One-Leg Latency Arbitrage — Strategy, Execution, and Anti-Detection

Among the five core arbitrage strategies covered in the HFT Arbitrage Leitfaden, the one-leg variant is both the most direct and the most operationally aggressive. A reference price moves; the broker’s quote lags; you transact at the broker before its quote catches up. There is no hedge to coordinate, no second account to fund, no inter-leg slippage to model. Whatever edge the broker’s feed lag produces, you capture in full.

That directness has two consequences. The first is excellent capital efficiency: one account, one position, the full move. The second is exposure: the broker sees one hundred percent of your trading and bears one hundred percent of the adverse selection. Brokers respond to that pattern, and they respond quickly. This guide explains the mechanics, presents the new limit-order execution model that meaningfully extends operational lifetime, and identifies the conditions under which one-leg is the correct choice.

What one-leg latency arbitrage is — exactly

in einbeiniger latency arbitrage, every trade has a single venue: the slow target broker. The reference price feed — typically institutional CQG oder Rithmic futures, LMAX spot, Integral OCX, oder cTrader Raw — is used for decision-making only. No trades are placed at the reference venue. When the reference moves, the platform fires an order at the broker; when the broker’s quote catches up, the position closes; net profit is the difference.

Contrast this with Zweibeiner (Hedge-Arbitrage), where an offsetting position is opened at a fast venue to neutralise market risk during the holding interval. In one-leg, that hedge is absent: you carry directional market exposure for the holding interval, typically a few hundred milliseconds to a few seconds. Most of the time that exposure is harmless because the broker’s quote catches up quickly. Occasionally the market keeps moving and the trade closes for less than expected — sometimes for a small loss.

How a one-leg trade actually executes

The mechanical pipeline of a single one-leg trade, end to end:

1. Reference tick. The fast feed publishes a price update that leads the broker’s quote.
2. Signal. The platform compares the reference to the broker’s most recent quote and, if the divergence exceeds the configured threshold, generates a signal.
3. Order construction. An entry order is built for the target broker.
4. Execution at broker. The order is sent and filled (or placed and filled — see the execution model below).
5. Position open. The position now sits at a price the broker’s quote has not yet caught up to. Market risk is present until the broker’s quote converges.
6. Exit. When the broker’s price reaches the take-profit target, a trailing stop activates, or a stop-loss triggers, the position closes via a market order.

The whole sequence is typically over in 200 milliseconds to a few seconds. The edge per trade is small in absolute terms — a few points on a major FX pair, a few tens of cents on gold — but the trade count is high and the edge is, on a confirmed slow feed, close to deterministic per trigger.

Why one-leg has the best capital efficiency

Among arbitrage variants, one-leg produces the most profit per dollar of margin. Three reasons:

• Single broker spread cost. Hedge mode pays the spread on two legs (target + fast hedge); one-leg pays it once.
• Single broker commission. Same logic for any commission.
• No capital fragmentation. Hedge mode requires capital split across two accounts — typically with the larger share at the slow target. One-leg deploys the full capital at the productive broker.

In practice, the same edge captured in one-leg mode produces roughly 30–60% more net profit per dollar than the same setup in two-leg hedge mode. The difference is meaningful at small capital where every dollar of fixed cost matters, and it is what makes one-leg the default starting point for new arbitrage operators despite its other drawbacks.

Risks specific to one-leg

1. Directional market risk during the hold

For the duration of the holding interval — typically 200 milliseconds to a few seconds — the position has full market exposure. Most of the time the market does not move enough to matter. Occasionally, particularly around news events or sudden order-flow imbalances, the market keeps moving in the direction of the reference, eating into the edge or reversing the trade outright.

2. Maximum detection signature

The broker sees one hundred percent of your trading. Every entry, every exit, every win, every loss. There is no hedge account to dilute the pattern. Detection happens faster, restriction follows sooner, and the operational lifetime at any single broker is shorter than in hedge mode.

3. Concentration risk at one broker

Capital sits at one venue. If that broker has a withdrawal problem, a regulatory event, or a sudden change in execution policy, the entire position — and the cash funding it — is at the same point of failure.

4. Quote-freeze hazard

If the broker freezes its quote during volatility (a common occurrence around scheduled news), naive software keeps generating signals against the frozen price. Production-grade one-leg requires explicit freeze detection and an auto-pause.

Execution: from market orders to limit orders with TTL

The most consequential change to one-leg execution in 2026 is the move from market-order entry to limit-order entry with a platform-side TTL. The strategy logic is unchanged. What changes is what the broker’s surveillance sees.

The toxicity of market-order entry

Consider what a market-order entry looks like from the broker’s side. The reference price moves. Ten to twenty milliseconds later, an account submits a market order, lifts the broker’s stale ask, and the position opens at a price the broker’s quote has not yet updated to reflect. Across many sessions the pattern is mechanically distinctive: aggressive taker, perfectly timed against external reference moves. No discretionary trader produces that pattern. Detection is fast.

The same edge can be captured with a different order type that produces a categorically different signature.

