SMT Divergence: What It Means and How to Read It on NQ and ES

SMT divergence is when two markets that normally move together disagree at a swing high or low. One makes a new high (or a new low) and the other fails to. On NQ and ES, it looks like Nasdaq futures trading through the morning low while S&P futures hold above theirs. ICT traders use it to confirm that a stop run might be the turn.

I build a chart grader, so before writing this I read the 19 sources that rank for SMT divergence. Nearly every one sells something: an indicator, a course, a scanner, a brokerage account or a prop challenge. SnapPChart doesn't detect SMT, so there's no detector to push here. The last section says what it does and doesn't do with an SMT trade.

Quick Answer: What Is SMT Divergence?

The classic pairs are ES vs NQ for index futures, EUR/USD vs GBP/USD for forex, and EUR/USD vs the Dollar Index with the logic flipped. Gold vs silver and BTC vs ETH come up too. Nobody has published a win rate for any of it.

What Does SMT Stand For?

SMT is usually expanded as Smart Money Technique. You'll also see Smart Money Tool, and the odd Smart Money Trading. None of the sources I read quotes ICT's own material for it, so I won't pretend to know which came first.

It comes from the Inner Circle Trader, the ICT teaching associated with Michael J. Huddleston, and it sits next to the rest of the ICT vocabulary, like liquidity sweeps, fair value gaps, order blocks and kill zones. If those terms are new, this walkthrough of the ICT concepts TJR teaches covers them first. Some sources give SMT a start year, and one says it first caught on in forex. I couldn't verify either, so there's no origin date here.

The "smart money" part is an interpretation

The name carries a story. In the ICT reading, large players push one market through an obvious high or low to trigger the stops parked there and fill their own orders, and the correlated market refusing to follow is the tell. None of the 19 sources shows evidence that any particular sweep was engineered. A divergence can also come from sector rotation (the Nasdaq-100 leans much harder on tech than the S&P 500), one heavyweight stock moving on its own news, or a headline that hits one market harder. You can trade the pattern without buying the story.

SMT is a different idea from RSI or MACD divergence

Same word, different comparison. RSI and MACD divergence compares price with an oscillator on one chart, like price making a higher high while RSI makes a lower high. SMT compares the prices of two different markets, with no indicator involved. When both show up at the same high, they're two separate reads that happen to agree.

Bullish vs Bearish SMT Divergence

The two versions mirror each other. A couple of the ranking pages define them backwards in one place and correctly in another, so check any cheat sheet you save against the version below.

Bearish SMT divergence

It forms at a swing high. One market makes a higher high, usually by taking out an obvious prior high. The other makes a lower high, or fails to reach its own prior high. Say NQ trades above yesterday's high while ES, in the same 15 minutes, tops out a few points under its own.

Bullish SMT divergence

It forms at a swing low. One market makes a lower low, sweeping the prior low, and the other makes a higher low or holds above its own. The diagram shows this case on NQ and ES, as a pullback inside an uptrend.

Both setups side by side, carried through to the trade plan most ICT sources attach to them.

The last two rows matter more than they look. The same pattern is a pullback signal in one spot and a top-picking call in another, and most sources quietly prefer the with-trend one.

How to mark it so it counts

Both swings have to form at roughly the same time, on the same timeframe. A Monday NQ high against a Tuesday ES high is two unrelated swings. Put the charts side by side, or overlay the second symbol on the first, which TradingView and most other charting platforms can do.

Count only confirmed swing points, with a few bars printed on either side. Sources split on whether the wick or the close defines the extreme. That's a definition, not a fact, so pick one and use it on both charts.

Which Markets Can You Compare for SMT?

Any two markets that normally move together, or reliably move opposite, can print SMT. The index correlations below come from a CME Group whitepaper on stock index spread opportunities, computed from weekly spot index values between January 2007 and April 2014, so they're background, not a live reading. The Dollar Index weights are from ICE's U.S. Dollar Index contracts FAQ, dated 2015. ICE says the index has no regularly scheduled rebalancing and was adjusted once, when the euro arrived.

