Best indicator for scalping: what a futures test shows after costs

Contents

1. Is there a best indicator for scalping?
2. Which indicators do scalpers use most, and what does each measure?
3. Which bar size should you use for scalping?
4. VWAP or EMA: which one suits scalping?
5. How much do commission and slippage cut a scalping trade?
6. What does a real test of a standard indicator rule show after costs?
7. How do you check a scalping signal is not overfit?
8. Frequently asked questions
9. References

Is there a best indicator for scalping?

There is no best indicator for scalping on its own. The only fair ranking is a backtest of the complete rule on futures with commission and slippage included. At scalping trade sizes, where each trade aims for a few ticks, the round-trip cost usually decides the result before the indicator does.

Scalping is a style of trading that aims to take a few ticks of profit from each trade and holds a position for seconds to a few minutes. An indicator is only one part of a scalping rule. The rule also needs an entry price, an exit, a bar size, a session and a cost model. Two traders can use the same indicator with different exits and get opposite results. So the question "which indicator is best" has no answer until the whole rule is written down and tested.

Think of an indicator as one ingredient in a recipe. Asking for the best ingredient does not tell you whether the dish tastes good. You have to cook the full recipe and taste it. For a scalping rule, "tasting" means a backtest where every trade pays commission and slippage, and the first number to read is the average trade after those costs. The free futures contract calculator turns a target of a few ticks into dollars so you can set that number against your round-trip cost before you test anything.

The same logic applies to slower intraday rules, and the article on what testing actually shows about the best intraday indicator walks through it on longer bars. Scalping is the extreme case: the smaller the move you aim for, the larger the share of it that costs take.

Which indicators do scalpers use most, and what does each measure?

Moving averages, VWAP, RSI and Bollinger Bands are the four indicators scalpers use most, and each answers a different question about price, so they are candidates to test rather than a ranking to trust. One tells you direction, one tells you the session's average price, one tells you how stretched recent gains are against recent losses, and one tells you how wide the recent range is.

	Moving averages	VWAP	RSI	Bollinger Bands
What it measures	Direction of recent price	Average price of the session, weighted by volume	Stretch of recent gains against recent losses	A band around a moving average set by recent volatility
Typical scalping use	Trend filter: trade only in the direction of the slope	Pullback to fair value: enter when price returns to VWAP	Reversal at an extreme reading	Entry or exit at a band touch
What breaks it in a scalp	Lag: the turn shows after the move	Resets at the session open, so early readings are thin	Stays at an extreme for a long time in a strong trend	The band widens after the move, not before it

Because the four measure different things, a fair test compares complete rules, not indicator names. A moving average trend filter combined with an RSI pullback is a different rule from either indicator alone, and each combination needs its own backtest. The broader list of technical analysis tools and indicators you can actually test explains which ones can be written as exact rules and which cannot. Oscillators built for faster signals, such as the one covered in the WaveTrend settings and cross-signal test, follow the same pattern: the settings and the exit decide more than the indicator's name.

Which bar size should you use for scalping?

Pick the bar size by how many bars a signal needs to form, how many trades it produces per session and how many ticks of move it can capture per trade. A 1-minute chart, a tick chart and a range chart give very different answers to all three at once, so the bar size is a setting to test, not a tradition to follow.

A time bar closes when the clock says so; a tick bar closes after a fixed number of trades; a range bar closes when price has moved a fixed distance. For a scalper the differences matter in three ways: trade count, how signals cluster around news, and how realistic the backtest fill is.

	1-minute bars	Tick bars (for example 500-tick)	Range bars
Bars per regular session (illustrative)	Fixed: 390 bars in a six-and-a-half-hour session	Varies with activity: many bars on a busy day, few on a quiet one	Varies with volatility: many bars when price moves, almost none when it stalls
How signals cluster around news	One bar per minute whatever happens; a news spike sits inside one or two bars	Many bars form in seconds, so one spike can fire several signals	Several bars form in seconds, each a full range step
How a backtest fill is modelled	At the next bar's open, a price that existed	At the next bar's open, often a fraction of a second later	At the next bar's open, which sits exactly one range step from the last close
Is next-bar execution realistic?	Yes, with one tick of slippage	Mostly, but the next open may arrive before an order could	Only if you accept that the open is the boundary price, which may have traded for an instant

A practical way to choose is to work backwards from ticks. Decide how many ticks a single scalp can realistically capture after costs, then pick the bar size on which a completed bar moves about that distance. A bar that moves two ticks on average cannot feed a rule that targets eight ticks without holding through several bars, and a rule that holds through several bars is no longer a scalp. Whatever you pick, test the same rule on at least one other bar size. A signal that works on 1-minute bars and fails on 2-minute bars is describing the bar boundaries, not the market.

