Backtest · NQ, ES and four more futures · 1-minute bars · 2024 → 2026

Does Scalping Futures Work? 66,000 Setups Tested on Real 1-Minute ES and NQ Data

Scalping — trades held for minutes, aiming for a few points — is the most common way people try to day-trade S&P 500 and Nasdaq-100 futures. We tested about 66,000 versions on two years of real 1-minute futures prices, then chased the few that looked good: we optimized them, ran them on older years, moved them to the micro S&P contract, and tried tight stops, hourly entries and a weekly loss limit.

Published October 14, 2026 · Stable Income Strategy research

The short answer

Almost never. Of about 66,000 scalping setups (trades held up to 30 minutes) tested on two years of real 1-minute bars for six futures markets, only one beat random look-alike trades: a break of the Nasdaq-100 futures’ 9:30–10:00 range, with a target of about 4.6 points and a stop of about 111. It made +1.9 and +2.0 points a trade after costs in the two years (about $1,100–$1,300 a year per MNQ, the micro Nasdaq-100 future at $2 a point), winning about 97% of trades but losing up to 215 points on the rest. It lost on the S&P 500 and four other markets, and on Nasdaq ETF minute bars it lost money in 2016–2021 — it looks fitted to 2024–26, and with 74% of its wins coming inside the minute the order filled, only tick-by-tick data can confirm it.

scalping setups tested on six futures markets: ~66,000
positive in both years outside Nasdaq day sessions — the same as random trades: 0–1%
points a trade after costs for the one survivor (NQ only), year 1 / year 2: +1.9 / +2.0
points a trade for the same rule on Nasdaq ETF bars, 2016–18 / 2019–21: −0.59 / −0.63

Key takeaways

About 66,000 scalping setups on NQ, ES, YM, RTY, gold and crude oil: outside Nasdaq day sessions only 0–1% were positive in both years — no more than random trades with the same exits.
The one survivor: a break of the Nasdaq-100 futures’ 9:30–10:00 range by stop order, target about 4.6 points, stop 0.35–0.5 ATR, out after 30 minutes — +1.9 / +2.0 points a trade after costs, 96.5% winners, losers up to 215 points.
Optimized (stop 1 ATR, 60-minute hold, entries until 2 pm) it made +$3,814 per MNQ over two years against +$2,160 — but on Nasdaq ETF minute bars both versions lost in 2016–18 and 2019–21 and were only thinly positive in 2022–24.
On the S&P 500 the Nasdaq rule lost $2,796 per MES over two years. The best S&P fit — buy above the first hour’s high, 30-point target and stop — made $1,858 a year per MES in 2024–26 but lost on S&P ETF minute bars in every 3-year block since 2016.
Buying every hour with a tight stop: 15 of 7,112 MES variants (0.2%) and none of 7,112 on micro natural gas were positive in both years; the one good MNQ slot (+$10.0k at 9:30) was the Nasdaq’s open-to-close drift.
One MES a day with a hard stop and a $100 weekly loss limit: the median setup lost $192–$491 a year; the best row (+$1,066 a year) was luck of the cap — the same trades without it were a coin flip.
Tight stops turned the survivor into a loser (−$11.1k to −$1.7k per MNQ at 0.05–0.15 ATR), and with tight brackets 14–32% of trades hit stop and target inside one minute — 1-minute bars cannot say which came first.

What we tested

Six studies on the same two years of real 1-minute futures bars (October 2024 to October 2026), each one a step further. Stops and targets are set in ATR — the average daily range over the last 14 days — so they scale with each market; on NQ the typical ATR was 310 points.

1The big scalping search

Tested: About 66,000 setups from seven families — VWAP bands, opening-range breaks, EMA pullbacks, RSI(2), prior-day high/low fades, round numbers and the opening drive — on NQ, ES, YM, RTY, gold and crude oil, in the day session and overnight. Holds up to 30 minutes; stops and targets from 0.0075 ATR (2.3 NQ points) to 0.5 ATR.

