In early 2024 I was scalping GBPUSD on a 5-minute system with a 62% win rate on demo. Moved it to live with a well-known ECN broker, and within three weeks my win rate had collapsed to 47% — same entries, same exits, same me. I blamed psychology. I blamed the VPS. I blamed Mercury retrograde. Then I pulled my execution logs and found the real culprit: 23% of my orders during the London open were being rejected or requoted, and the fills I did get were slipping 0.8 pips on average against me. My broker's liquidity providers were exercising "last look" — a 100–300 millisecond window where they could look at my order, check if the market had moved, and reject it if the move favored me. Winners got rejected; losers got filled. My edge wasn't broken. It was being taxed at the fill level, invisibly, trade after trade. Understanding last look vs no-last-look brokers changed my broker selection more than any spread comparison ever did.
Feeding the Signal Calculator a last-look reality check
Here's the thing most traders miss: your position-size math assumes the fill you planned. Last look means the fill you planned and the fill you got are different numbers — and the difference compounds. I now run every scalping plan through the calculator twice: once with advertised costs, once with measured costs. The field meanings shift slightly for this exercise.
| Input Field | What it means | Sample value |
|---|---|---|
| Symbol | The pair you're scalping — majors suffer last look most | GBPUSD |
| Direction | Your scalp direction; rejection rates differ slightly by side in fast markets | Long |
| Entry price | Your signal level — then add measured slippage for the realistic version | 1.2710 (+0.8 pip slippage) |
| Stop-loss distance | Scalp stops are tight — which is exactly why fill quality dominates | 8 pips |
| Take-profit distance | Typical scalp target; small targets amplify cost drag | 12 pips |
| Account balance | Current equity | $7,500 |
| Risk per trade | 0.5% — scalps get more attempts, so each risks less | 0.5% |
The calculator said 4.69 lots for the advertised-cost version. Then I re-ran it mentally with the measured reality: 0.8 pip slippage on an 8-pip stop and 12-pip target doesn't just cost money — it warps the R:R from 1:1.5 to roughly 1:1.3. That's the difference between a working system and a donation scheme. The DarkWolf MT5 Signal Calculator can't measure your broker's last look for you, but it forces you to type honest numbers instead of brochure numbers.
Last look vs firm liquidity: how your fill actually happens
Let me demystify the plumbing, because once you see it, you can't unsee it. When you click buy on an ECN/STP broker, your order typically goes to one or more liquidity providers — big banks and non-bank market makers. With firm liquidity (no last look), the price you see is the price you get: the LP's quote is binding, and your fill happens in single-digit milliseconds. With last look, the LP gets a final peek — usually 100 to 300 milliseconds — after receiving your order. If the market moved in your favor during that window, they can reject the trade ("price moved, try again"). If it moved against you, they fill it happily. Guess which happens more.
| Feature | Last look | No last look (firm) |
|---|---|---|
| Quote binding? | No — LP can reject within the hold window | Yes — displayed price is executable |
| Typical hold time | 100–300 ms (sometimes more in news) | None — fills in low single-digit ms |
| Rejection pattern | Skewed: winning entries rejected more often | Rare; usually only on genuine price gaps |
| Spreads | Often tighter on screen (the bait) | Slightly wider, but honest |
| Who it hurts | Scalpers, news traders, latency-sensitive EAs | Nobody structurally — costs are visible |
| Who it barely touches | Swing traders holding days | — |
The dirty secret: last-look venues often advertise the tightest spreads in the industry. Of course they do — the spread is only half the price you pay. The CFTC has taken enforcement actions over abusive last-look practices in FX, which tells you this isn't a conspiracy theory — it's a documented market-structure issue. The FCA likewise requires firms to disclose execution practices, though you'll need to dig through order-execution policies to find the actual hold times. Nobody puts "we reject your winners" on the homepage.