VWAP or EMA: which one suits scalping?

VWAP anchors to the session and resets each day, while an EMA follows recent price at any length you choose, so the choice is about the kind of signal you want to test, and either one only means something after costs. Neither is better; they produce different signals from the same bars.

VWAP (volume-weighted average price) is the average price paid during the session so far, weighted by the volume traded at each price. An EMA (exponential moving average) is an average of recent closes that gives more weight to the newest bars. The two differ in what they are anchored to. VWAP is anchored to the session open, so at 09:35 it reflects five minutes of trading and by the afternoon it reflects the whole day. An EMA is anchored to its length: a 20-bar EMA at any time of day reflects roughly the last 20 bars.

	VWAP	EMA
Anchor	Session open; resets every day	Chosen length; no reset
Settings to test	Almost none (the session definition)	Length, and often two lengths for a crossover
Typical scalping signal	Price pulls back to VWAP and holds	Price crosses the EMA, or a fast EMA crosses a slow one
Weak spot in a scalp	Early session readings are built on little volume	Lag: a crossover confirms a move that has partly happened
Overfitting risk	Low on the indicator itself; it lives in the exit and filters	Higher: every length is one more thing to tune

For a test, the difference in settings matters as much as the difference in signals. A VWAP pullback rule has almost no indicator settings, so most of its tuning sits in the exit and the entry filter. An EMA rule has at least one length to choose, and a crossover has two, which gives the optimizer more room to fit noise. If the lag of an EMA is what bothers you, the DEMA crossover test that shows how it cuts EMA lag is the natural next variant, with the same caution that less lag also means more false crosses on fast bars.

How much do commission and slippage cut a scalping trade?

A scalp aims for a few ticks of profit, so a round trip of commission plus one tick of slippage on each side can be larger than the gross average trade. This is why the average trade must be measured against the round-trip cost before any other number in the report is read.

Slippage means the gap between the price the backtest assumed and the price a real order would actually have received. For market and stop orders on a liquid future, one tick per side is a common starting assumption. Commission and exchange fees are charged on entry and on exit. Put together, the only formula this article needs is:

net average trade = gross average trade − commission and fees per round trip − slippage per round trip

Here gross average trade is the average result per trade before any cost, commission and fees per round trip is what the broker and exchange charge for one entry and one exit, and slippage per round trip is the ticks lost on entry plus the ticks lost on exit, in dollars.

Tick values in the example follow the exchange's published contract specification for the E-mini Nasdaq-100 future, where one tick is 0.25 index points worth $5.00. Exchanges can change tick sizes, fees and contract terms, so check the exchange's contract page for the market you test; the futures contract calculator linked in the first section shows the current tick and point values.

The lesson is about ratios, not about one market. A rule whose gross average trade is only two or three times its round-trip cost depends on cheap fills, and a small change in slippage decides whether it wins or loses. A rule whose gross average trade is many times the cost can absorb a bad fill. This is also why costs belong in the search from the first run, not in a check at the end: a search that ranks candidates before costs will favour the rules with the most and smallest trades, exactly the ones costs remove first. AlphaPrime (Valynox) charges the commission and slippage you set inside every evaluation of its backtesting, with next-bar execution, so a scalp that only works on free fills never reaches the top of the candidate list.

What does a real test of a standard indicator rule show after costs?

The run behind this article is not a scalping test. It tested a stand-in rule built from a standard indicator on ES daily bars, and it is here to show what a report after commission and slippage looks like, and why one candidate kept out of many is flattered by the selection itself.

The run used ES daily bars from 2019-01-02 to 2025-06-30, long and short, one contract, with $20 commission and $25 slippage per trade. The rule was an Information Entropy (Variant 2) rule as the only entry and exit, with no stop, no target and no filters. It is a stand-in for the subject of this article, not a scalping signal: daily bars and 44 trades over six and a half years are the opposite of a scalp. The search ran 1,024 candidates and kept only this one, judged on return relative to drawdown.

Two things in this report carry over to scalping. First, the average trade of $2,922.33 is measured after $45 of commission and slippage per trade, so costs are a small fraction of each trade on daily bars. A scalping rule inverts that ratio, which is the whole point of the previous section. Second, the net profit of $128,582.5 against a max drawdown of $55,412.5 belongs to the one candidate kept out of 1,024. Picking the best after the fact flatters it: among a thousand variants of any rule, some will look this good on one history by chance alone, and the report cannot tell you which kind this one is. The rule did not earn that money, and no reader should expect it; the number is a description of one history, not a forecast.

