What Is the Forex Spread: Fixed vs. Variable Options

Before a position moves in the trader’s favour, it must clear that gap.

For most major pairs, one pip is the fourth decimal place: 0.0001. If EUR/USD is quoted at 1.08420 bid and 1.08430 ask, the bid-ask spread in trading is 1 pip. A market buy opens at 1.08430. Immediate liquidation occurs at the bid, 1.08420. The position starts 1 pip below break-even, before swaps, commissions, or slippage enter the ledger.

That mechanism is simple. The cost profile is not. A fixed quote creates a stable entry threshold but can embed a larger baseline charge. A variable quote can be extremely narrow during liquid hours, then widen sharply when liquidity disappears or news hits. The spread label on an account page is therefore not a complete pricing specification. The order-routing model, session, instrument, and execution logic matter.

A quoted spread is not a broker fee schedule in miniature. It is a live measurement of the cost to cross the market at that moment.

The mechanics of bid-ask spreads in currency markets

Every FX quote has two executable sides:

• Bid: the price at which the trader can sell the base currency.
• Ask: the price at which the trader can buy the base currency.
• Spread: ask minus bid, normally expressed in pips.

For a pair quoted to four decimal places, the calculation is direct:

For USD-denominated accounts trading a standard lot of EUR/USD, a pip is commonly worth about $10. A 1-pip spread therefore begins as roughly a $10 round-trip friction in price terms for that position size. The exact monetary value changes with pair structure, trade size, account currency, and conversion rate. The pip count is the more portable metric.

The spread pays for several things at once. It can reflect external liquidity-provider pricing, a broker’s mark-up, internalisation risk, and the cost of making a continuous two-sided market. The trader does not need to identify each component to calculate the immediate cost. But the distinction becomes material when comparing account types.

A standard account may show EUR/USD with a typical spread-only cost around 0.8 pips. AUD/USD may average around 1.0 pip; NZD/USD around 1.6 pips. These are useful reference points, not fixed execution guarantees. A platform’s charting stack may display an attractive average while the order ticket records a different price at the moment the order reaches the matching layer.

This is why spread analysis should use the prices available to trade, not a mid-price chart. A mid-price is an analytical construct: halfway between bid and ask. It is useful for indicators and visual cleanliness. It is not an executable exit for a long position or an executable entry for a short one.

The distinction is especially visible in short-horizon systems. A strategy targeting 2 pips on EUR/USD is not operating in the same cost regime as a position strategy targeting 150 pips. For the first system, spread, fill quality, and order-routing delay are core variables. For the second, they may be secondary, though never absent.

Fixed spreads: stable arithmetic, not guaranteed best pricing

A fixed spread remains unchanged under the broker’s stated normal conditions. Major pairs are commonly offered in a range of roughly 1.5 to 3.0 pips. The account holder sees the same nominal entry cost whether the London market is active or a major data release is due.

This is usually associated with market-maker or dealing-desk execution. The broker may internalise client flow rather than pass each ticket directly into external liquidity. That structure allows it to publish a fixed bid-ask differential. It does not mean that the external market itself has stopped moving. It means the broker has chosen to absorb, reprice, delay, reject, or otherwise manage the gap between its client quote and available hedging prices.

The operational advantage is predictable cost modelling. An automated strategy can budget a known spread into its entry and exit conditions. A trader testing a bot on historical data has fewer moving parts in the cost model. The backtest still needs realistic assumptions for fills and latency, but the spread parameter is not changing every few seconds.

Fixed pricing is also mechanically easier to audit. If EUR/USD is specified at 2 pips, the expected opening disadvantage is clear. There is no need to reconstruct a time series of quote widths to estimate the baseline transaction cost.

That simplicity has limits. A fixed spread does not guarantee that every order will receive the displayed price. During a fast market, the execution engine may apply requotes, reject a marketable order, or fill it after a delay. A stable quote paired with unstable execution is not stable trading cost.

