Verdict: doesn't work Backtest only

I tested 10 SPY scalping setups on second-by-second quotes. None beat the bid/ask spread.

Backtested on SPY and QQQ quotes recorded every two seconds, 41 sessions, July to September 2026. Updated 1 Oct 2026.

Short answer: no. I tested ten common scalping setups on SPY, each with seven different profit-target and stop combinations: 70 versions in all, on second-by-second quotes with real bid/ask fills. Six of the 70 had a small positive average, and none of those was distinguishable from luck. The other 64 lost.

The reason is the same almost every time: getting in and out of 100 SPY shares cost about $1.90 in bid/ask spread, and none of the setups earned more than that. Among the ten setups, the best "edge" was worth about $1 a trade before costs.

70versions tested (10 setups × 7 exits)

0meaningfully profitable after the spread

$1.90average round-trip spread on 100 SPY shares

~$1.00the best edge before costs among the 10 setups

What is scalping?

Scalping means trading for very small moves, a few cents a share, and holding for seconds or minutes. It's built on doing that many times a day, being right a little more often than wrong. SPY, the most traded ETF in the world, is the natural place to try it: it's liquid, the spread is a single cent most of the time, and it moves all day.

But when the target is ten or twenty cents, a one-cent spread on the way in and out is a big share of it.

The exact rules I tested

The common rules

Data: SPY (and QQQ, for one setup) quotes every two seconds, 41 sessions.
Trading window: signals from 09:35 to 15:45 New York time.
Fills: 100 shares, bought at the ask or sold at the bid on the quote after the signal, so no setup gets the price it saw.
Exits: seven profit-target and stop pairs, in cents per share: 15/15, 15/10, 20/10, 30/10, 30/15, 45/15 and 50/20. A trade not done in 10 minutes is closed.

The ten setups

Momentum burst: SPY moves at least twice its normal six-second move (and at least 5 cents) in six seconds; follow it.
Burst fade: the same signal, traded the other way.
VWAP breakout: a one-minute candle closes at least 5 cents across the day's VWAP; follow the cross.
VWAP reversion: price stretches two standard deviations from VWAP; fade it.
Bollinger fade: price pierces the 20-period, 2-standard-deviation Bollinger band on one-minute candles; fade it.
Order book imbalance: the quoted size on one side is at least 80% of the total for three ticks in a row; go with the heavier side.
Round-number fade: the first touch of a whole-dollar price after approaching from at least 25 cents away; fade it.
Round-number break: the same touch, traded as a breakout.
QQQ lead, SPY lagging: QQQ makes a burst and SPY hasn't followed yet; trade SPY the same way.
QQQ lead, any: the same, whether or not SPY has started moving.

Results

Setup	Signals	Best version, $ per trade	Versions with a positive average
Momentum burst	8,872	-1.01	0 of 7
Burst fade	8,919	-2.46	0 of 7
VWAP breakout	288	-1.46	0 of 7
VWAP reversion	173	+1.34	2 of 7
Bollinger fade	1,022	+0.93	1 of 7
Order book imbalance	3,392	-1.97	0 of 7
Round-number fade	73	-2.29	0 of 7
Round-number break	73	+2.34	3 of 7
QQQ lead, SPY lagging	2,162	-2.11	0 of 7
QQQ lead, any	8,570	-1.69	0 of 7

Per 100 shares, after the bid/ask spread. The three setups with a positive best version had few signals and were nowhere near statistically meaningful: with 70 versions, a handful will look positive by chance.

The fixes I tried afterwards all made things worse or changed nothing:

Trend filters: taking only trades in the trend's direction (above or below the open, VWAP, the last 15, 30 or 60 minutes, moving averages) lost more than taking every signal.
Limit orders instead of paying the spread: worse. The signals that filled were mostly the losers; the best ones ran away before my order was filled.
Fading stretched moves (new 10-minute highs, staircases of higher highs and lows, 5-minute swing points): some had a real edge before costs, and the best of them cleared the spread by about 50 cents a trade, but that was small, inconsistent and not statistically meaningful.

Why it failed

The spread is bigger than the edge. Of the ten setups, the momentum burst was the most promising: about +$1 a trade at the midpoint, before costs. But crossing the spread cost $1.90. Every other setup was weaker.
Fading is just momentum backwards. When following a burst has a small edge, fading it has the same edge the other way, before paying the spread twice.
Small targets need big win rates. With targets of 15 to 50 cents and a cent or two of cost each way, even a fair coin flip loses steadily. Win rates ran from about 26% to 44%.
Hindsight is the only filter that worked. Knowing the next 30 minutes' direction in advance made the setups profitable. No filter available at the moment of the trade did.

Did I test it wrong?

Quite possibly. I tested the version of SPY scalping described above, with settings I thought were fair, and it didn't work. That doesn't prove every version fails. I may have picked the wrong settings, the wrong market or timeframe, or missed a filter that experienced traders take for granted. Many strategies also depend on judgement that a set of fixed rules can't capture.

If you trade SPY scalping and it works for you, I'd genuinely like to hear how. Tell me the chart and timeframe, what counts as a setup, any filters, where you get in, where your stop and target go, and when you get out. I'll test your version on the same data, the same way, and publish the result here, whichever way it comes out.

Send me your version

What would change my mind

A scalping setup with an edge clearly larger than about $2 per 100 shares per trade before costs, on tick data, that holds up across months. Or a way to get filled inside the spread without picking up mostly the losing trades.

Questions

What is scalping?

Trading for very small moves, often a few cents a share, and holding for seconds to minutes. It depends on being right slightly more often than wrong, or winning slightly more than you lose.

Why is scalping SPY so hard?

Because the bid/ask spread is a large share of the move you're trying to catch. On 100 shares it cost about $1.90 to get in and out, and no setup I tested earned more than that.

Does adding a trend filter help scalping?

In my tests it made things worse. Taking only trades in the direction of the trend, by any of several definitions, lost more than taking every signal.

What about using limit orders to avoid the spread?

That was worse too. The signals that filled were mostly the ones that went against me; the best ones moved away before my order was filled.

How it was tested

Data: SPY and QQQ bid/ask quotes recorded about every two seconds, 31 July to 28 September 2026 (41 sessions). Fills on the quote after each signal. No commissions.
Backtest, not live trading: these results come from historical prices with trading costs included. Backtests usually flatter a strategy, so a failed backtest is a strong "no".

This page records my own testing. It isn't financial advice.

Disclosures · Privacy · [email protected]