Found: Nasdaq day session: 5% positive in both years, against 2–3% for random look-alike trades. Every other market and window: 0–1%, the same as random. One rule beat the controls.

2Optimizing the survivor

Tested: 107,520 versions of the Nasdaq opening-range break — range 5–60 minutes, entry window, sides, target 0.0075–0.05 ATR, stop 0.1–1 ATR, hold 5–60 minutes — on the real 9:30 range and on fake 11:00 and 13:00 ranges.

Found: A wider stop (1 ATR), a 60-minute hold and entries until 2 pm made +$3,814 per MNQ over two years, against +$2,160 for the rule as found — partly fitted.

3Older years on ETF bars

Tested: Both versions on Nasdaq-100 ETF (QQQ) minute bars scaled to futures points, 2016 to October 2024 — years no search had seen.

Found: Lost in 2016–18 and 2019–21, thinly positive in 2022–24. Not confirmed outside 2024–26.

4The S&P 500 (MES)

Tested: The Nasdaq rule on ES; the same 107,520-setup grid on ES; 34,560 setups with bigger targets; a 5,760-setting sweep; the best fits on S&P 500 ETF minute bars since 2016.

Found: The Nasdaq rule lost $2,796 per MES. The best S&P fit, a first-hour breakout, held in both futures years but lost on the older ETF years.

5Buying every hour with a tight stop

Tested: Buy MES, MNQ or micro natural gas (MNG) at 9:30, 10:00 … 3:00 pm with a 0.05–0.5% stop and a 1–3% target, flat by 4 pm; 7,112 variants on MES and on MNG, plus a stop-size sweep on MES.

Found: MES: 15 of 7,112 positive in both years. MNG: none. The one good MNQ slot is the Nasdaq’s open-to-close drift, not the stop.

6One MES a day with a $100 weekly limit

Tested: Eight day-trading rules (opening range, momentum, VWAP, gap fill and others) plus buy or sell the open, each with a fixed 5, 10 or 20-point stop and at most 20 stop-points ($100) of losses a week: 468 capped setups, against 200 runs of random entries.

Found: The median setup lost money. The best row was the luck of which trades the weekly cap let through.

Seven scalping families, six markets

Every family was run on NQ, ES, YM, RTY, gold and crude oil, with many stop, target and entry variants each. “Controls” are the same orders at shifted or random times — what luck alone earns.

Scalping families tested and the best result of each
Family	What it does	Best result
First-minutes range break	Buy a break above, or sell a break below, the high or low of the first minutes of the session, with a stop order	NQ’s 9:30–10:00 break with a tiny target: the only survivor. The same rule loses on ES, YM, RTY, gold and crude.
Opening drive	Follow a strong first 5 or 15 minutes	NQ: +1.3 to +3.8 points a trade, but only p ≈ 0.10 against the controls. NQ only.
Round numbers	Fade a touch of a round price (NQ 100/50, ES 50/25, gold $50/$25 …) with a limit order	Gold at $50 levels: about +$6 / +$4.5 a trade per MGC, p 0.07, negative in 2026 Q3. Index versions lose.
VWAP bands	Fade moves 1, 2 or 3 standard deviations from VWAP (the day’s volume-weighted average price)	No market beat the controls.
EMA 9/20 pullback	Buy a pullback to the 20-bar average in the direction of the 60-minute trend	No market beat the controls.
RSI(2)	Trade a very short overbought / oversold gauge on 1- and 5-minute bars	No market beat the controls.
Prior-day high/low fade	Fade a touch of yesterday’s day-session high or low	No market beat the controls.
Overnight (6 pm–9 am)	The same families at night	Nothing robust. Best: NQ’s 6 pm range break, +0.5 / +0.6 points a trade.

Of 83 best-of-family picks made on the first year, 9 were positive in the second. YM fails on costs alone: micro fees are 3.2 index points a round trip.

