Scalping vs Momentum Trading: Which One Pays After Costs?

Michael Whitman · Trading since 2017, full-time since 2020 · Published 2026-09-08 · Updated 2026-10-22 · Reviewed 2026-10-22

Momentum trading and scalping differ in how much movement each trade needs, not in speed. On SPY a round trip of about $0.026 a share was 11.83% of a median five-minute move and 0.22% of a median one-month move, and five-minute moves reversed slightly more often than they continued, in SPY and in three other index funds tested on this page.

What each one is

Momentum trading holds a move while it runs. The entry is into strength and the exit is when the move stops, so a single position can last days, weeks or months. Momentum trading covers it.

Scalping takes a very small amount out of many trades. The target is a few cents or ticks, repeated often, with a stop that is tighter still, and positions last seconds to minutes. Scalping covers that style.

The shared surface is activity. Both demand attention and both produce a lot of decisions, which hides the two questions that actually decide which is workable: does the move continue over the holding period, and how big is the cost next to that move?

Where they differ

How much movement each trade needs. Momentum takes whatever a real move offers, and over weeks a typical move in SPY was measured in dollars. Scalping works inside moves of a few cents, while paying the same per-share cost to get in and out.

How many round trips you pay. A scalper may pay for dozens of round trips a day; a momentum trader pays for a handful a month. Cost per trade times trades per month is the bill, and it is the larger number by a wide margin for the scalper.

What each assumes about the market. Momentum assumes a move that has started tends to keep going over its holding period. Scalping assumes very little about direction and lives on the spread and on tiny moves, which makes its costs the whole story.

What each needs from execution. Momentum tolerates a slightly late fill. Scalping does not, because a fraction of a cent of slippage can be a large share of the entire target.

Where they agree

Both need a written invalidation before entry. The distance differs enormously; the requirement, and the position-sizing arithmetic that runs off it, are the same.

Both produce a reviewable record. Hundreds of scalps a month build one fastest, while a momentum record takes longer to accumulate, but either one can be measured once it exists.

Neither gets the cost back on a losing trade. It is charged whether you are right or wrong, so a method with a tiny edge per trade needs that edge to survive the fee on every trade.

Which one to use: Momentum Trading or Scalping?

Use momentum-style holding when your costs are ordinary retail costs. Over a month, a round trip was well under 1% of a typical SPY move, so the result depends on your decisions rather than on your fee schedule.

Use scalping only when your cost per round trip is close to zero next to a five-minute move, which usually means one very liquid instrument, a tight spread and a fee structure most retail accounts do not have.

Scalp when fast feedback matters most to you and the arithmetic still works, because hundreds of trades a month produce a record quickly.

And when the appeal of scalping is that the risk per trade is small, look again. A small risk taken many times a day is the same risk taken more often, plus a cost each time.

The original data

Two measurements on SPY, the S&P 500 fund, answer the two questions above. Both use Yahoo Finance bars downloaded on 3 October 2026: daily adjusted closes from 29 January 1993 to 2 October 2026 (dividends reinvested), and 4,680 five-minute bars from 60 full sessions, 10 July to 2 October 2026.

First, did a move continue over the next equal period? Each period’s return was paired with the next one, and the table counts how often the two went the same way. If moves were unrelated, the expected share depends only on how often the market rose in that period, and that is the comparison column. Five-minute and 30-minute pairs stay inside one session; daily and longer periods are back-to-back blocks counted back from 2 October 2026, so no period is used twice. The p column uses the normal approximation to the binomial.

No horizon shows moves carrying on more often than chance. The five-minute figure is the only clear departure, and it runs the other way: consecutive five-minute moves reversed slightly more often than they continued, a correlation of -0.035. Days and weeks lean the same way by a small margin (p = 0.033 and 0.048, borderline with six periods tested), and months sit inside the noise. A one-year row is in the published file too, but with 32 pairs it is too small to read.

The one place a trend did separate what followed was the trailing year. At each month-end from January 1994 to August 2026, SPY’s return over the past twelve months was positive 322 times and the next month rose in 220 of them, 68.32%. After the 70 month-ends with a negative year it rose in 35, 50.00%. A two-proportion test gives p = 0.004. Treat it with care: the 70 negative months come from a few bear markets, so they are not 70 independent cases, and next-month swings after a negative year were wider (a standard deviation of 6.23% against 3.72%).

