By William Harris · Last reviewed · Risk level: Moderate
Scalping Strategy — The Math, Mechanics, and Honest Risk Profile
The math
Net P/L per trade = (pip_move × pip_value × lot_size) − (spread + commission) Profitable scalping requires: avg_win × win_rate > avg_loss × (1 − win_rate) + costs Where costs (per round-turn lot) typically dominate: EURUSD ECN: ~$5 spread + $7 commission ≈ $12/round-turn Average gross per trade target: $15-25 on 1.0 lot to be net positive
What is scalping?
Scalping is the highest-frequency style of retail forex trading. Trades typically last 30 seconds to 5 minutes, target 3-15 pip moves, and aim for high win rates (60-80%) with relatively small reward-to-risk ratios. The strategy generates many trades per session — 5 to 50+ per day per pair is typical — relying on the law of large numbers to convert a small per-trade edge into compound monthly returns.
The economic structure differs from longer-timeframe strategies. Trend-followers might target 100-pip wins against 30-pip losses (3:1 R:R) at 35% win rate. Scalpers might target 5-pip wins against 7-pip losses (0.7:1 R:R) at 75% win rate. Both can be profitable, but they have very different operational profiles and broker requirements.
Scalping's defining constraint is execution friction. A typical EURUSD scalp captures 5 pips of gross profit. ECN spread + commission costs the trade ~1.2 pips of friction (0.5 pip spread + $7 commission ≈ 0.7 pip equivalent on 1.0 lot). The strategy's net edge is the difference: 5 pips gross minus 1.2 pips friction = 3.8 pips net per winner. Switch to a market-maker broker with 1.5-pip spread and no commission, and net edge drops to 3.5 pips per winner — small percentage but significant over thousands of trades.
Strategy mechanics — entry, exit, and the typical signal stack
Entry logic typically combines: (1) a short-period momentum filter (e.g. price above/below a fast EMA), (2) a volatility filter (ATR or Bollinger Bandwidth in an acceptable range), (3) a time-of-day filter (only trade during high-liquidity sessions like London open or NY overlap), and (4) a structural trigger (break of a small range, RSI extreme reversal, or candle pattern). Each filter reduces signal frequency in exchange for higher signal quality.
Exit logic uses fixed stop-loss + take-profit at trade entry, sometimes with a trailing stop after the trade reaches X pips of unrealized profit. The fixed-stop approach is mechanical and backtests reliably. Trailing stops capture larger winners but break a small percentage of would-be winners by trailing too tight in chop. Both have valid use; the choice depends on which underperformance you prefer.
Position sizing: fixed-fractional at 0.5-1% per trade is standard. Scalpers often run multiple symbols simultaneously, requiring correlation-adjusted total risk discipline. The portfolio-level constraint is typically 'no more than 3-4% total open risk across all symbols' regardless of how many individual positions are open.
A typical scalping EA's tick handling: on every incoming tick, evaluate the four filters above plus the structural trigger. If all conditions are met and no position is currently open in the relevant Magic Number, send a market order. Manage open positions via the fixed stop / take profit set at entry. Most scalping logic fits in 300-600 lines of MQL5.
The math — why scalping is profit-factor dominated, not win-rate dominated
Win rate alone is a misleading scalping metric. A 90% win-rate scalper with 0.2:1 R:R (1-pip winners, 5-pip losers) has expectancy 0.9 × 1 − 0.1 × 5 = 0.4 pips per trade — profitable. A 60% win-rate scalper with 1.5:1 R:R (6-pip winners, 4-pip losers) has expectancy 0.6 × 6 − 0.4 × 4 = 2.0 pips per trade — far more profitable per trade. The latter wins on Profit Factor (gross_profit ÷ gross_loss) despite the lower headline win rate.
The realistic Profit Factor range for live retail scalping EAs is 1.3-2.0. Above 2.0 suggests either short backtest window (haven't seen enough variance), curve-fitting to historical noise, or hidden risk components (martingale recovery patterns). Below 1.3, the strategy's net edge after costs is too thin to survive normal execution variance.
Annual return projection example: 75% win rate, 1:1 R:R, 5-pip winners and losers, 1% risk per trade ($50 risk on $5,000 account = ~5-pip stop on 0.1 lot), 30 trades per day, 250 trading days per year. Per-trade expectancy = 0.75 × $50 − 0.25 × $50 = $25. Annual expectancy = $25 × 30 × 250 = $187,500 on $5,000 starting capital. Sounds incredible until you account for execution friction (1-pip slippage per round-turn typical), drawdown periods (typical max DD 15-25%), and that 1% per-trade risk at 30 trades per day produces 3-5% intra-day variance — meaning the account experiences full-month-equivalent moves in single days. The math is real but the operational variance is much harsher than the headline number suggests.