What you can take from the report is the way to read one. Net profit is only meaningful next to drawdown. The trade count tells you how much weight the average trade can bear: 44 trades is a small sample, and a few trades swapped would move the average a long way. A scalping report with several thousand trades has the opposite problem: the average trade is stable, but it is tiny, and the question becomes whether the cost model is right to within a fraction of a tick.

How do you check a scalping signal is not overfit?

Run the same checks in the same order every time, stop at the first failure, and write the pass rule before you look at any result. A scalping backtest with thousands of trades can produce a smooth equity curve and still describe noise, because many small trades average out the randomness in each one without removing it from the rule.

Overfitting is tuning a rule so closely to one stretch of history that it fits the noise in that stretch rather than a repeatable pattern. Bailey, Borwein, López de Prado and Zhu (2014) show that the best result from a large search is inflated by the search itself, and that the inflation grows with the number of variants tried. Sullivan, Timmermann and White (1999) found the same effect when they tested a large universe of technical trading rules: the best rule in-sample looked far less impressive once the size of the search was taken into account. A scalping search is a large search by nature, because every indicator length, bar size and exit distance multiplies the candidates.

1. Search with costs in. Commission, slippage and next-bar execution are set before the first candidate is evaluated, so the ranking already reflects what can be traded.
2. Split by date. Tune on the earlier part of the history and run the rule unchanged on the later part. Write down the split date and the acceptable drop in average trade before you run it.
3. Rerun with doubled slippage. A scalp that turns negative with one more tick per side was depending on the fill, not the signal.
4. Test the neighbouring settings. If a 9-bar EMA passes and 8 and 10 fail, the setting is a lucky spike. A plausible edge sits on a plateau where neighbours behave alike.
5. Remove the top five winners. A scalping rule should not depend on a handful of trades; if the result falls below zero without them, the edge was a few news spikes.

The order matters because each check spends some of the data's information. Running the cheap checks first (costs, doubled slippage) rejects most candidates before they touch the out-of-sample period, which keeps the number of rules that are compared on that period small. The pass rule written in advance is what stops the last step from becoming "it looks fine to me". Nothing in this process proves that a rule works; it removes the rules that can be shown not to.

A concrete next step: take the scalping rule you trade or plan to trade, write its gross average trade and your real round-trip cost side by side, and rerun the backtest with one extra tick of slippage per side. If the average trade after that rerun is still positive and still several times the cost, move on to the date split with a pass threshold you have written down first.

Frequently asked questions

Which indicator do most scalpers use?

Moving averages and VWAP are the most common, usually as a direction or fair-value reference, with RSI or Bollinger Bands added for entry timing. Popularity says nothing about results. The same indicator produces different outcomes with different exits, bar sizes and cost settings, so the only useful comparison is between complete rules tested after commission and slippage.

What time frame is best for scalping?

There is no single best time frame. Choose the bar size from the number of ticks one trade can realistically capture after costs, then pick bars that move about that distance when they complete. One-minute bars give a fixed count of bars per session; tick and range bars give more bars when the market is active and fewer when it is quiet. Test the same rule on a second bar size before trusting it.

Can you scalp with RSI alone?

RSI alone can be written as a scalping rule, but its weak spot is that it stays at an extreme reading for a long time in a strong trend, so a reversal rule keeps selling into rising prices. Most tested RSI scalping rules pair it with a direction filter and an exit distance. Whether the combination holds up is decided by the average trade after costs on your own data.

Why does a scalping backtest look good but lose in live trading?

Usually because the backtest filled trades at prices a real order would not get. Filling at the signal bar's close, assuming every limit order fills when price touches it, or leaving out one tick of slippage per side all inflate a scalp more than any other strategy type, because the gross profit per trade is only a few ticks. Rerun the test with next-bar execution and doubled slippage; if the result collapses, the live result was the honest one.

References

• Bailey, D. H., Borwein, J., López de Prado, M., & Zhu, Q. J. (2014). Pseudo-Mathematics and Financial Charlatanism: The Effects of Backtest Overfitting on Out-of-Sample Performance. Notices of the American Mathematical Society, 61(5), 458–471.
• Sullivan, R., Timmermann, A., & White, H. (1999). Data-Snooping, Technical Trading Rule Performance, and the Bootstrap. The Journal of Finance, 54(5), 1647–1691.
• CME Group. E-mini Nasdaq-100 Futures Contract Specs. https://www.cmegroup.com/markets/equities/nasdaq/e-mini-nasdaq-100.contractSpecs.html