the next failed trade.
Immediate order types
FOK and IOC are native, so “no fill” is an available outcome instead of an unavailable one.
Measurable path
Every order and fill is a message with a timestamp, which makes the latency budget auditable rather than asserted.
Fill quality beats raw speed
A system that is 20 ms faster but accepts bad fills will lose to a slower system that refuses them. The reason is arithmetic: on a short horizon the edge per trade is a few pips and the cost of one bad fill is larger than the profit of several good ones. So the execution policy matters at least as much as the speed.
Fill or kill
Take the price that was quoted or take nothing. This is how you stop a fast strategy from being paid in slippage.
Nothing rests
Resting orders advertise your position and can be picked off. An HFT system wants an immediate outcome, not a queue position.
Risk fixed first
The stop and the maximum loss are decided before the order exists, so no fill can create uncontrolled exposure.
Algorithmic trading, quant and where HFT sits
HFT is a subset of algorithmic trading, and algorithmic trading is the execution arm of quantitative trading more broadly. The distinction is useful when evaluating a provider: a slow, well-researched systematic strategy and a microsecond execution engine are different products with different costs, and marketing language blurs them deliberately.
A reasonable test is to ask what the system does when the market does not cooperate. An HFT system should be able to decline the trade — that is what FOK is for. A discretionary EA can only take the price and hope. Whichever it is, the outcome should be measurable on a live account rather than in a screenshot.
Quant
Decides what to trade and how much, from data and validation rather than opinion.
Algo
Decides how the order reaches the market: work it, chase it, or refuse it.
HFT
Compresses both into a horizon where execution quality is the entire edge.
How PULSE-X is built
PULSE-X is XTRSK’s high-frequency system. It runs over the FIX 4.4 protocol, sends fill-or-kill orders, and keeps a strict latency budget between the tick and the acknowledgement. One instrument, one position at a time, a hard stop written before entry, and no grid, martingale or averaging down anywhere in the design.
Execution is routed with direct market access so the cost of a round trip is known rather than estimated, and the result is published on the live account page. If you want the research method behind the systems rather than the plumbing, read how XTRSK does quantitative trading.
High-frequency trading questions
How fast is “high frequency” in FX?
It is a horizon rather than a number. In retail-accessible FX, HFT usually means holding seconds to minutes with a round-trip latency budget measured in tens of milliseconds rather than microseconds.
Do I need the FIX API to run an HFT system?
You need an execution path that supports immediate order types and does not sit behind a dealer or a bridge. In practice that means a FIX API route; MT4/MT5 can be bridged into one, but the platform adds delay.
Is colocation necessary?
Not for every strategy, but it is the only way to remove the transport component of the latency budget. If the opportunity is short-lived, distance decides the outcome.
Can HFT work on a small account?
The mechanics do not depend on account size, but the cost of execution does. Below a certain size, spread and commission can exceed the average move being captured.
Does XTRSK guarantee HFT returns?
No. Faster execution improves the odds that a defined edge is captured; it does not create one, and leveraged trading can lose money quickly.
Further reading
The protocol reference for the order types described above is the FIX Trading Community standards library. For what a carrier-neutral colocation facility actually is, see Equinix’s London data centres.
High-frequency execution changes how an order is filled, not whether the market moves in your favour. Trading leveraged FX and metals involves significant risk of loss and is not suitable for everyone.
Chat with XTRSK
Chat ready
Start a chat and the XTRSK team will be notified immediately.