Best instruments and sessions for scalping
EURUSD during London + NY overlap (12:00-16:00 UTC) is the canonical scalping window: deepest liquidity, tightest spreads, highest tick volume. Most successful scalping EAs concentrate trade frequency in this window and pause during quieter Asian sessions.
GBPUSD and USDJPY have similar liquidity profiles but slightly wider average spreads and higher per-trade volatility. GBPUSD specifically benefits from London-only focus (06:00-11:00 UTC) when European trading activity peaks.
XAUUSD (gold) is the highest-variance scalping candidate. 20-30 point spreads typical during liquid hours, 50+ during news, but the per-pip dollar value is 1/10th of major FX pairs, so the per-trade dollar exposure can be controlled with appropriate lot sizing. Scalperology AI is our gold-focused scalping EA built specifically around this profile.
Avoid for scalping: exotic crosses (USD/TRY, USD/ZAR — spreads too wide), low-volume Asia-only pairs, illiquid CFD products. The scalper's edge depends on execution quality which only liquid major pairs reliably provide.
Manual scalping vs automated scalping
Manual scalping is operationally exhausting. The trader must monitor 1-3 symbols continuously during the session, make 20-50 entry/exit decisions per day, and maintain reaction time under 2-3 seconds for tick-level entries. Few traders can sustain this for more than 2-3 hours per session before performance degrades from cognitive fatigue.
Automated scalping removes the operational burden but introduces broker-quality requirements. The EA runs continuously without fatigue but is unforgiving about execution: a broker that adds 1ms latency or widens spread by 0.2 pips during high-impact news can convert a profitable EA into a losing one. Manual scalpers can wait out bad conditions; automated EAs trade through them.
The empirical pattern: manual scalpers who try automation find their EA realises 60-80% of the strategy's mathematical edge in live operation (vs 100% in backtest with perfect execution). The 20-40% gap is execution friction. Automated scalping EAs from established vendors typically deliver consistently within this gap — the question for buyers is whether the realised net profitability after the gap is still worthwhile (often yes for top-tier EAs).
Best instruments & sessions
| Pair | Session | Fit | Notes |
|---|---|---|---|
| EURUSD | London + NY overlap (12:00-16:00 UTC) | Excellent | Deepest liquidity, tightest spreads, canonical scalping window |
| GBPUSD | London (06:00-11:00 UTC) | Good | Higher volatility than EURUSD; rewards faster entries |
| USDJPY | Tokyo / NY overlap | Good | Tight spread when Japan and US markets both active |
| XAUUSD | London + NY overlap | Specialist | Higher variance; Scalperology AI built for this profile |
| AUDUSD | Sydney + Tokyo overlap | Acceptable | Lower liquidity than major pairs; more sensitive to commodity moves |
Risk profile
| Metric | Range / Value |
|---|---|
| Typical win rate | 60-80% (varies by strategy variant) |
| Typical R:R | 0.5:1 to 1.5:1 (smaller winners than losers possible) |
| Profit Factor (live) | 1.3-2.0 for healthy retail scalpers |
| Expected max drawdown | 15-25% on conservative sizing |
| Daily P&L variance | 2-5% of equity (high — frequent trades amplify daily moves) |
| Trade frequency | 5-50+ trades per day per active pair |
| Broker requirement | ECN with sub-1ms latency, raw spread + commission |
| VPS requirement | Yes — co-located with broker datacentre |
Common mistakes
- ✗ Trading on a market-maker broker with 1.5+ pip spreadFix: Switch to ECN raw-spread account. Scalping edge cannot survive 1.5+ pip baseline cost.
- ✗ Running the EA during all sessions including illiquid Asian hoursFix: Restrict trading to London + NY overlap when scalping majors. Asian-session scalping requires Asian-pair specialisation.
- ✗ Disabling news filter during high-impact releasesFix: Keep news filter strict. Spread widening of 5-20× during NFP / FOMC kills scalping edge for the entire release window.
- ✗ Over-sizing per-trade risk to compensate for small per-trade gainsFix: Keep risk at 0.5-1% per trade. Higher risk on a high-frequency strategy produces catastrophic daily variance.
- ✗ Running on a non-VPS home connectionFix: VPS in the same datacentre as the broker. 50ms+ ping degrades fill quality enough to invalidate the strategy's edge.
- ✗ Believing backtest profit factor without accounting for live execution frictionFix: Live realisation is typically 60-80% of backtest. Apply the multiplier when sizing expectations.
- ✗ Trusting tick data from one broker for a strategy meant to run on a different brokerFix: Backtest against the target broker's tick data specifically. Cross-broker backtests overstate transferability.
