Scalping expert advisor: the stats of a forex scalping EA and why spread decides everything

What a scalping expert advisor does

A scalping expert advisor aims to capture a few pips, many times a day, and to be out of the market within minutes. It trades the noise rather than the direction: small pullbacks during a quiet session, the rebound after a spike, the tick-level imbalance around round numbers. Targets of 3 to 10 pips, stops of similar size or a little wider, and position durations from seconds to an hour. Most scalping forex robots on MT4 and MT5 trade the Asian session or the late New York afternoon when ranges are tight and spreads on a good broker are tightest.

The whole category is a cost arbitrage. The strategy's edge per trade is measured in pips; so are the spread, the slippage and the commission. The same trading bot with the same signals makes money at one broker and loses it at another, and the difference is a fraction of a pip.

The statistics you will see

Honest scalpers show a win rate of 55–75%, an average win close to the average loss (payoff ratio 0.7 to 1.3), and a profit factor of 1.1 to 1.5. Expectancy in currency is small, a few dollars per 0.1 lot, and it only becomes meaningful because the EA trades 40 to 150 times a week. The review classifies any average hold under one hour as Scalping, and the Activity distributions histogram fills the under-1-minute, 1–5 minute, 5–15 minute and 15–60 minute buckets with almost nothing beyond.

The equity curve is a gentle upward slope with small, frequent drawdowns of 2–8% that recover in days. Deep drawdowns are unusual unless the EA has a hidden loss-holding mode. The Z-score is usually near zero, since consecutive scalps are close to independent. The hour chart is sharply peaked on the traded session, and the weekday chart is often weak on Friday. Open positions are rarely visible at all, because nothing stays open long enough for a viewer to see it; when one does appear it is a single small position.

The number that matters most is not on the Trade statistics card. It is in the Execution quality section: typical spread in points for the scalped symbol, at the hours the EA trades, alongside signed open and close slippage.

A worked example: the cost ratio

The EA averages 4.0 pips per winning trade on EURUSD. The broker's spread at 02:00 UTC is 1.2 pips, and the Execution panel shows average slippage of 0.3 pips on entry and 0.3 on exit. Round-trip cost is 1.8 pips, 45% of the average winner, before commission. Move the same EA to a broker whose overnight spread is 2.0 pips and the cost becomes 2.6 pips, 65% of the winner, and a profit factor of 1.3 turns into something under 1.0. Nothing about the strategy changed.

Broker cost audit explains the rule of thumb: when round-trip costs exceed about a third of the average win, the result belongs to the broker, not the EA. For a scalper, that line is usually close.

Advantages

• Losses are small and frequent rather than rare and enormous, so the worst case is a slow bleed you can see coming.
• Hundreds of trades arrive in weeks, so the statistics become meaningful quickly and Monte Carlo in Risk Lab has a real sample to work with.
• Minimal overnight exposure: no swap to speak of, no weekend gaps.
• Easy to evaluate with Custom Analysis, because filtering by hour or symbol gives large subsets.

Disadvantages and failure modes

Fragility to costs is the main one, and it is not a tail risk: it is present on every trade. Brokers can and do widen spreads, add slippage or restrict scalping on accounts that are consistently profitable at it. The second failure is volatility regime change. A scalper tuned to a 30-pip Asian range gets stopped repeatedly when that range becomes 80 pips, and the small, frequent losses become a steady decline over weeks.

The third failure is the disguised one. Some "scalping" EAs take a small profit fast but hold a losing trade until it comes back, sometimes adding to it. The closed-trade stats then show a scalper's win rate with a swing trader's average loss, and the holding-time histogram grows a tail into the 1–7 days bucket. That EA is a mean reversion or grid system with a scalping label, and the average-loss-to-average-win ratio gives it away.

What the AI review and Score tend to flag

Strategy type Scalping, typically with the informational fully automated flag. Concentration is the common watch flag, since many scalpers trade one pair; above 85% on a single symbol it appears. The tail flag should not appear on an honest scalper; if it does, read the previous section again. Stop-loss usage varies: some scalpers manage exits on time or by the opposite signal, so a low stop-loss percentage is a question to ask rather than an automatic rejection, but under 15% raises the high-severity flag either way.

The Score reaches its longevity trade-count threshold fast and may look mature at three months. Profitability is capped by a modest profit factor. Risk control is usually fine. What limits the Score is consistency when a month or two turns red after a spread change.

Red flags when buying a scalping EA

• Demo-only, or a Real account at a broker you have never heard of.
• Average win under 3 pips on a symbol whose Execution panel shows a spread over 1 pip.
• Positive average slippage. Real brokers do not pay you to trade; demos and some manipulated feeds do.
• A holding-time histogram with a tail into days alongside a 90% win rate.
• A vendor who says the EA works at any broker. It does not, and an honest one names the broker and account type.

How to check on an account page

Trade statistics: win rate, payoff ratio, average trade length under an hour, and trades per week. Execution quality: pick the most-traded symbol, read the typical spread at the EA's hours and the signed slippage, and compute the cost ratio against the average win in pips. Activity distributions: confirm the holding-time buckets stop at 15–60 minutes and the hour chart matches the vendor's claimed session. Monthly returns calendar: a scalper should be green most months with small numbers; one giant month is suspicious. By symbol shows whether the edge exists on more than one pair. On deeper plans, Risk Lab's Monte Carlo is genuinely useful here because scalps are close to independent, so the expected maximum drawdown and the −25% ruin rung mean something. If the owner allows Custom Analysis, filter to the most-traded hour and check the edge survives on its own. Then compare with the news-driven cousin in news trading EA and the broker-hostile extreme in arbitrage and latency EA.