Two details in that table that most SMT guides skip. The euro is 57.6% of the Dollar Index and the pound only 11.9%, so EUR/USD against DXY is close to mechanical while GBP/USD against DXY is a much looser link. And in CME's sample the S&P 500 and the Dow (0.979) tracked tighter than the S&P 500 and the Nasdaq-100 (0.924). NQ vs ES is the popular pair. It isn't the tightest one, and some SMT tools default to ES vs YM instead.

Inverse pairs flip the second chart

When EUR/USD makes a new high, the dollar is weakening, so DXY should make a new low at the same time. If DXY holds a higher low instead, the dollar didn't confirm the euro's move, and that's bearish SMT on EUR/USD. Flip DXY in your head and it reads like any other pair. The EUR/USD setup grading guide covers the forex side on a single chart, and forex trading sessions covers when London and New York overlap.

Triple SMT and the other pairs

Triple SMT adds a third market. One index takes its high while the other two fail (ES, NQ and YM), or EUR/USD sweeps while GBP/USD and DXY don't confirm. ICT traders read more markets disagreeing as a stronger signal. Nobody has measured whether it is.

You'll also see crude (WTI vs Brent), 10-year vs 30-year Treasury futures, and AUD/USD vs NZD/USD. Same logic. For metals, the gold trading strategy guide and this primer on trading silver cover each side of that pair. If you trade the indexes as CFDs (US500, US100) rather than futures, the guide on how to trade indices covers the differences.

Whatever the pair, the correlation has to hold now, on the timeframe you're reading. A correlation figure is an average over years, and it hides the weeks when two markets go their own way. Loose or drifting pairs print divergences constantly, and nothing says those mean anything.

How to Read SMT Divergence on NQ and ES

This is the sequence most ICT sources describe, in the order you'd check it. Most of the work happens before the sweep, deciding whether a divergence would even matter. If NQ and ES are new to you, start with how futures trading works.

1. Set the higher-timeframe bias. On the 1-hour, 4-hour or daily, decide which way the market is trending. An SMT that lines up with that bias is a with-trend read. One that fights it is a counter-trend read, which most sources treat as weaker. The multi-timeframe analysis guide covers setting the bias without overthinking it.
2. Mark the levels worth sweeping. The previous day high and low, the overnight and session highs and lows, equal highs or equal lows, and any higher-timeframe order block or fair value gap. SMT counts at those levels. In the middle of the range it's noise. For picking levels, see support and resistance basics.
3. Watch the session. Most sources give the most weight to SMT around the London and New York opens (the ICT kill zones) and little to SMT during the Asian session. ICT sources print different clock times for those windows, so I'm not adding a third set. The ICT concepts section of the TJR guide explains the kill zones, and the guide to futures trading hours covers when NQ and ES actually trade.
4. Line the charts up. Same timeframe, same window, same contract month. NQ and ES are quarterly contracts, and around the roll a continuous chart can jump where it switches months. NQ and MNQ are quoted in the same index points (so are ES and MES), so the swing levels match whichever size you chart. Micro futures, explained has the contract details.
5. Spot the divergence. One market sweeps the level and the other doesn't. That's the SMT. It's the same stop-run idea behind a liquidity sweep setup, with a second market as the witness. The post on stop-loss hunting explains why stops cluster where they do.
6. Wait for the structure shift. On the lower timeframe (1 to 15 minutes), wait for price to break the last swing against the sweep. After a bullish SMT that means breaking the last lower high. Sources call it a market structure shift (MSS), a change of character (CHoCH), a change in state of delivery (CISD) or displacement. Different names, same step. If the labels blur together, reading price action walks through swing structure.
7. Enter on the retrace. The structure shift usually leaves a fair value gap or an order block behind it, and the entry most sources describe is the retrace into one of them. The fair value gap strategy and order block strategy pages cover what makes each one valid.
8. Put the stop beyond the swept extreme. Below the sweep low on a bullish SMT, above the sweep high on a bearish one. If price gets back past that extreme, the read was wrong. In dollars, using CME's contract specs, NQ is $20 per index point, so a 0.25-point tick is $5.00 (see the E-mini Nasdaq-100 contract specs). ES is $50 per point ($12.50 a tick, per the E-mini S&P 500 contract specs). MNQ is $2 per point ($0.50 a tick, per the Micro E-mini Nasdaq-100 contract specs). A 25-point NQ stop is $500 on one NQ contract and $50on one MNQ. That's arithmetic, not a suggested stop size.
9. Target the opposing liquidity. The other side of the range, the opposing session high or low, or the next obvious pool of stops. Check the reward to risk before you take it. A 25-point stop aimed at a target 20 points away is a bad trade however clean the divergence looked.