The relevant question is therefore not merely “is the spread fixed?” It is:

1. Does the platform maintain the quote during fast price updates? A fixed spread is less useful if the order ticket repeatedly requotes.

2. What is the fill policy for market orders? Positive and negative slippage rules matter more than branding language.

3. Are stop orders triggered from bid, ask, or a separate reference price? A wide fixed spread can affect stop placement from the first second of the trade.

4. Does the broker widen or suspend fixed pricing outside normal liquidity windows? “Fixed” often describes ordinary dealing conditions, not every possible market state.

5. Can the API endpoints return the same executable quote shown in the terminal? Strategy logic and order execution should operate on consistent price data.

Fixed spreads are often suitable for systems that require a known initial cost and place relatively few trades. They can also make sense for new traders who need clean arithmetic while learning position sizing. Neither point establishes that fixed pricing is cheaper. It establishes that the cost is easier to model.

Fixed spread pricing reduces variance in the fee input. It does not remove execution risk from the system.

Variable spreads: the liquidity-sensitive model

Variable, or floating, spreads change in real time. They are most often associated with ECN and STP account structures, where quoted prices are linked more directly to available liquidity-provider quotes.

Under deep liquidity, a major pair can show a variable spread from 0.0 to 0.5 pips. The London–New York overlap is the standard example: multiple large pools are active, quote competition is stronger, and the order book usually has more depth close to the top of book.

The same pair can widen materially when the market thins out. This occurs around session transitions, daily rollover, holidays, and high-impact releases. During major news events, variable spreads can reach 8 pips or more. The precise peak depends on the pair, venue, liquidity conditions, and the broker’s quote aggregation rules. There is no universal maximum.

The technical logic is straightforward. Liquidity providers widen their own quotes when the probability of adverse selection rises. If a bank is at risk of buying just before a rapid fall or selling immediately before a rapid rise, it demands more distance between bid and ask. The retail platform receives that wider input. Its aggregation layer may pass it through, add a mark-up, or filter it according to its configuration.

A tight minimum spread is not an average spread. An average spread is not a maximum spread. And neither statistic indicates how the quote behaves precisely when a strategy is most exposed.

This is where broker comparison pages often collapse distinct data points into one headline number. “Spreads from 0.0 pips” states a floor. It does not state the duration of that floor, the average during the trader’s active session, the spread during scheduled releases, or the fill probability at the best displayed level.

For a variable-spread account, the useful observation window is the trader’s actual operating window:

• A European morning strategy should inspect the spread distribution during that period, not only the 24-hour average.
• A news strategy must measure quote widening before, during, and after releases. A normal-session average is irrelevant to its principal risk.
• An Asian-session system needs data for its specific pairs. EUR/USD may remain workable while less-liquid crosses behave differently.
• A scalper needs tick-level evidence where possible: spread, latency, rejected orders, and slippage belong in the same test set.

Charting can obscure this. Many platforms plot bid-derived candles by default. A long trade exits at bid, but enters at ask. A short trade enters at bid, but exits at ask. If the charting stack does not show an ask line or does not retain historical spread data, visual backtesting will understate friction.

A variable spread is therefore not a defect. It is a market-sensitive input. The fault lies in treating the narrowest observed quote as a permanent property of the account.

Raw accounts separate spread from commission

Raw-spread accounts move the pricing architecture one level closer to the underlying market. The variable spread may begin at 0.0 or 0.1 pips, but the broker charges a separate commission per lot traded.

This is not zero-cost trading. It is a different cost layout.

A standard spread-only account combines market spread and broker compensation into one quoted bid-ask gap. A raw account exposes more of the market spread, then places the broker’s compensation in the commission column. That separation is operationally useful because it allows the trader to model the components independently.

The all-in cost must include both:

For a raw account, the correct calculation is:

All-in cost in pips = live spread in pips + round-trip commission expressed in pips.