The one survivor: the Nasdaq 9:30–10:00 range break

The rule as tested on NQ: at 10:00 ET, mark the high and low of the first 30 minutes (the opening range). Until noon, a stop order buys a break above the high or sells a break below the low. The target is 0.015 ATR — about 4.6 points — and the stop 0.35 ATR, about 111 points; a trade still open after 30 minutes is closed. In 2024–26 it made +1.9 and +2.0 points a trade after costs (daily Sharpe 0.13 and 0.11), about 284 trades a year, positive in every quarter and in 18 of 25 months.

The Nasdaq opening-range scalp and its variants, per MNQ, October 2024 to October 2026
Version	Trades	Won	Per MNQ, 2 years	Max drawdown
As found: stop 0.35 ATR, out after 30 min	567	96.5%	+$2,160	$528
Same, stop 0.25 ATR (about 85 points)	568	–	+$2,412	$395
Optimized: stop 1 ATR, hold 60 min, entries until 2 pm	624	98.1%	+$3,814	$747
Optimized, only on days above the 200-day average	540	–	+$3,733	$245

Max drawdown is the deepest fall from a high point. The wins average about 4.5 points and the losses 62–68, so about 10 losing trades a year decide the result. Adding a second day filter (trade only against the opening gap) left 4 small losers and a $29 drawdown in two years — too good: the filters were chosen after seeing the ~20 big losses. On Nasdaq ETF bars the 200-day filter added about 0.2 points a trade in 2022–24, nothing before, and lowered the plain rule in 2024–26.

The S&P 500: nothing carried over

ES is the most traded index future, and MES — the micro S&P 500 future, $5 a point — the usual small-account contract. Nothing the Nasdaq showed worked on it.

The Nasdaq rule on MES

The optimized Nasdaq breakout lost $2,796 per MES over two years: 549 trades, 93.3% won, profit factor 0.36 (dollars won per dollar lost). On ES its target is only 3–4 ticks, while fees and a tick of slippage cost about 0.57 point, so it needs about 97% winners to break even.

Searching ES on its own

107,520 breakout versions picked on year 1 had a year-2 daily Sharpe of −0.11, worse than a fake 11:00 range. Only 0.1% were positive in both years, against 0.6% on the fake range (0.7% vs 2.6% even before fees). With bigger targets, 34,560 versions: 6.5% positive in both years, the same as the fake range’s 6.7%.

The best S&P fit: a first-hour breakout

Buy a break above the 9:30–10:30 high, target and stop 0.5 ATR (about 30 points), out after 2 hours: 302 trades, 57% won (+$74 / −$70 on average), profit factor 1.40, $1,858 a year per MES ($2,002 then $1,633), max drawdown $778. With fills 2 ticks worse: $1,616 / $1,358. Later fake first hours and random-minute buys with the same exits all made less (0 of 20).

But not on older years

The same search on fake 11:00–13:00 ranges found rules making up to about $800 a year in their weaker year, so part of the result is selection luck. On S&P 500 ETF minute bars, never seen by the search, the same shape reproduced in 2024–26 but lost in every earlier 3-year block: −0.33, −0.52 and −0.19 points a trade in 2016–18, 2019–21 and 2022 to October 2024. It is fitted to 2024–26.

Is it luck? How we checked

Test 66,000 setups on two years and some will look great by chance. These checks separate a pattern from a lucky pick — and show where 1-minute bars stop being able to tell.

Random look-alike trades

Every setup was rerun with its orders moved to the same minute 5–61 sessions away (20 shifts) and at random minutes and sides (10 runs), with the same exits. Outside Nasdaq day sessions, real setups were positive in both years no more often (0–1%). The survivor beat all 30 controls in year 2.

Fake opening ranges

The same breakout from other 30-minute ranges (10:00–14:00) earned about its costs, and none of 10 random-time runs did as well. In the optimization, rules picked on year 1 kept an edge in year 2 on the real 9:30 range (daily Sharpe 0.091 for the top 10) and lost on fake 11:00 and 13:00 ranges (−0.080, −0.099).

