Transaction Costs in Retail FX: Spreads, Commissions, Swaps, Slippage and the Cost of a Busy Signal

Retail forex transaction costs are the sum of four things: the spread between the bid and ask price, any commission, the overnight rollover (swap) charged or paid on positions held past the daily cut-off, and slippage between the price you expected and the price you got. Each looks tiny per trade, but they are charged on the full leveraged position, so a signal that trades several times a day can lose more than the whole account to costs in a year even if its trades are slightly better than random.

In short: a typical round trip on one standard lot of EUR/USD costs a retail trader around 1.5 pips (about $15) once spread and a modest allowance for slippage are included, roughly 1.4 basis points of the $110,000 position. That is small for a trade held for weeks and enormous for a scalp aiming at 5 or 10 pips. In our worked example, a signal trading four round trips a day at one lot on a $10,000 account pays about $15,300 a year in costs, 153% of the account. With a 10-pip target and stop, it needs a 57.5% win rate just to break even. For comparison, Levich and Thomas (1993) estimated that a large institution trading currency futures 65 times a year paid about 1.6% of the position a year.

Signal sellers rarely show costs. This piece explains how each one is built, what the research says, and how to estimate the drag before you subscribe.

The misconception: “spreads are a fraction of a pip, costs don’t matter”

Retail platforms advertise spreads “from 0.0 pips” or “as low as” a few tenths of a pip. Those figures describe the best moment in the best pair, not the average fill of a signal follower. More importantly, a cost that is small relative to the price can be large relative to the trade’s expected profit. A signal with a 10-pip target that pays 1.5 pips round trip is giving up 15% of every winner and adding 15% to every loser.

The academic literature has known this for decades. Almost every serious study of FX trading rules reports results net of costs, and the fastest strategies are usually the ones that fail the test. (The trading-rule evidence is in do technical trading rules work in FX?)

Context: where retail sits in the FX market

The Bank for International Settlements’ 2025 Triennial Central Bank Survey put global over-the-counter FX turnover at $9.51 trillion a day in April 2025 in its final detailed tables (the preliminary release of September 30, 2025 had rounded to $9.6 trillion), up about 27% from $7.5 trillion in 2022. Spot trading was $2.95 trillion a day (31% of the total), outright forwards $1.75 trillion and FX swaps $4.02 trillion. Trading with reporting dealers was 47% of turnover, with other financial institutions 49% (hedge funds and principal trading firms alone 8%), and with non-financial customers 5%.

Retail is a small slice of that. When the BIS first collected a retail breakdown, in the 2013 survey, Dagfinn Rime and Andreas Schrimpf reported in the BIS Quarterly Review (December 2013) that “retail trading accounted for 3.5% and 3.8% of total and spot turnover, respectively,” and 19% of spot turnover in Japan. In the 2025 tables, “retail-driven” turnover was about $242 billion a day, roughly 2.5% of the total, of which $109 billion was spot (3.7% of spot). The 2013 article also described the “retail aggregators” that sit between small traders and the dealer banks: they pool many small client trades, internalize some of the flow and lay off the rest with dealers. It judged that retail platforms’ internalization ratios differ widely but “are unlikely to exceed 50%.”

That structure explains the cost stack. The interbank market for EUR/USD is extraordinarily liquid; the retail price you see is the dealer’s price plus a markup that pays for the dealer’s hedging, risk and profit. How dealers decide which flow to hedge is covered in forex broker conflicts of interest.

The four costs, one at a time

1. The spread

The spread is the gap between the price at which you can buy (the ask) and sell (the bid). You pay half of it, on average, when you open and half when you close, so a round trip costs one full spread. On EUR/USD, one pip is $10 per standard lot of 100,000 euros. (Pip arithmetic is in pips, lots and signal math.)

US retail dealers generally publish two account structures:

Spread-only accounts, where the dealer’s compensation is built into a wider spread and no separate commission is charged. Published figures for EUR/USD on such accounts are typically expressed as “as low as” values around 1 pip, and actual spreads widen around news, at the daily rollover and in thin markets.
Commission accounts (“raw” or “core” pricing), where the spread is closer to interbank and a fixed commission is charged on each side of the trade, per 100,000 units of base currency. Our worked example assumes $5 per side, so $10 per standard-lot round trip on top of the spread; actual schedules differ by dealer and change over time.

The two structures often cost about the same for a standard lot of EUR/USD. They differ in which pairs and trade sizes they favour and in how visible the cost is.

2. Commission

Commissions are explicit and easy to model: a fixed dollar amount per 100,000 units per side, sometimes per million. Some dealers also publish volume rebates for very active clients, paid per million traded. Rebates matter for the signal industry because introducing brokers and affiliates are commonly paid per lot their referrals trade, which gives a signal seller a reason to prefer frequent trading. See how signal providers make money.

3. Rollover (swap)

A spot FX trade settles two business days after the trade date for most pairs. A retail position that stays open past the daily cut-off (5 pm New York time is the usual convention) is rolled forward, and the account is credited or debited the interest differential between the two currencies for the extra day, adjusted by the dealer’s markup.

