Sessions & the cost stack

BeginnerDuration ~25 min video + 20 min hands-onTools Your broker-spec.md, A calculator

Why this lesson

A strategy that looks profitable on gross numbers routinely dies on net. Spread plus swap plus slippage can quietly exceed the entire per-trade edge of an intraday system — and, worse, the costs are not constant: they spike at exactly the moments many systems trade. Itemize the stack now and you’ll never again be surprised by the gap between a backtest and a bank statement.

Watch

Three short Darwinex segments make the cost stack concrete. Watch for when costs spike, not just that they exist.

Segment: 0:00–13:00 — the rollover spread spikewatch full video

Watch for: Why FX spreads blow out for roughly five minutes every day at the 5pm ET (NY close) rollover: interest-rate-parity mechanics reset the overnight swap rate at that instant, and liquidity providers can't hedge the unknown new rate, so they quote defensively wide. This is the daily, predictable version of session-driven cost. The tail of the video is company Q&A — skip it.

Segment: first ~23 min — slippage from first principleswatch full video

Watch for: Slippage has two structural causes: latency (the price you see is already stale) and finite liquidity (a price only exists for a given volume). Crucially, an MT4/MT5 'stop-loss' or 'limit' is converted to a market order once triggered, so it is never truly guaranteed — the same point from Lesson 1.2, now with the order-book reasoning. The last ~10 minutes are unrelated company Q&A.

Segment: ~29:39–31:45 — a real session-end blowoutwatch full video

Watch for: A September 2017 case study: a large market order three seconds before the session close swept the available depth and triggered a ~100-pip spike that chained into stop-outs. A concrete worked example of why thin-liquidity, session-end periods blow out spreads and hunt stops — systemic to OTC fragmentation, not any one trader's fault.

The explainer

The market trades around the clock but not evenly. Three sessions dominate: Tokyo (Asian), London (European), and New York. Liquidity — and therefore tight spreads — concentrates when the big centers overlap, above all the London/NY overlap in the early US morning; it thins out in the Asian session and around the daily rollover (forex trading sessions, EUR/USD session behavior). This matters because your costs are session-dependent, so the first thing to internalize is that “the spread” is not one number.

Cost 1 — spread (variable). On a good raw-spread account, EUR/USD runs about 0.1–1.0 pips during London/NY liquidity, but the spread is a variable spread: it widens materially in Asian hours and blows out 10–20+ pips at the 5pm ET rollover as liquidity providers reset (rollover spread behavior). You pay the spread on every entry, embedded in the bid/ask gap from Lesson 1.1. Here is the trap that makes this a whole module: a breakout system, by design, fires when volatility jumps — and volatility spikes are exactly when spreads widen and liquidity thins. The system trades precisely when the trade is most expensive. That correlation between signal and cost is why a naive backtest using one favorable spread number lies.

Cost 2 — commission. On a raw A-book account you pay a per-trade fee, typically $3–$7 per round-turn standard lot (broker execution models). “Round-turn” means the fee covers both opening and closing, so watch whether a broker quotes it per side (double it) or per round turn. This is a commission per round-turn lot, flat regardless of how the trade goes.

Cost 3 — swap (rollover). Hold a position past the daily rollover and you pay or receive interest based on the rate differential between the two currencies — the swap or rollover. On majors it’s roughly −$3 to −$10 per standard lot per night depending on direction, and it is tripled on Wednesday to pre-pay the weekend’s settlement (the triple-swap Wednesday) (FP Markets rollover guide). A swing system holding trades for days must model swap per instrument and per direction — sometimes you collect it, usually on a CFD account the markup means you bleed it.

Cost 4 — slippage. As Lesson 1.2 established and the video above reasons out, your fill drifts from your intended price. It’s near zero (0–0.5 pips) for retail-size market orders on majors in liquid hours, but it leans negative on stops — 57% of stop orders got negative slippage in FXCM’s published stats (FXCM slippage statistics) — and around news, 10–20 pips is normal.