Three more funds and a later window give the same short-term answer. To check that the five-minute figure was not a quirk of one fund or one stretch of weeks, the same count was run on QQQ (the Nasdaq-100 fund), IWM (Russell 2000) and DIA (the Dow), using Yahoo Finance five-minute bars for the 60 full sessions from 30 July to 22 October 2026, downloaded 22 October 2026.

Consecutive five-minute moves went the same way 47.95% of the time in QQQ (4,511 pairs), 48.06% in IWM (4,409) and 48.40% in DIA (4,461), against about 50.0% expected if moves were unrelated. The two-sided p values are 0.006, 0.010 and 0.032. All four funds leaned toward reversal at five minutes, and none toward continuation.

Daily moves since each fund listed gave the momentum side nothing either. On adjusted daily closes, consecutive QQQ days went the same way 49.91% of the time against 50.46% expected (6,871 pairs from March 1999), DIA’s 49.26% against 50.32% (7,162 from January 1998) and IWM’s 48.98% against 50.21% (6,600 from May 2000, a borderline lean toward reversal at p = 0.047).

Nasdaq’s own price history for the last ten years agreed with Yahoo’s closes to within 11 cents a share and gave identical daily counts, so the result does not rest on one data vendor. Every row is in the four-fund table.

Second, how big is a round trip next to a typical move? Take 100 SPY shares at the 2 October 2026 close of $769.64. Crossing an assumed one-cent spread costs $1.00. On the sale, the SEC’s Section 31 fee of $20.60 per million (from 4 April 2026) is $1.5855, and FINRA’s Trading Activity Fee of $0.000166 a share is $0.0166. Both apply to sales only; the SEC charges the exchange and FINRA charges the broker, and this example assumes both are passed on to you. That is $2.60, or about $0.026 a share, spread included, before any commission. The spread is an assumption: under SEC Rule 612 quotes move in steps of $0.01, or $0.005 for the stocks with the tightest average spreads, so it cannot be narrower than that step.

The intraday moves come from the 60 five-minute sessions and the longer ones from the 252 sessions from 1 October 2025, as absolute close-to-close changes over every window of that length. The longer windows overlap heavily: the three-month median rests on 190 overlapping windows inside one year. The intraday trading page measures the same cost problem with bar ranges by time of day; this table extends it out to months. Both measurement tables are published with every cost row.

The video record is lopsided too. Of the 24,971 videos in the search study behind this site, 1 compares the two directly in its title, at 11 views. Scalping is in 709 titles at a median of 23,380 views across 353 channels; momentum trading in 82 at a median of 6,167 across 70.

Why the cost decides it

Because it is fixed per share and the move is not. On the table above, the same $0.026 is more than a ninth of a typical five-minute move and about a fifth of one percent of a typical month, and no amount of skill changes that ratio.

And because direction offers no help at the short end. A scalper cannot count on five-minute moves to continue; in SPY they reversed slightly more often than not, so the edge has to come from somewhere other than riding the move.

When it fails

The first failure is scalping on a retail cost base, and it looks like a skill problem when it is an arithmetic one. Entries hit the levels, exits hit the targets, stops are honored. The account still shrinks, because a large share of every small gain is handed back at both ends of every trade. Take 100 round trips of 100 SPY shares that win and lose the median five-minute move of $0.22 equally often: the gross result is zero, and the cost at $2.60 a round trip is $260.

The second is running momentum where moves do not continue. On SPY, days and weeks leaned slightly toward reversal (a borderline result), so buying strength on a short horizon was buying into a small headwind, not a tailwind. At five minutes the lean was clearer and it held in every fund tested: buying a strong five-minute bar in QQQ meant betting on a move that repeated its direction 47.95% of the time.

A third is scalping a wide-spread instrument. If the spread is several cents, it can exceed the whole target.

A fourth is trusting the trailing-year result. It rests on a few bear markets, on one fund, and a rule built on it would have faced the same handful of episodes.

And a fifth is treating small risk per trade as small risk. Frequency multiplies it.

Related

Momentum trading covers holding a move while it runs. Scalping covers the very short holding period. And day trading covers the wider category both belong to.

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