The commission must be converted using the instrument’s pip value and trade size. A commission figure without its charging convention is incomplete. “Per lot” may mean per side or per round turn. It may use a standard lot definition. It may vary by account tier, instrument, or monthly volume. The trade ticket and fee schedule need to agree.

Raw pricing often benefits active intraday traders because a small reduction in all-in cost can matter across hundreds of entries. But this only holds if the strategy trades when liquidity supports the narrow underlying spread and if the commission does not erase the advantage.

There is another limitation. Raw accounts do not neutralise market stress. A 0.1-pip normal quote can still widen when liquidity withdraws. The fixed commission remains fixed; the underlying spread does not. A system that assumes “raw” means permanently tight has confused account nomenclature with live market conditions.

For API-driven trading, fee transparency also needs to be visible in the data flow. The execution report should make it possible to reconcile:

• submitted order timestamp;
• acknowledged order timestamp;
• fill price and quantity;
• bid and ask at the decision point;
• commission booked;
• any partial fills;
• realised swap or financing charge for positions held past rollover.

Without that record, the trader can see total P&L but cannot identify whether underperformance came from signal decay, wider spreads, poor order routing, or commission drag.

Matching the spread model to the trading system

Fixed versus variable spread forex pricing is not a general ranking exercise. It is a compatibility test between cost behaviour and strategy behaviour.

A low-frequency trader who holds positions for days may accept a wider fixed spread if the entry cost is minor relative to the expected move and the broker’s overnight swap rates are competitive. For that trader, financing can exceed the difference between 0.5 and 1.5 pips over the life of the position.

A scalper has the opposite profile. The spread is often the dominant controllable cost. A 1-pip difference is not marginal if the system’s gross target is only several pips. Variable or raw pricing can be more efficient during liquid sessions, provided the platform has stable order routing and the trader avoids periods when spreads expand.

Automated systems divide further:

• Mean-reversion bots can be damaged by sudden spread expansion because entries and exits cluster near short technical thresholds.
• Breakout bots need explicit news and liquidity controls. A wide spread can trigger entries at a much worse effective level than the signal model assumed.
• Market-making or grid logic is highly spread-sensitive. A static backtest using a single 0.5-pip assumption is not a meaningful production test.
• Swing systems should include spread but place more weight on swaps, margin rates, and the broker’s handling of stops through rollover.

The platform itself is part of the fee system. A terminal that displays only a static “from” spread but offers no historical spread metric leaves a material gap in cost analysis. A DOM depth panel can show visible liquidity, but it may not represent all executable liquidity or guarantee a fill at the top level. API endpoints can provide useful quote and fill telemetry, but only if timestamps are precise enough to distinguish market movement from platform lag.

Execution delay becomes expensive when spreads are moving. If the trading interface freezes for a fraction of a second while the quote stream changes, the nominal spread at signal generation is not the spread at fill. The resulting cost is not always recorded as a separate fee. It appears as a worse execution price. From the strategy’s perspective, it is still transaction friction.

The correct comparison is all-in and time-specific

The question “what is the forex spread?” has a narrow answer: the bid-ask difference. The useful answer is broader. It is the immediate price of crossing from one side of the FX market to the other, with behaviour determined by account type and liquidity conditions.

Fixed spreads offer a known quoted cost, commonly around 1.5 to 3.0 pips on major pairs. Variable spreads can contract to 0.0–0.5 pips in deep liquidity and can widen sharply in unstable conditions. Raw accounts expose a tighter variable spread but add commission. None of these labels, on their own, predicts the final cost of a completed trade.

The stable selection rule is binary. Use fixed pricing when the system needs predictable spread inputs and the quoted level fits its expected return. Use variable or raw pricing when the strategy can trade liquid windows, measure all-in costs, and tolerate a spread that changes with the market.

If the broker cannot provide reliable execution records, coherent fee entries, and a trading interface that remains responsive while quotes move, the spread type is secondary. The pricing model is not stable enough for serious cost analysis.