GTC limit entry on MT4 / MT5 — patient, then cancel

Auf MT4 und MT5 — the platforms most slow-feed brokers offer — the new model submits a GTC (Good Till Cancelled) limit order at an offset inside the broker’s spread, and the platform enforces a TTL (Time To Live) cancellation timer locally. If the broker’s quote crosses the limit price within the TTL, the order fills. If the TTL expires with no fill, the platform cancels the order. There is never a naked-leg outcome from this design.

Two properties of this design are important. First, the broker sees only an ordinary GTC limit order — exactly the order type a discretionary trader uses when they want to be filled at a specific price. The TTL is platform-side; the broker is not told about it. A cancellation after a fraction of a second is also ordinary — many discretionary traders re-price aggressively. The execution profile contains no exotic instructions and no broker-visible automation hints.

Second, on FIX-API connections — used by a smaller proportion of operators — the same logic can be expressed natively as an IOC-style order with a short validity window, and FOK semantics are available for use cases where any partial fill is unacceptable. FOK is, however, rarely used on MT4/MT5 because those platforms do not natively support it; on retail platforms, GTC + platform TTL is the practical model.

The slippage offset — placing the limit inside the spread

Where exactly the limit is submitted matters. Too far from current price and it never fills; at the touch it shows aggression. The platform’s Schlupf parameter, configured per symbol, defines an offset from the broker’s current price at which the limit is placed.

A limit at the mid is not a market-aggressor — it is a price-improving order, the kind a market-maker would submit. The broker’s surveillance categorises it accordingly. Typical slippage settings:

• 0 points (order at the touch): maximum fill probability, most aggressive signature. Rarely useful.
• Half-spread (mid): best balance of fill probability and signature quality. Recommended starting value.
• Half-spread + small margin: more market-maker-like, but fill probability drops if the broker’s quote doesn’t catch up.

Exits — trailing stop, take-profit, stop-loss (market orders)

The exit side of one-leg uses a different mechanism. Limit-order exits create a problem the entry side does not have: an unfilled close leaves the position open through subsequent volatility. The practical solution on MT4/MT5 is straightforward — exits are configured as trailing stop, take-profit, oder stop-loss levels attached to the open position. When any of these triggers, the broker closes via a market order.

The market exit is appropriate here for the same reason limit entry is appropriate: incentives match the requirement. On entry, the platform can afford to wait — a missed fill is just no trade. On exit, the platform cannot afford to wait — an open position must be flat when the strategy says so. Trailing-stop and TP/SL market exits are immediate and unambiguous; the position is either still open at the broker or it is flat.

On FIX API integrations, where FOK semantics are natively available, FOK can be used for close orders when complete-fill-or-cancel behaviour is required (for example, when partial fills would interfere with downstream logic). On MT4/MT5, FOK is not a native option; the trailing-stop / TP / SL mechanism handles the same operational requirement with native platform tools.

The new settings — what each parameter does

Trading options panel (global)

• Open with limit + GTC — checkbox + dropdown. Enables limit-order entry; GTC is the order type submitted to the broker. The platform-side TTL field enforces the cancellation timer.
• Close with limit + FOK — applicable to FIX API connections only. Leave unchecked on MT4/MT5 (exits use trailing/TP/SL market orders instead).
• Min. order lifetime + random — minimum holding time and randomisation window. Both contribute to making holding times look discretionary rather than mechanical.
• TTL (ms) — platform’s cancellation timer on the GTC open limit. Typical range 200–2000 ms.
• Ignore min lifetime for S/L — emergency exits bypass the minimum holding rule.

Per-symbol grid

• Lot size — first numeric column.
• Volume mode — Fixed or one of the dynamic modes.
• Slippage (points) — offset that defines where the limit is placed in the spread.
• Distance (points) — minimum reference-vs-broker gap required to trigger a signal on this symbol.
• Take-profit / stop-loss / trailing-stop (points) — the exit triggers; firing any of them produces a market close.
• Random delay, random size — anti-detection jitter on order timing and lot size.

Recommended configuration by broker profile

Broker profile	Slippage offset	TTL (ms)	Anmerkungen
Slow third-party feed (200+ ms lag)	Half-spread	1000–2000	Generous TTL lets the broker quote catch up; high fill rate.
Mid-tier hybrid (100–200 ms lag)	Half-spread	400–800	Tighter TTL because the lag closes faster.
Borderline (50–100 ms lag)	Touch or +1–2 points	200–400	Limit close to ask; signature gain smaller but still meaningful.
Fast ECN (under 25 ms lag)	—	—	Not a viable one-leg target regardless of order type.

Market-order entry vs limit-order entry — the honest comparison

Metrik	Market-order entry	GTC limit-order entry
Fill rate on triggered signals	~99%	70–90% depending on TTL and offset
Average entry price	Stale ask (worst in spread)	Mid or better (price improvement)
Signaturerkennung	High — aggressive taker	Low — passive in spread
Operational lifetime at one broker	Weeks	Months
Naked-leg risk on entry	Niedrig	None (TTL expiry cancels cleanly)

The lower fill rate is more than offset, on a slow target broker, by the better fill price and the extended operational lifetime. Fewer triggered signals convert into trades, but each completed trade contributes more, and the account survives meaningfully longer before broker-side throttling. Where the choice goes the other way — borderline-fast brokers where the lag is shorter — the fill-rate gap dominates and market-order entry can produce better per-session yield.