Direction matters. Holding the higher-yielding currency earns the differential; holding the lower-yielding one pays it.
The markup always works against you. The dealer pays less than the differential on the receiving side and charges more on the paying side. On a book that is long as often as it is short, the differentials roughly cancel and what remains is the markup, paid every night.
Wednesday counts triple. Because of two-day settlement, a position rolled on Wednesday moves its value date across the weekend, so the rollover covers three days.

As an illustration, OECD data via FRED put US 3-month interbank rates at 3.94% and German (euro) 3-month rates at 2.64% in September 2026. With an assumed 1% a year financing markup on each side, one standard lot of EUR/USD ($110,000 at 1.10) would cost about $6.93 a night held long and earn about $0.90 a night held short; the long Wednesday rollover would be about $20.79. On a position held half the time long and half short, the average cost is the markup alone, about $3.01 a night. Markups differ between dealers and are disclosed on their rollover pages; the 1% here is an assumption, not a quoted rate.

4. Slippage

Slippage is the difference between the price a signal names and the price you actually receive. It is usually small for market orders in liquid pairs at quiet times and can be large for stop orders in fast markets, at weekend gaps and around news. Signal followers face an extra layer, because by the time a signal is posted, read and executed, the market has often moved. That mechanism is worked through in slippage and execution on copied trades.

Our worked example assumes 0.3 pips per round trip (0.15 pips per fill). That is deliberately mild; a follower acting on delayed alerts would usually do worse.

How institutions measure the same costs

The research literature gives a useful yardstick for what costs look like at the professional end of the market.

Levich and Thomas (NBER Working Paper 3818, 1991; Journal of International Money and Finance, 1993) estimated the cost of a currency futures transaction for a large institution in 1976-1990 at about 2.5 basis points (4 basis points as a conservative figure), built from bid-ask spreads of $0.0001 to $0.0002 and a brokerage commission of $11 per round trip. For a rule trading about 65 times a year, that cut profits by “1.62% per year or 2.60% per year.”
Neely, Weller and Ulrich (2009), Journal of Financial and Quantitative Analysis, summarized the evidence that the cost of a switch from long to short “declined from about 10 basis points in the 1970s to about 2 basis points” by the mid-2000s, with spot spreads of “2 basis points or less, for transactions in the $5 to $50 million range.”
Neely and Weller (2013), Journal of Banking and Finance, compared quoted Bloomberg forward spreads with traders’ screens and concluded that “actual spreads were roughly one third of the quoted spreads.” Citing Burnside and coauthors (2007), they also note that emerging-market bid-ask spreads over 1997-2006 were two to four times those of developed currencies.
Lustig, Roussanov and Verdelhan (2011), Review of Financial Studies, used Reuters and Barclays quotes and noted Richard Lyons’ finding that such indicative spreads are “roughly twice the size of inter-dealer spreads,” so their net returns are conservative.
Menkhoff, Sarno, Schmeling and Schrimpf (2012), Journal of Financial Economics, showed how decisive costs can be: applying the full quoted spread cut their best currency momentum strategy from “nearly 10% to about 4% p.a.” and “wipe[d] out most of the profit of many other strategies.”

The lesson from this literature is twofold. Costs per trade fell sharply as FX trading went electronic. But even at institutional prices, a strategy that trades often must clear a meaningful hurdle, and retail costs per trade are not lower than an institution’s.

Worked example 1: the cost of one trade

All figures are for one standard lot of EUR/USD at 1.10 (notional $110,000, $10 per pip), computed in Python.

Account type	Spread	Commission (round trip)	Slippage	Total per round trip
Spread-only	1.2 pips ($12)	none	0.3 pips ($3)	$15.00 (1.5 pips, 1.36 bp)
Commission	0.2 pips ($2)	$10 ($5 per side)	0.3 pips ($3)	$15.00 (1.5 pips, 1.36 bp)

The 1.2-pip and 0.2-pip spreads are illustrative values consistent with published “as low as” figures for each account type; real average spreads vary by dealer, time of day and market conditions.

Worked example 2: annual cost drag by signal frequency

Take a $10,000 account that follows every signal at one standard lot. With a 10-pip stop, each trade risks $100, or 1% of the account, a conventional sizing rule; the position is 11 times the account in notional terms, well inside the US 50:1 limit for major pairs. Costs are $15 per round trip as above, plus $3.01 a night of net financing markup when positions are held overnight.

Signal style	Round trips a year	Overnight holding	Trading costs	Financing	Total cost, % of account
Scalper, 4 a day	1,000	10% of trades, 1 night	$15,000	$301	153%
Day trader, 1 a day	250	20% of trades, 1 night	$3,750	$151	39%
Swing, 3 a week	150	all trades, about 3 nights	$2,250	$1,356	36%
Position, 2 a month	24	all trades, about 10 nights	$360	$723	11%

The scalping signal turns over about $220 million of notional a year on a $10,000 account. Its costs are 13.6% of one lot’s notional per year, about eight times the 1.62% Levich and Thomas calculated for an institution trading 65 times a year. Note how the cost mix shifts: fast signals die of spreads, slow ones of financing.

