What Is Scalping in Trading? How It Works and Its Break-Even Math

Michael Whitman · Trading since 2017, full-time since 2020 · YouTube creator, 137K subscribers · Published 2026-09-05 · Updated 2026-11-02 · Reviewed 2026-11-01

▶ Watch: Scalping for Beginners (Full Course)

Scalping is a trading style that holds positions for seconds to a few minutes and takes many small price moves in a day. Every trade pays the same spread and fees, so the smaller the move a scalper aims for, the larger the share of it those costs take.

Most explanations stop at “small profits that add up.” This one puts numbers on the part that decides whether they do: what a round trip costs, how often a small target is reached before an equal stop, and the hit rate a scalper needs just to stand still. Everything below comes from 60 sessions of SPY, the S&P 500 fund, from August to October 2026.

How it works

A scalper trades the smallest moves on the fastest charts. Positions last from a few seconds to a few minutes, often a few cents to a few tens of cents a share, and a busy scalper places dozens of trades a session. Positions are closed before the session ends, so scalping is the fastest form of day trading: a day trader may hold for hours, a scalper rarely past a few bars of a one- or five-minute chart, or a tick chart that prints a bar every set number of trades.

The approaches differ in where the small move comes from. Common ones:

• Trading the spread. Resting a bid and an offer and earning the gap when both fill, which is the market maker’s job and the hardest to compete in.
• Momentum or breakout scalps. Joining a move as price clears a level, for a few cents of follow-through.
• Mean-reversion scalps. Fading a stretch away from a short-term average, back toward it.
• Order-book scalps. Trading around visible size in the order book.

Liquidity is the price of entry. A one-cent bid-ask spread on SPY is a rounding error to a swing trader and the largest single cost a scalper pays, and in a thin stock the spread and slippage can be larger than the move being traded.

The cost of every round trip

Each trade pays a spread, any commission and the regulatory fees on the sale. On 1 November 2026 the SEC’s Section 31 fee was $20.60 per million dollars of sales, in force since 4 April 2026, and FINRA’s Trading Activity Fee on stock sales was $0.00 a share from 1 October to 31 December 2026. With an assumed one-cent spread and no commission, a round trip in SPY cost $0.01 + 0.0000206 × the price per share: $0.0255 at the period’s lowest close of $750.64 and $0.0261 at the highest, $781.42.

Against a five-minute move, that is a lot. Each move below runs from one window’s close to the next, with a session’s first window measured from the 9:30 AM open; one-hour windows cover 9:30 AM to 3:30 PM. The last two columns compare each move with the cost at its own closing price.

A third of five-minute moves (1,525 of 4,680) were no bigger than five round trips, the size at which cost eats a fifth of the move or more. At one hour that fell to 36 of 360. The cost table has the counts.

A worked example

Friday 30 October 2026, the first five-minute bar. SPY opened at $761.38 and the 9:30 bar closed at $760.01. Buy 200 shares there: $152,002 of stock.

Step 1, the cost. Spread: 200 × $0.01 = $2.00. Section 31, charged on the sale and worked here at the $152,002 entry value: $3.1312 (either exit price changes it by about a tenth of a cent). FINRA fee: $0.00. Total $5.1312, or about $0.0257 a share.

Step 2, the bracket. Target $760.26, stop $759.76, 25 cents each way. A win makes 200 × $0.25 = $50.00, minus $5.1312, for $44.87. A loss costs $50.00 plus $5.1312, $55.13.

Step 3, the break-even hit rate. Wins must cover losses: $55.13 ÷ ($44.87 + $55.13) = 55.13%. With a 50-cent bracket the same arithmetic gives $105.13 ÷ $200.00 = 52.57%. The cost, not the market, sets that hurdle above 50%.

Step 4, what happened. The 9:35 bar traded from $759.44 to $760.83, through both the target and the stop inside the same five minutes. From five-minute bars alone there is no way to say which came first. That is the honest limit of judging a 25-cent scalp from a five-minute chart, and the test below counts such cases separately rather than guessing.

The original data

The sample. Yahoo Finance five-minute bars for SPY over the 60 regular sessions from 7 August to 30 October 2026: 4,680 bars, 78 a session, after one 4:00 PM stamp was dropped. SPY went from a $771.02 open to a $755.71 final five-minute close, down 1.99%, which matters for the result below.

The bracket test. From every five-minute close except the session’s last, a long trade was given a target and a stop the same distance away. Each later bar was checked in order: target first, stop first, both inside one bar, or neither by the close. That is 4,620 entries per bracket size.

The finding: entries chosen by the clock were a coin flip, and the cost set the bar above the coin. Percentages are of resolved trades (4,101, 4,476 and 3,852 for the three sizes), leaving out both-in-one-bar cases and the 3, 84 and 759 trades still open at the close.

Shorts did slightly better, for a reason that is not skill. Mirror each trade into a short and the target-first shares become 51.84%, 52.39% and 51.90%. SPY fell over the period, and only the $1.00 short cleared its 51.29% hurdle, by less than one standard error of about 0.8 points. Neighboring entries share most of their path, so the true uncertainty is larger than that.