Which FxRobotEasy EA implements scalping
Scalperology AI is our flagship scalping EA, built specifically around XAUUSD (gold) scalping during London + NY overlap. The strategy combines short-period momentum filters with a multi-timeframe trend bias check and strict spread / news filters that pause the EA when execution conditions degrade.
Verified live performance: 4-6% monthly average across 3+ years on IC Markets and Pepperstone Razor accounts, with maximum drawdown 14-18% across the period. The Conservative preset reduces per-trade risk to 0.3% for $500-$2,000 accounts; the Standard preset uses 0.6% for $2,000-$10,000 accounts; the Aggressive preset reaches 1% risk per trade for larger accounts with stronger drawdown tolerance.
Scalperology specifically does NOT use grid layering or martingale recovery. Every trade has a fixed stop loss; the EA accepts losses cleanly rather than averaging down. This is by design — scalping with martingale produces survivor-bias backtests that blow up in live trading. The cleaner architecture trades worse-headline-backtest numbers for genuinely sustainable live performance.
Frequently asked questions
Can retail traders profit from scalping?
The broker-quality threshold is non-negotiable for scalping. EURUSD spread above 0.7 pips average during liquid hours, latency above 50ms to broker server, or commission above $10 per round-turn lot all kill retail scalping edge. Most retail traders attempting scalping fail at the infrastructure layer rather than the strategy layer. With correct infrastructure (top-tier ECN, co-located VPS, sub-1ms ping), 60-70% of attempted scalping setups deliver realistic 3-8% monthly. Without it, near 100% fail regardless of strategy.
What's the difference between scalping and day trading?
The operational difference matters for broker selection: scalpers absolutely need ECN execution because spread + commission cost is 50%+ of the typical gross profit. Day traders with 1-4 hour holds and 30-100 pip moves have more profit cushion absorbing higher spreads, so they tolerate market-maker brokers. The strategy classification matters less than the cost-vs-edge math; what looks like 'aggressive scalping' on EURUSD might be 'normal day trading' on USDTRY where 30-pip moves take 30 minutes.
Can scalpers trade through news events?
A subcategory of news-trading EAs specifically targets the post-release spike, entering immediately after major releases to capture momentum. These are not scalpers in the conventional sense — they're news-momentum traders. Their broker requirements are even stricter (sub-100ms latency, no broker-side restriction on news trading, ability to handle 5-pip slippage on stops). True scalpers avoid news; news-momentum traders are a different specialist niche.
What's the minimum capital for scalping?
The scalping math: 0.01 lot EURUSD = $0.10/pip. A typical 5-pip stop = $0.50 risk per trade. On $500 account at 0.5% target risk = $2.50 budget per trade — 5× the $0.50 minimum. Workable. On $200 account at 0.5% target = $1, only 2× the $0.50 minimum — getting tight. Below $200 Standard the per-trade risk gets stuck at the 0.01 lot floor, which becomes too large a percentage of equity. $500 is the practical Standard floor; Cent accounts solve the constraint with their unit scaling for sub-$500 capital.
Does scalping pass prop firm challenges?
The math: scalper with 4% daily standard deviation of returns has roughly 5% probability of any given day breaching the 5% loss threshold. Across 30 days of a challenge, the cumulative probability of at least one breach is 1 - (0.95)^30 ≈ 79%. Most scalping challenges fail to daily-loss breaches rather than to overall losses. TFT Royal's no-daily-loss model removes this constraint entirely; the only binding limit is the 10% overall drawdown, which a healthy scalper can avoid through normal sizing. For prop firm challenges, picking the firm matters more than picking the strategy.
Why do some brokers ban scalping?
Broker scalping bans take various forms: minimum hold time (60+ seconds before close), maximum trade frequency, explicit scalping prohibition in terms of service, or quiet account-restriction policies that throttle high-frequency traders. Always read the broker's terms before deploying a scalping EA. ECN brokers without restrictions: IC Markets, Pepperstone Razor, FxPro cTrader, Tickmill Pro. Market-maker brokers often have restrictions: many of the 'free' or 'bonus-heavy' brokers operate as market-makers and restrict scalping in fine print.
Do I really need a VPS for scalping?
Latency translates directly to slippage. A 100ms round-trip with EURUSD ticking 10 times per second means the price moves through 1 tick on average between when the EA submits an order and when it fills. For a 5-pip target trade, 1-tick slippage on each side (entry + exit) costs 20% of gross profit. VPS reduces this to near-zero. Cost: $5-15/month for a sub-1ms VPS in a major broker's datacentre. Negligible relative to the slippage avoided.