Which chart do you actually trade?

Sources disagree, and most don't say so. Some enter on the market that swept. Some buy the market that held (the stronger one) after a bullish SMT and short the one that failed (the weaker one) after a bearish SMT. Others enter from the fair value gap on whichever market is lagging. No data settles it. The rule that holds either way is to put the stop and the target on the structure of the chart you're trading. Trade ES after NQ swept and the stop goes beyond ES's own low, a different price and a different distance.

Which timeframe?

The common ground is a split. The 1-hour, 4-hour or daily sets the bias and the levels, and 1 to 15 minutes handles the execution. Where the SMT itself should be read is contested. Some sources say 15 minutes or lower, because higher-timeframe SMT takes too long to form. Others say 4-hour and daily SMT carry more weight and the lower timeframes throw off more false signals. Neither camp has published data.

Do you need an SMT indicator?

TradingView scripts, and indicators for the other common futures and forex platforms, can mark SMT for you. That helps, because watching two charts at once is tiring and the trigger can come fast around the open. None of them can tell you whether the divergence sits at a level that matters or whether the correlation still holds this week.

Where Does SMT Divergence Fail?

Plenty of SMT divergences go nowhere. Not every one reverses, and the ways they fail are predictable enough to list. Two are already covered above (mid-range SMT and quiet-session SMT), so here are the rest.

• It just doesn't reverse
  Both markets make another high and the divergence is gone. SMT never said price would turn. It said one market was weaker at that moment.
• The divergence persists
  Markets can disagree for days, and SMT says nothing about when the turn comes or at what price. Even a tight pair drifts apart. In CME's whitepaper the S&P 500 and Nasdaq-100 correlated at 0.924 on weekly data, yet over the three months to April 30, 2014 the S&P 500 returned +1.58% (total return) and the Nasdaq-100 -2.98%. Over that quarter they finished on opposite sides of zero.
• The correlation drifted
  If the two markets haven't been tracking each other lately, a divergence between them is just two markets doing different things.
• It printed on the news
  A CPI or NFP release can hit one market harder for a few minutes and print a divergence that means nothing. Some traders read news-time SMT as a sign the move is fake. Neither reading has data behind it, so treat it as lower quality and check the calendar first.
• It fights the higher timeframe
  A bearish SMT at the top of a strong uptrend is a top-picking trade. Sometimes it works. The trend can also just resume and run the short over.
• The swings weren't the same swing
  Different windows, different timeframes, or different contract months around the roll. Mismatched charts invent divergences that never happened.

Failed highs and lows cut both ways, and the bull trap and bear trap guide walks through what happens to traders caught on the wrong side of one.

Is There Any Evidence SMT Divergence Works?

Not publicly. Of the 19 ranking sources I read, none quotes a win rate, and two say outright that no statistic exists. There's no control group and no published backtest with costs. Phrases like "high probability" show up anyway, and one page says a three-way divergence makes a reversal more likely, all with no number behind them. This page makes no probability claims for the same reason.