Put the four together and you have the whole retail cost stack. The reason it gets its own module is the distinction it forces: gross vs net. Gross results are what the price chart alone would give you — entry to exit, no frictions. Net results subtract spread, commission, swap, and slippage. Every honest evaluation from here reports net. A system with a small gross edge and a large cost stack has no edge, and the only way to know is to itemize. That itemization is a per-instrument, per-session table — which is exactly what you’ll build in code in Lesson 1.5. First, build it by hand once.

Do it

Cost out a single realistic swing trade, in full, on paper.

1. Set the trade: long 1 standard lot EUR/USD, entered during the London/NY overlap, held 4 nights including a Wednesday, exited via a stop. Pull your broker’s numbers from broker-spec.md; use these defaults where you lack a figure.
2. Spread cost: assume a 0.8-pip spread at entry. At $10/pip on a standard lot, spread cost = 0.8 × $10 = $8.
3. Commission: $6 per round-turn standard lot = $6.
4. Swap: −$5/night × 4 nights, but one of those nights is Wednesday (×3), so effective nights = 3 normal + 1 triple = 6 swap-units. −$5 × 6 = −$30.
5. Slippage: the exit is a stop, so assume 2 pips of negative slippage = 2 × $10 = $20.
6. Total cost stack = 8 + 6 + 30 + 20 = $64. Now the punchline: at $10/pip, $64 is 6.4 pips. Write this sentence in your notes: “This trade must clear 6.4 pips before I make a single cent.” If the system’s average winner is 15 pips, nearly half the edge is gone before variance even shows up.
7. Repeat step 2 with the rollover spread (say 12 pips) instead of 0.8, to feel what happens if the system trades at the wrong minute. The stack more than doubles.

Save this as cost-stack-by-hand.md. Lesson 1.5 automates exactly this.

Terms introduced

Terms introduced in this lesson

• sessions (Tokyo/London/NY)
• variable spread
• commission per round-turn lot
• swap/rollover
• triple-swap Wednesday
• gross vs net

Each term links to the glossary, where it is defined by the way you just used it.

Check

Check yourself

1. EUR/USD spreads on a raw-spread account are typically tightest during:

   the London/NY overlapthe 5pm ET rolloverthe Asian (Tokyo) sessionthe weekend

2. Swap (rollover) on a standard lot of a major pair is roughly, per night:

   −$100 or more−$3 to −$10 depending on directionalways positivezero on all pairs

3. Swap is charged at triple size once a week — on Wednesday — because:

   Wednesday is the most volatile daybrokers make extra profit midweekit covers the weekend's settlement in advanceit resets the leverage limit

4. Commission on a raw A-book account is typically quoted as:

   a flat monthly feea percentage of profitper round-turn lot (roughly $3–$7 per standard lot)per pip of movement

5. A breakout system is especially exposed to cost because it tends to fire:

   in the calmest, tightest-spread hoursonly on weekendswhen commission is waivedexactly when volatility rises and spreads widen

6. The difference between gross and net results is:

   gross is after costs, net is beforenet is gross minus spread, commission, swap and slippagenet includes only commissionthey are the same thing

Check answers

You can move on when you can… itemize every cost on one trade — spread by session, commission, swap (and triple-swap Wednesday), slippage — and explain why breakout systems fire exactly when spreads widen.

Mark this lesson complete

Go deeper

• FP Markets rollover/swap guide and FXCM’s published slippage statistics — real numbers, not round-figure guesses.
• The Darwinex channel is the rare broker that teaches transaction costs honestly; the three segments above are the core, and the full videos reward a second watch once the vocabulary sticks.
• The book that owns this topic: Stefan Jansen’s Machine Learning for Algorithmic Trading ch. 18 (transaction costs) is the most systematic written treatment — you don’t need the ML to use the cost chapter.