Living with last look: what I changed about my trading
I didn't switch brokers immediately — switching has its own costs, and I wanted to quantify first. So I adapted. Change one: I stopped trading the first 15 minutes of London on the last-look account. That's when rejection rates peaked (that 23% figure), because volatility gives LPs maximum option value on their hold window. Change two: I widened my scalp targets from 10 to 15 pips. Bigger targets dilute the per-trade cost drag — slippage is a fixed tax, so larger gross profits carry it better. Change three: I moved limit orders instead of market orders wherever the strategy allowed. A limit order at my price either fills at my price or doesn't fill; it can't be slipped against me, though it can still be rejected.
Change four was the big one: I split my trading. Scalping and news-adjacent strategies moved to a no-last-look venue with slightly wider spreads but honest fills. Swing trades stayed on the last-look broker, where the tight spreads genuinely help and the hold window is irrelevant over a 200-pip target. This split-broker setup sounds fussy, but it matches the tool to the job. My scalping broker guide goes deeper on what to test during a broker trial — including the execution-log analysis that exposed my 23%.
The true cost of a last-look fill, in numbers
Let's put algebra on the damage so you can compute it for your own broker. The effective cost of a trade isn't the spread — it's the spread plus slippage plus the hidden cost of rejections (missed winners you had to chase at worse prices):
| Variable | Meaning | My measured value |
|---|---|---|
| Spread | Advertised round-trip spread on the pair | 0.4 pips (GBPUSD) |
| Avg Slippage | Mean adverse fill difference vs. requested price, from your logs | 0.8 pips |
| Reject Rate | Share of orders rejected or requoted | 23% at London open, 6% off-peak |
| Chase Cost | Extra pips paid re-entering after a rejection | ~1.5 pips average |
| Effective Cost | True per-trade friction | 0.4 + 0.8 + (0.23 × 1.5) = 1.55 pips |
Read that last number again: 1.55 pips true cost vs. 0.4 advertised. On a 12-pip scalp target, friction eats 13% of every winner before you even start. My system needed a 55% win rate to break even at 0.4 pips; at 1.55 pips it needed 62%. I had 62% on demo and 47% live — and suddenly the entire mystery was just arithmetic. Measure your own numbers from MT5's trade history (right-click, save report, check fill vs. requested prices). Ten minutes of spreadsheet work beats a month of self-blame.
Testing my scalping costs through the calculator
Once I had measured numbers, I ran three scenarios through the calculator to decide what to do. Same $7,500 account, same 8-pip stop / 12-pip target GBPUSD scalp, 0.5% risk.
| Scenario | Assumed costs | Calculator verdict | Decision |
|---|---|---|---|
| Brochure numbers | 0.4 pip spread, no slippage | 4.69 lots, R:R 1:1.5 | Fantasy — ignore |
| Measured, off-peak | 0.9 pips effective | 4.69 lots, R:R ~1:1.4 | Tradeable — scalp off-peak only |
| Measured, London open | 1.55 pips effective | 4.69 lots, R:R ~1:1.3 | Skip — edge destroyed |
This table made the decision for me: keep scalping the last-look broker only in calm hours, move the aggressive sessions elsewhere. Investopedia has a solid primer on ECN vs. market-maker execution if you want the full taxonomy, and my offshore broker breakdown covers the regulatory trade-offs when you go hunting for better execution venues. If your scalping numbers look like my row three, the DarkWolf plans include execution-quality presets that bake realistic slippage into the sizing automatically.
What my rejected fills taught me about broker choice
Mistake one was mine: I chose that broker purely on spread. 0.2 pips on EURUSD! I bragged about it. I never asked about last look, hold times, or rejection rates — the questions that actually determine scalping profitability. Spreads are the sticker price; execution is the out-the-door price. Now my broker trial checklist has five items: advertised spread, measured slippage over 50 trades, rejection rate at the session open, behavior during one news event, and withdrawal speed. A broker can fail the spread test and pass everything else and still be the better choice.