What does exist is the correlation itself, plus evidence that it breaks. CME's numbers say NQ and ES usually move together, and the 2014 quarter shows that "usually" has real exceptions.

So treat SMT as a filter you test yourself. Log every SMT you see for a month: the pair, the level, the session, with or against the higher-timeframe trend, whether a structure shift followed, and what happened next. That's the only win rate that applies to how you trade it. Testing a rule on history and checking one trade before you take it are different jobs, covered in backtesting vs grading each setup before entry.

SMT works like any other confluence factor, one more reason and never the whole reason. If you score setups, the six ICT and SMC setup criteria (the sweep, the gap, the order block, the kill zone and the rest) are where an SMT read would slot in.

Can AI Detect SMT Divergence From a Chart Screenshot?

Not SnapPChart, and I'd rather say that plainly than let you find out. It grades one uploaded chart screenshot of one instrument and never sees a second market. Upload NQ and it doesn't fetch ES, check DXY or compare anything, and a split-screen screenshot of two symbols isn't a supported input. It doesn't detect SMT, doesn't alert you when one forms, and doesn't tell you when an SMT entry triggers.

From that one chart, it does read most of the ICT vocabulary around an SMT trade. The liquidity sweep (price spiking through a level and snapping back on a wick, with the level's price and side). Fair value gaps. Order blocks. Swing structure, including break of structure vs change of character. And the session kill zone, but only when the time axis or session shading makes the time readable. It doesn't guess the session. Those write-ups are display-only. They show next to the grade and are stripped out before grading, so the sweep write-up never lifts the grade and a missing one never drags it down. The grade still scores the trend itself, higher highs and higher lows (or lower highs and lower lows) on the visible chart. It just doesn't score these write-ups. This piece on what AI can read in smart money concepts goes further into that.

The grade itself covers with-trend continuation setups only: a pullback in an uptrend (long) or a rally in a downtrend (short). Every SMT guide I read pitches it as a reversal tool. A bearish SMT at the top of an established uptrend is a reversal short, which is outside what SnapPChart grades. By default it reads that chart as a long that is stopping out, not as a short. The SMT that lines up with the grade is the with-trend one, a bullish SMT at a pullback low inside an uptrend or the bearish mirror at a rally high inside a downtrend. That's the same continuation logic the momentum trading strategy is built on.

One catch. The grader decides the trend from what's on the screenshot. It never sees your 1-hour bias. So the chart you upload has to show the uptrend as well as the pullback, the way the diagram above keeps the higher highs and higher lows in view before the sweep. (The diagram is there to show the trend context, not a shape that would grade well: a third pullback that gives back most of the leg counts against a setup.) Crop it to just the pullback leg, like a 1-minute chart of only the worked example's morning drop, and all that's left is lower highs and lower lows. The grader reads that as a short that's covering, not as your long.

So pairing it with SMT looks like this.

1. Check the correlated chart yourself.ES when you're trading NQ, GBP/USD or DXY when you're trading EUR/USD. Same timeframe, same window.
2. Upload the chart you intend to trade, zoomed out to show the trend. Only if the other market failed to confirm. Keep the higher highs and higher lows before the pullback on the screenshot (for a short, the lower highs and lower lows before the rally).
3. Read what comes back. SnapPChart shows the sweep, gap, order block, structure and (if the axis shows it) the kill zone on that chart beside the grade, and grades the setup as a pullback in whichever trend that screenshot shows.
4. Keep the SMT call yours.Whether ES confirmed is something the grader doesn't know, and it won't pretend to.

Two more things to know. Futures, index, forex, metals and crypto charts have their grade capped at B, so a clean NQ pullback earns a B at most. Your first grade is free and shows the setup's grade and the general read. Entry, stop, targets and the trade plan come with a paid plan.

For the wider picture, there's more on what AI chart analysis reads off a screenshot, the AI trading guide's section on the limits of AI on a chart, and a comparison of AI futures trading tools for NQ and ES traders weighing the options.