Worked example 3: the break-even win rate

For a signal with a symmetric target and stop of T pips and a round-trip cost of c pips, a win nets T minus c and a loss costs T plus c. The win rate needed to break even is (T + c) / 2T.

Target and stop	Cost 1.5 pips	Cost 2.0 pips
5 pips	65.0%	70.0%
10 pips	57.5%	60.0%
20 pips	53.8%	55.0%
50 pips	51.5%	52.0%

A 10-pip scalping signal that wins 55% of the time sounds good and has a genuine gross edge of 1 pip a trade (0.55 x 10 minus 0.45 x 10). After 1.5 pips of costs it loses 0.5 pips a trade, or $5,000 a year at 1,000 round trips and one lot. It needs 58% to net 0.1 pips a trade ($1,000 a year) and 60% to net 0.5 pips ($5,000). Small differences in cost assumptions swing the result from profit to loss, which is why a signal seller’s gross results tell you very little. Win rate on its own is misleading for the reasons covered in why win rate is misleading.

Worked example 4: costs measured in R

A cleaner way to compare signals with different stops is to express the round-trip cost as a fraction of the amount risked, R. A trade risking a 10-pip stop and paying 1.5 pips in costs gives up 0.15R every time, win or lose.

Stop distance	Cost per trade at 1.5 pips	Cost per trade at 2.0 pips
5 pips	0.30R	0.40R
10 pips	0.15R	0.20R
20 pips	0.075R	0.10R
50 pips	0.03R	0.04R
100 pips	0.015R	0.02R

This is the right unit for comparing costs with edge. As our deep dive on the statistical significance of track records notes, a gross expectancy of 0.10R to 0.20R per trade would already be a strong result for a forex strategy. A 10-pip scalping signal therefore spends most or all of a strong edge on costs before it starts, while a 100-pip swing signal spends a small fraction of it. The same arithmetic explains the trading-rule literature: the fastest rules failed after costs first, because their gross profit per trade was smallest relative to the cost of each switch.

It also shows why the frequency of trading and the size of the stop matter more than the headline spread. Halving the spread from 1.2 to 0.6 pips saves 0.06R per trade on a 10-pip stop. Widening the stop to 20 pips, with the position size cut in half to keep the dollar risk constant, halves the cost per R without touching the spread at all, although it changes the strategy itself.

The whole calculation fits on one line. Annual cost as a share of the account is approximately:

(round trips a year x lots per trade x round-trip cost per lot + nights held a year x lots x financing cost per night) / account size

To fill it in, you need four numbers that any honest signal service can supply: how many trades it sent over the past year, the typical position size it assumes relative to account size, the average holding period, and the pairs traded. You need two numbers from your own dealer’s current disclosures: the average (not minimum) spread or commission for those pairs, and the rollover rates including markup. Then compare the result with the signal’s claimed gross return. If the provider reports results in pips without stating whether spreads were deducted, assume they were not and subtract them yourself. A claimed 40 pips a week from 20 trades is 2 pips a trade gross, which leaves about 0.5 pips a trade at our 1.5-pip cost, a quarter of the headline.

What can go wrong

Variable spreads at the wrong moment. Spreads widen at the daily rollover, around major data releases and in thin holiday markets. A stop triggered when the spread is wide fills at the worse side of it.
Stops slip more than limits. A stop becomes a market order in a falling market. Weekend gaps can fill stops far beyond their level.
Delayed signals. A follower who acts a minute after the alert pays the move that happened in that minute, every time, on top of the spread.
Hidden financing. Swap markups are charged daily and appear as small line items. On slow strategies they can exceed trading costs.
Currency conversion. Profits and losses in yen or Swiss francs are converted to the account currency, sometimes at a markup.
Cost changes. Commission, markup and spread schedules are set by each dealer and change over time; check the current disclosure, not a review site.
Incentives toward volume. When a signal provider earns per lot traded, frequent trading pays the provider whether or not it pays you.

The short version

Retail FX costs are spread, commission, overnight financing and slippage. For one standard lot of EUR/USD the trading part is roughly 1.5 pips a round trip in our assumptions, small in price terms but large relative to a 5- or 10-pip target. Research going back to Levich and Thomas (1993) shows costs falling from about 10 to about 2 basis points per switch for institutions, yet still deciding whether fast strategies survive; Menkhoff and coauthors found the full quoted spread cut currency momentum returns from about 10% to about 4% a year. In our worked example, a four-trades-a-day signal costs 153% of a $10,000 account a year and needs a 57.5% win rate on 10-pip trades to break even. Estimating costs before following any signal is basic arithmetic; this article explains it and does not recommend any dealer or account type.