Scalping

Scalping is the family of trading approaches that tries to collect an edge in small pieces, many times, over very short holding periods. A scalp typically lasts from a few seconds to a few minutes and almost never survives the session; the target is a small fraction of the instrument's daily range. What separates scalping from other short-term styles is not only speed but proportion: because the distance to target is small, spread, commission and slippage are a material part of each trade's outcome rather than a rounding error. Any description of a scalping system that ignores execution cost is incomplete.

## How it works

Most scalping systems share the same skeleton. A context filter decides when the strategy is allowed to trade at all: a session window, a higher-timeframe trend condition, a volatility floor, or a distance from a reference level. Inside that context, a trigger fires on a low timeframe — commonly 1, 3 or 5 minute candles, sometimes tick or range bars. The trigger can be a break of a reference price, a pullback into a moving average or channel, a candlestick or micro-structure pattern, an indicator crossover, or a reading from volume and order flow.

Risk is defined at entry, usually as a structural stop (behind the level, the wick, or the range) or a volatility multiple such as a fraction of ATR. The exit is normally mechanical: a fixed reward-to-risk multiple, the next liquidity level, a trailing stop, partial profit-taking, or a time stop that closes the position if the move does not develop within a set number of bars. Because the holding period is short, trade management rules carry an unusually large share of the results — two systems with identical entries can behave very differently depending only on how they exit.

## Main variants

Across the 86 decoded versions linked from this page, most implementations fall into a handful of recognisable families:

**Level-based scalping.** Entries reference a pre-computed price: the previous day's high and low, the opening range, a session open, VWAP, or a prior swing. The level provides both the trigger and the natural stop.

**Momentum and breakout scalping.** The system waits for expansion — a range break, a volume surge, a strong candle — and joins the move for a short, measured target.

**Mean-reversion and range scalping.** The opposite premise: price is expected to return to a mean or to the middle of a band after stretching. Compression zones, channel extremes and oscillator extremes are typical triggers.

**Indicator-driven scalping.** Channels (Keltner, Bollinger), MACD variants, EMA stacks and composite signal indicators bundle several conditions into a single entry arrow. The logic is explicit, which makes these versions the easiest to code and test.

**Structure and pattern scalping.** Liquidity sweeps, stop runs, outside days, engulfing patterns and similar formations, often framed as traps where the entry is taken against the failed move.

**Time-anchored and event scalping.** The setup is tied to a specific clock time (an index open, a scheduled release) rather than to a chart condition.

## What typically differentiates implementations

Two versions branded with the same name can be very different systems. The variables that matter most are the base timeframe, the instrument and its typical spread, the session window, the confirmation filter layered on top of the trigger, where the stop is placed, whether the target is fixed or adaptive, whether partials or a trailing stop are used, and how many trades per day the rules allow. Position sizing and the maximum number of consecutive losses tolerated also change the character of the strategy far more than the entry signal does.

## Common mistakes

The recurring errors are structural rather than exotic. Ignoring transaction costs turns a marginal system into a profitable-looking one on paper. Testing on the closing price of a low timeframe hides the fact that stop and target could both be touched inside the same candle. Over-filtering until only a handful of historical trades remain produces a curve that describes the past, not the method. Trading outside the liquid hours the strategy was designed for changes the spread and the behaviour of the level being traded. And moving or removing the stop after entry — common when targets are small and losses feel recoverable — destroys the asymmetry the system depends on.

## How to evaluate and backtest versions of it

Start by writing the rules until they are unambiguous: exact timeframe, exact session, exact stop and target, exact conditions for skipping a signal. Then test with realistic assumptions — actual spread, commission, and a slippage estimate consistent with the instrument and the time of day. Intrabar resolution matters: use tick or a lower timeframe to resolve the order in which stop and target were reached. Because scalping generates many trades, a usable sample is available quickly, but that sample should span different volatility regimes, not just one calm or one trending stretch. Compare results across sessions, days of the week and instruments to see whether the edge is a property of the logic or of one specific market condition. Finally, check the sensitivity of parameters: a version whose results collapse when a threshold shifts slightly is describing noise, not structure.

Strategies in this concept (199)

Frequently asked questions

How is scalping different from day trading in general?

Both close positions within the session, but scalping sits at the shortest end of the spectrum: holding periods of seconds to minutes, targets that are a small fraction of the daily range, and many trades per session. The practical consequence is that transaction costs and execution quality weigh far more on the result than they do in a strategy that holds for hours.

What timeframes and instruments are typically used?

Charts of 1, 3 or 5 minutes are the norm, sometimes with tick, range or second-based bars. Instruments are chosen for liquidity and tight spreads — index futures and CFDs, major forex pairs, gold, and high-volume crypto pairs are the usual candidates. Wide-spread or thin instruments rarely leave room for a small target to be profitable after costs.

Does scalping require an indicator, or can it be purely price-based?

Both exist. Level-based and structure-based versions use only price references such as previous highs and lows, opening ranges or session opens. Indicator-based versions use channels, oscillators or composite signals. Neither category is inherently better; what differs is how explicit the rules are, and explicit rules are easier to test and reproduce.

Why do scalping backtests often look better than live results?

Three causes dominate. Costs are underestimated or omitted; the test resolves candles at close, so it never has to decide whether the stop or the target was hit first inside a bar; and the strategy is optimised on a narrow historical window. Adding realistic spread, commission and slippage, and testing with intrabar data, usually explains most of the gap.

How much data is enough to evaluate a scalping strategy?

Trade count accumulates fast, so the binding constraint is usually regime coverage rather than sample size. A useful test spans quiet and volatile periods, trending and ranging phases, and ideally more than one year including seasonal or session shifts. A large number of trades drawn from a single market condition is not a diverse sample.

Can scalping strategies be automated?

Many can, and the mechanical nature of most rules makes them natural candidates. The limits are practical: latency, order fill quality, and the fact that some discretionary versions rely on reads of order flow or context that are hard to specify. Before automating, the rules need to be written to a level where two people coding them independently would produce the same system.

All strategy concepts · Explore all strategies