Picked on year 1, run on year 2

Year-1 stars of the optimization — 99% winners, no losses, daily Sharpe about 1.0 — fell to about 0.13 in year 2, no better than the plain rule. Positive in both years: 11.8% of versions on the real range, 3.8% and 0.8% on the fake ones.

Older years on ETF bars

On QQQ minute bars × 41.3, the rule lost 0.59 and 0.63 points a trade in 2016–18 and 2019–21 ($344 and $359 a year per MNQ) and made +0.69 from 2022 to October 2024 ($406 a year), with a 95% range of −0.28 to +1.53 that includes zero. Caveats: the target was only 0.7–0.8 point in 2016–17, below the 0.8-point fee, and the ETF reads about 1.2 points a trade lower than real NQ on the same days (12 stop-outs against 6).

Slightly worse fills

Two extra ticks of slippage still left +1.4 / +2.1 points a trade. But moving the fill and the bracket 2 ticks worse turns 3–4 winners into losers and halves year 1 (+1.87 → +0.89 points a trade).

The bar-order problem

A 1-minute bar shows the open, high, low and close, not which came first. With brackets under 0.025 ATR (about 8 NQ or 1.5 ES points), 14–32% of trades hit stop and target in the same minute, and assuming the better or the worse came first moves the result by 2–3.5 NQ points a trade — more than the target. A common shortcut gave tight breakouts a daily Sharpe of 0.5–0.7 — and made random controls look profitable too. The survivor has no such minutes, but 74% of its wins land inside the fill minute. Tick-by-tick trades with bid/ask quotes, or at least 1-second bars, are needed to settle it.

What did not work

Ideas tested along the way, on the same two years of real 1-minute bars, with costs.

Day-trading and scalping ideas that failed
Idea	What happened
Buy every hour (9:30 → 3 pm) with a 0.05–0.5% stop and a 1–3% target, on MES, MNQ and micro natural gas	Targets almost never filled (MNQ at 9:30 with a 0.5% stop: the 1%, 2% and 3% targets hit on 93, 10 and 1 of 498 days) and always cut profit. MES: 15 of 7,112 variants positive in both years; natural gas: 0 of 7,112. Hours picked on year 1 lost in year 2 (average of the top 10: MES −$4.2k, MNQ −$2.2k).
MNQ at 9:30 with a 0.5% stop, no target	+$10.0k over two years, but $5.5k came from 2025 Q2, and entering 10–15 minutes later turned it to −$1.9k / −$3.1k, while plain holding 9:30 → 4 pm stayed +$3k to +$5k at every start. It is the Nasdaq’s open-to-close drift, not the stop.
MES with any hourly stop from 0.05% to 1.5%	Each stop averaged −$2.1k to −$3.7k per hourly slot over two years. Costs, about $4 a trade, take roughly $2k per slot — about MES’s whole open-to-close gain.
One MES a day, hard stop 5/10/20 points, $100 weekly loss limit	4 of 33 rule × stop rows positive in both halves, none with a t-statistic above 1.2; 30 of 468 capped setups (6%). The best row (+$1,066 a year) without the cap: profit factor 1.02, +$2,188 then −$1,818; with the week starting on Tuesday: −$310 a year.
Trailing stop instead of the quick target (NQ breakout)	Every trail made less: best +$1,169 against +$2,160 per MNQ, with a losing year 1. The edge is the first pop. On the opening drive a wide trail made +$4,219 but the drawdown rose to $1,649 from $497.
A tight stop to start small (NQ breakout)	Stops of 0.05, 0.1 and 0.15 ATR lost $11.1k, $5.3k and $1.7k per MNQ; fixed stops of 60 points or less lost. A tight stop is a different, losing rule — not a smaller version.
Trade only below (or above) the 50/100/200-day average, or near the lower Bollinger band	About 6 of 117 combinations at p ≤ 0.05 — what chance gives. The breakout did worse below the 200-day average. The opening drive below the 100-day (56 trades, 91% won) rests on about 8 sell-offs.
Bigger targets on MES	The Nasdaq shape lost on MES at every target from 1 to 30 points.