One-leg vs hedge — when each is the right choice

Both variants exploit the same broker feed lag. They differ in how they manage the risk and detection that come with it.

Faktor	One-leg fits	Hedge fits
Kapital	Unter $10k insgesamt	$10k+ insgesamt
Profit per dollar	Higher	Herunter
Per-trade risk	Higher (directional)	Lower (market-neutral)
Signaturerkennung	Stronger (mitigated by limit-order model)	Weaker by design
Lebensdauer pro Broker	Weeks (months with limit-order entry)	Months
Betriebliche Komplexität	Simple — one account	Zwei-Makler-Koordination
Prop-firm compatibility	Arm	Possible (2-legs latency 3 variant)

Common mistakes in one-leg setups

1. Targeting a fast broker. No order-type sophistication compensates for the absence of feed lag. Verify the lag before committing capital.
2. Using market-order entry on a slow broker. Leaves substantial signature on the table; switch to limit entry as the default once the broker is confirmed slow.
3. Too-tight TTL. If the broker’s lag is 800 ms and the TTL is 300 ms, almost nothing fills. Match TTL to observed lag.
4. Zero randomisation. Even with limit entry, perfectly regular timing and sizing reveal mechanical execution. Use the random-delay and random-size parameters.
5. Skipping news blackouts. Limit orders left in the book through major releases get filled at terrible prices or left stranded. Disable trading 30–60 seconds around scheduled red-impact events.
6. Keine tägliche Verlustbegrenzung. One-leg’s directional risk means a feed glitch or broker quote freeze can produce a string of losses. Hard caps on daily P&L are standard.
7. Single-broker concentration. One-leg means one venue means one point of failure. Rotate target brokers; never deploy the full operation at one account.

Häufig gestellte Fragen

When is one-leg better than hedge mode?

At small capital (under roughly $10k total), when broker selection has already produced a clearly slow target, and when prop-firm compatibility is not required. At those conditions the simpler operational footprint and the higher yield per dollar of one-leg outweigh its detection and concentration disadvantages.

Does GTC limit entry work on every MT4/MT5 broker?

Limit orders are universally supported; the only complication is that some MT4 broker configurations restrict pending orders inside the spread (the “stops-level” or “freeze-level” settings). On such brokers the slippage offset must be set just outside the touch rather than at the mid. The platform configuration accommodates this per symbol.

Why not always use limit-order entry?

On borderline-fast brokers — where the broker’s lag is short — the fill-rate cost of limit entry can exceed the signature benefit. Market-order entry on those brokers may produce better per-session yield even though the signature is worse. Configure per broker rather than as a global default.

What is the minimum capital for one-leg latency arbitrage?

$2,000–$5,000 is the practical minimum after accounting for the fixed costs of a fast reference feed, VPS in LD4 oder NY4, and a meaningful test budget. Below that, fixed costs absorb too much of the gross profit. The strategy becomes meaningfully more efficient at $10,000+ where lot sizes can be increased without crossing broker-flagging thresholds.

Can one-leg arbitrage pass prop-firm rules?

Generally no. The directional, single-broker, short-hold profile of one-leg matches almost every prop firm’s prohibited-practices clause for latency arbitrage and tick scalping. For prop-firm operation the appropriate variant is hedge arbitrage in the 2-legs latency 3 configuration — see the Hedge Arbitrage guide.

How does the limit-order model interact with news blackouts?

It does not change the requirement for news blackouts; it makes them more important. A limit order resting in the book through a major release will either fail to fill (broker quote skipped over the level) or fill at a price that is no longer reasonable when the quote returns. Disable trading 30–60 seconds around scheduled high-impact events as standard practice regardless of order type.

What about volatile but unscheduled moves?

The TTL handles them cleanly. A limit that does not fill within the TTL expires; the platform simply produces no trade for that signal. There is no naked-leg or stranded-order scenario as a structural matter, regardless of how the market behaves between signal and TTL expiry.

Will brokers eventually detect the limit-order pattern too?

Specific accounts can be profiled even with the new model — particularly if other parameters (lot size, timing, instrument selection) remain mechanically regular. That is what the randomisation controls are for. The categorical shift from “aggressive taker” to “passive market-maker” in the order-type fingerprint is the largest single signature improvement available; combining it with timing and sizing jitter materially raises the operational lifetime ceiling.

Weiterlesen

• HFT-Arbitrage – Der vollständige Leitfaden — pillar covering every arbitrage type
• Latenzarbitrage erklärt — the broader mechanism that one-leg belongs to
• Hedge Arbitrage — Two-Broker Strategy — the alternative when capital, prop-firm compatibility, or detection risk demand it
• Wie wählt man einen Broker für HFT-Arbitrage aus — selecting the slow target broker
• Live-Performance-Überprüfung
• Glossar für HFT und Arbitrage-Handel