Scalping is trading tiny moves on 1–5 minute charts, holding positions for minutes and finishing every session flat. It compresses all of trading — analysis, execution, risk, psychology — into its fastest, least forgiving form. That makes it terrible as a first style learned live, and outstanding to train in a simulator, where a full session of reps costs nothing.

The math that decides everything: costs

Before any setup, the arithmetic most scalping guides skip. If your average winner is 0.4% and round-trip fees plus slippage cost 0.15%, costs consume nearly 40% of every win. Scalping edges are small; fees eat small edges.

Practical consequences: scalp only liquid majors with tight spreads (BTC, ETH on top exchanges), use limit orders where possible, and demand setups with at least 1.5R after costs. Run your backtest numbers net of fees — a strategy that's profitable gross and negative net is just exercise.

One setup, repeated: the pullback scalp

Scalpers don't need ten setups; they need one they can execute half-asleep. A robust choice — the trend pullback:

• Timeframe: 5m for context, 1m for entry
• Setup (long): clear 5m uptrend — price above rising VWAP and 20 EMA
• Entry trigger: 1m pullback to VWAP or the 20 EMA that prints a rejection — a hammer or engulfing candle — enter on its close
• Stop-loss: below the rejection candle (typically 0.1–0.3% away)
• Target: 1.5–2R — at the recent high or the next visible level; take it and reset
• Session stop: 3 losses in a row = done for the session, no exceptions
• Short: mirror below falling VWAP

Why the session-stop rule is part of the strategy

Scalping's speed means tilt compounds within minutes: one revenge trade becomes five before you'd have finished your coffee. Losses cluster when the market regime stops fitting your setup — chop after a trend, news spikes. The 3-loss stop isn't psychology fluff; in backtests it measurably improves expectancy by cutting exactly those regime-mismatch streaks.

How to practice scalping in replay

Replay is the only sane way to learn scalping — live practice burns money during the (long) incompetent phase, and demo-live is too slow to build volume:

1. Open chart replay on 1m candles at a random point in history.
2. Trade one session per sitting (e.g. 100–200 candles) with full rules, per the deliberate practice guide.
3. Log every trade including the session-stop events.
4. Target 100+ replay trades before touching even a live demo. Scalping generates samples fast — use that.

Common mistakes

1. Scalping against the 5m trend — fighting flow at 1m speed is the fastest way to donate.
2. Widening a stop "because it's only a scalp" — small trades with big losses is the worst combination in trading.
3. Trading through chop — no trend, no pullback setup, no trade. Flat VWAP = closed for business.
4. Ignoring fees — see the math above; it is the difference between a job and a hobby.
5. Learning scalping live — every mistake in this list is free in replay and expensive on an exchange.

FAQ

Is scalping profitable for beginners?

Not at first — it's the most skill-intensive style, and fees punish every error. But it's very learnable in simulation: high trade frequency means fast feedback and fast improvement, at zero cost. Get profitable over 100+ replay trades before going live small.

What is the best timeframe for scalping?

The 5m/1m combination: 5m defines trend and levels, 1m times the entry. Pure 1m trading without higher-timeframe context is noise-trading.

How many trades does a scalper take per day?

Quality scalpers often take just 3–10 A-grade setups per session, not fifty. More trades means more fees and more noise — the session stop and setup filter exist precisely to prevent quantity replacing quality.