How much money does a small account need?

Only the Nasdaq breakout had results worth sizing, and only in the 2024–26 backtest it was fitted to. With that warning, here is what the numbers say about small accounts.

Backtest return per MNQ

The optimized breakout on days above the 200-day average made about $1,900 a year per MNQ ($1.4k, then $2.3k) with a $245 max drawdown: about 190% a year on $1,000, 95% on $2,000, 55% on $3,500.

Chance of losing half

With every fill $1 worse and one extra −$600 day a year, $1,000 had a 7.5% chance of halving and $500 a 23% chance.

Stop size and a $1,000 start

With a 0.25 ATR stop (about 85 points, about $170 on one MNQ), $1,000 had a 0.5% chance of falling to $300 within a year (2.6% with fills 1 point worse), against 3.2% and 8.8% with the 0.35 ATR stop. The 0.25 setting was picked after seeing the results.

One bad trade

On ETF bars in 2022–26 the worst single trade cost $304–$542 per MNQ, depending on the version — 30–55% of a $1,000 account.

Just holding one MNQ

Holding one MNQ made $17.7k over the same two years but fell $10.8k on the way (April 2025), so it needs about $12,000 or more per contract.

MES with $1,000 and a $100 weekly limit

The weekly cap held the worst week to −$103 to −$113 (fees and slippage on top of the 20 points), but losing streaks ran 5–15 weeks and cost $251–$1,573 — and the median setup lost money.

The risks

Two years of real futures data, mostly a rising market (ES rose 24%). On older ETF data the Nasdaq rule lost in 2016–2021, and the best S&P fit lost in every 3-year block.
Rare big losses: about 97% of trades win around 4.5 points, the losers average 62 points and the worst was 215; the best 5 days made 102 points, the worst 5 lost 685.
Fills decide it: 74% of the wins come inside the minute the order filled. Fills 2 ticks worse halved year 1, and fast breakouts can slip more than the one tick assumed.
Fitting: on ETF bars the optimized version made +2.65 points a trade in the years it was tuned on and +0.86 in 2022–24; filters chosen after seeing the ~20 big losses look perfect and mean little.
NQ only: the same rule lost on ES, YM, RTY, gold and crude oil.
Costs add up: about $4 a round trip on one MES; at 45–225 trades a year that is $200–$900 before any edge.

Questions

Does scalping futures work?

Rarely, in our test. We ran about 66,000 scalping setups — holds of up to 30 minutes on NQ, ES, YM, RTY, gold and crude oil, day and night — on two years of real 1-minute bars (October 2024 to October 2026), with costs. Outside Nasdaq day sessions only 0–1% were positive in both years, no more than random trades with the same exits. One Nasdaq rule survived, and on older Nasdaq ETF data it lost money in 2016–2021.

What is the best opening range breakout strategy for NQ?

The only scalp that survived our test: mark the high and low of 9:30–10:00 ET, enter on a break either way with a stop order until noon, target about 4.6 points (0.015 ATR), stop about 111 points (0.35 ATR), out after 30 minutes. In 2024–26 it made +1.9 / +2.0 points a trade after costs (567 trades, 96.5% won, +$2,160 per MNQ over two years). A 1-ATR stop, 60-minute hold and entries until 2 pm made +$3,814. Neither was confirmed on Nasdaq ETF bars before 2022.

Can you day trade MES profitably?

Not with any rule we found. The Nasdaq breakout lost $2,796 per MES over two years, and 107,520 versions searched on ES did no better than fake ranges. The best S&P fit — buy a break of the first hour’s high, 0.5 ATR (about 30 points) target and stop, out after 2 hours — made $1,858 a year per MES in 2024–26 but lost on S&P 500 ETF minute bars in 2016–18, 2019–21 and 2022–24. With one MES, a hard stop and a $100 weekly loss limit, the median setup lost $192–$491 a year, depending on how the limit was applied.