Mistake two: I ran the same strategy on demo and assumed live would match. Demo fills are fantasy — no last look, no real liquidity constraints, instant fills at mid. My 62% demo win rate was never real; it was a backtest with extra steps. Now I forward-test every strategy on a small live account for at least 100 trades before sizing up. The $200 I risk finding out is tuition; the $2,000 I used to risk finding out was something else.
The human-value quirk I'll leave you with: check your broker's order execution policy for the phrase "asymmetric price improvement." That's the polite term for the LP keeping favorable moves and passing through unfavorable ones — last look's business model in a suit. If you find it, you're not being paranoid. You're reading.
A $100,000 scalping week: last-look broker vs no-last-look
Same system, same week, split across two accounts — $100,000 each, 0.5% risk per scalp ($500), GBPUSD 8-pip stops / 12-pip targets, 40 trades each. The only variable was the venue.
| Metric | Last-look broker | No-last-look broker |
|---|---|---|
| Advertised spread | 0.2 pips | 0.6 pips |
| Avg slippage | 0.8 pips adverse | 0.1 pips |
| Rejection rate | 19% | 2% |
| Win rate | 48% | 61% |
| Week P/L | −$1,840 | +$2,260 |
Same trader, same system, same week: a $4,100 swing explained entirely by execution quality. The "expensive" broker with 0.6-pip spreads made money; the "cheap" one lost it. I've never seen a clearer demonstration that for scalpers, the broker IS the strategy's largest variable cost. If you scalp, run this experiment yourself for one week on demo-vs-live or broker-vs-broker. The numbers will change how you shop for brokers permanently.
Last look broker questions, answered
What is last look in forex trading?
It's a practice where a liquidity provider gets a brief window — typically 100–300 milliseconds — after receiving your order to decide whether to fill or reject it. During that window they check if the market moved. If it moved in your favor, they often reject; if against you, they fill. The result is asymmetric: your winners get filtered out while losers sail through, which quietly degrades any short-term strategy.
Is last look legal?
Generally yes, when disclosed — it's a standard institutional FX practice that migrated to retail. What crosses the line is abusive last look: excessive hold times, using the window to trade on your order information, or rejecting at rates far beyond what's needed for risk management. Regulators including the CFTC have fined firms for exactly these abuses. The practice itself isn't the scandal; undisclosed or predatory versions of it are.
How can I tell if my broker uses last look?
Pull your execution data. In MT5, compare requested price vs. fill price across 50+ trades and count rejections/requotes, especially around session opens and news. Rejection rates above 10% in volatile periods, or slippage that's consistently adverse (never in your favor), are the fingerprints. Also read the order execution policy — search for "last look," "hold time," or "asymmetric price improvement." Honest firms disclose it.
Does last look affect swing traders?
Barely. If your stop is 150 pips and your target is 300, a 0.8-pip slippage and occasional rejection are rounding errors. Last look is a scalper's problem — it lives in the world of 8-pip stops and 12-pip targets, where 1.5 pips of hidden cost is 13% of the trade's economics. I keep swing positions on my last-look broker deliberately: the tight spreads help, and the hold window is irrelevant at that scale.
What's better for scalping: ECN with last look or STP without?
For pure scalping, firm pricing usually wins despite wider advertised spreads — my split-week test showed a $4,100 difference on identical systems. But "ECN" and "STP" are marketing labels, not guarantees; what matters is measured execution: slippage, rejections, fill speed. Test with a small live account, measure for two weeks, and let the numbers pick your broker. Never choose on spread alone — that's the mistake that cost me three weeks of confusion.
Can I avoid last look entirely as a retail trader?
Mostly, yes. Several retail brokers offer no-last-look execution on majors, and futures markets (CME FX futures) have no last look at all — every quote is firm, though you'll pay exchange fees and face larger contract sizes. The realistic retail answer: use limit orders where possible, avoid the most volatile 15 minutes of each session, and put strategies with sub-20-pip targets on the venue with the best measured fills, not the best advertised spreads.