The smallest scalps cannot be judged from these bars. At 10 cents, 2,164 of 4,620 entries (46.84%) saw both levels inside one bar, so these bars cannot score them. The bracket table is published.

Why small targets suffer most. A move’s size grows roughly with the square root of time: a window 13 times longer should move √13 = 3.61 times as far. Over the same 60 sessions, a 65-minute window moved a median 3.31 times as far as a five-minute one, and fitting all eight window lengths gave an exponent of 0.50.

So trading 13 times as often means paying 13 costs against moves only about a third as large. The open-to-close move ran above the rule (10.17 times the five-minute median against 8.83) and the 195-minute window below it (5.16 against 6.24). The window-length table lists all eight.

Path against progress. Adding up every five-minute move in a session, regardless of direction, gave a median of $21.17 a share; the median open-to-close distance was $2.145. The median session traveled 11.55 times its net move, all of it raw material for a scalper, and every piece of it paid for separately.

Checked against Nasdaq. Nasdaq’s daily SPY open was within 5 cents of the first five-minute open on 57 of the 60 days, and its median open-to-close move was 0.285% against 0.278% from the five-minute bars.

The teaching is short on failure. In this site’s study of 24,971 trading videos, 60% of those with scalping in the title never pass 50,000 views, and of the 136 that carry description text, 3 mention invalidation, failure, being wrong, a loss, a mistake or risk.

Rules that changed in 2026

The $25,000 pattern day trader minimum is gone. FINRA’s Rule 4210 amendments, Regulatory Notice 26-10, replaced the pattern day trader designation and its $25,000 minimum with intraday margin based on actual exposure, effective 4 June 2026. Firms may phase it in until 20 October 2027, so a broker can still apply the old count. Many guides still say scalping needs $25,000.

The FINRA sale fee is zero for the rest of 2026. Its Trading Activity Fee on stock sales is $0.00 a share from 1 October to 31 December 2026 (filing SR-FINRA-2026-021), and its usual $0.000195 a share returns on 1 January 2027. The Section 31 fee, which FINRA collects from firms for the SEC, still applies. Recheck both before working out a cost.

Scalping itself is legal. Manipulation is not. In US futures, the Commodity Exchange Act names spoofing, “bidding or offering with the intent to cancel the bid or offer before execution”, as unlawful, and the market manipulation page covers the stock-market equivalents.

When it fails

The hit rate needed is above 50% before any skill shows up

A symmetric bracket hands the cost straight to you. At 25 cents each way the hurdle was 55.17% and the measured rate 48.16% long, 51.84% short. Any edge has to lift the hit rate about five points above a coin flip across every trade taken, not just over a good week.

The thin moves come in the afternoon

The share of five-minute moves no bigger than five round trips rose through the day: 20.56% in the first hour from 9:30 AM, 30.42% from 11:30 and 41.53% from both 1:30 and 2:30 PM, before 35.28% in the last half hour. The quiet stretch, when a bored scalper is most tempted, is when the cost takes the largest bite.

The result depends on the period’s direction

Longs and shorts split 48 to 52 here most likely because SPY fell 1.99%. A rising quarter would likely flip the gap. Calling that an edge is mistaking the market’s direction for skill.

Five-minute bars hide the order of events

At 10 cents a side, nearly half the test’s entries could not be scored. A backtest on bars that cannot see the sequence will quietly count some of those as wins.

Questions people ask

Is scalp trading illegal?

No. Buying and selling quickly for small moves is legal in US markets. What is illegal is manipulation, such as placing orders meant to be canceled before they fill to move the price.

Is scalping profitable?

It can be for a trader with a real edge and low costs, but the arithmetic is against the average attempt. On SPY in fall 2026, a 25-cent scalp needed a 55.17% hit rate to break even; entries taken at every five-minute close reached their target first 48.16% of the time going long.

Is scalping like gambling?

Without an edge, the outcome of a symmetric bracket behaves like a coin toss with a fee on every toss. With an edge and costs counted, it is a business with thin margins. The trade count makes either result show up quickly.

Is scalping high risk?

Each trade risks little, but the trade count makes small errors compound. Leverage, slippage on fast moves and a stop that fills worse than planned matter more at this speed than at any slower one.

What time frame do scalpers use?

Tick, one-minute and five-minute charts. A regular US session of 390 minutes holds 78 five-minute bars.

Related

How to scalp a market turns the cost check into steps.

Day trading is the same idea one step slower, and the momentum trading vs scalping comparison measures cost against moves from five minutes out to three months.

Timeframes explains why one market looks so different on a one-minute and a daily chart. Swing trading holds for days, where a two-and-a-half-cent cost barely registers, and the intraday trading page measures how bar ranges change through the session.

And risk management matters most at this speed, because a high trade count turns a small sizing error into a large one fast.

This page is educational, not financial advice. Test every idea on your own charts before risking money.