How much money do you need to day trade micro futures?

It depends on the worst losses, not the average. For the Nasdaq breakout in its 2024–26 backtest (about $1,900 a year per MNQ, max drawdown $245), $1,000 had a 7.5% chance of losing half in a year and $500 a 23% chance, with every fill $1 worse and one extra −$600 day a year. On ETF data the worst single trade in 2022–26 cost $304–$542 — 30–55% of $1,000. Simply holding one MNQ made $17.7k in two years but fell $10.8k on the way, so it needs about $12,000 or more.

What is the best time to day trade futures?

In our data the only scalping edge was at the Nasdaq open: the 9:30–10:00 range break. The same breakout from 30-minute ranges later in the day (10:00–14:00) earned about its costs, and no overnight scalp (6 pm–9 am) held up on any of the six markets. Buying MNQ at 9:30 with a 0.5% stop made +$10.0k in two years, but that was the Nasdaq’s open-to-close drift: entering 10–15 minutes later turned it negative.

Do tight stops work for scalping futures?

Not in our tests. On the Nasdaq breakout, stops of 0.05, 0.1 and 0.15 ATR lost $11.1k, $5.3k and $1.7k per MNQ over two years, and fixed stops of 60 points or less lost; 0.25 ATR and wider kept the edge. Buying every hour with a 0.05–0.5% stop was positive in both years on 15 of 7,112 MES variants and none on micro natural gas. Very tight brackets also cannot be judged on 1-minute bars: 14–32% of trades hit stop and target in the same minute.

Can you backtest a scalping strategy on 1-minute bars?

Only roughly. A 1-minute bar does not say whether the high or the low came first. With stops and targets under about 8 NQ points, 14–32% of trades hit both in one minute, and the ordering assumption moves results by 2–3.5 points a trade — more than the target. A common shortcut gave tight breakouts a daily Sharpe of 0.5–0.7 — and made random trades look profitable too. A fair test needs tick-by-tick trades with bid/ask quotes, or at least 1-second bars.

Why does the scalp work on NQ but not on ES?

Costs and follow-through. A round trip (fees plus a tick) costs about 0.57 ES point, about 0.0095 of ES’s daily range, against about 0.0034 on NQ — roughly three times heavier. And ES breakouts carried on less: the tiny target was hit 93% of the time on ES against 98% on NQ, while about 97% is needed to break even. Even before fees, the ES search found nothing better than fake ranges.

Is this investment advice?

No. It is a backtest of past prices for education. Futures are leveraged — one trade can cost a large part of a small account — and past results do not promise future ones.

How the test works

Prices: real 1-minute futures bars for NQ, ES, YM, RTY, gold and crude oil (and micro natural gas for the hourly test), whole sessions from 6 pm to 5 pm, 498 sessions from October 11, 2024 to October 7, 2026, split into two years at October 2025.
Older years: Nasdaq-100 ETF (QQQ) minute bars × 41.3 and S&P 500 ETF minute bars × 10.06, 2016 to October 2024 — never seen by any search.
Costs: $0.80 a side per micro contract plus one tick of slippage on market and stop orders (about $4 a round trip on MES); limit orders fill only when the price trades a tick through.
Stops and targets in ATR, the average day-session range over 14 days, so they scale with each market and year.
Fills inside a minute: when a stop and a target fall in the same bar, the worse is assumed first; the minute an entry fills is checked against every path the bar’s open, high, low and close allow.
Controls: the same orders moved to other sessions or to random minutes and sides, fake opening ranges later in the day, and rules picked on the first year then run on the second.
Dollars are per contract: MNQ is $2 a point, MES $5 a point.

Not investment advice

This is a study of past prices, published for education. Leveraged funds, options and futures can lose most of their value quickly, and a pattern seen in the past can break. Nothing here is a recommendation to buy or sell anything. Decide with your own judgment, or with a licensed adviser.