SMT Divergence, Using Correlated Pairs as a Lie Detector
SMT divergence is a technique where you compare two correlated instruments to spot weakness or strength that a single chart might hide.
The basic idea: correlated pairs, such as EURUSD and GBPUSD, normally make similar highs and lows. If one makes a new extreme and the other fails to, it may signal that the move lacks real strength.
Example (bearish):
- EURUSD makes a higher high, sweeping a previous high
- GBPUSD fails to make a higher high
- This divergence suggests the sweep might be a trap, and a reversal could follow
Example (bullish):
- EURUSD makes a lower low
- GBPUSD holds above its previous low
- This divergence may signal weakness in the sell-off
How to use it:
1. Choose correlated pairs. EURUSD and GBPUSD, or a pair and the dollar index (inverted).
2. Mark key highs and lows on both.
3. Look for divergence at liquidity levels. It carries more weight at meaningful zones.
4. Wait for confirmation. A structure shift on your entry timeframe.
5. Manage risk normally.
Warnings:
- Divergence alone is not a signal
- Correlations can change
- Timing can be imperfect
Think of SMT as a second opinion. When two related markets disagree, it's worth paying attention.
Add it as a confluence factor, not as a standalone strategy.
SMT Divergence, Using Correlated Pairs as a Lie Detector
-
PropScalpDesk
- Posts: 521
- Joined: Sat Sep 19, 2026 7:50 pm
Re: SMT Divergence, Using Correlated Pairs as a Lie Detector
The part that trips people with SMT is timing. Compare the two pairs on the same candle, on the same chart timeframe. EURUSD taking out its high at 08:20 while GBPUSD made a lower high at 08:05 is not a divergence, it's two different moments. I keep both charts synced to the same crosshair so I'm reading the same bar.
Also compare structure, not size. GBPUSD normally moves more pips than EURUSD, so "failed to make a higher high" means it stayed below the earlier swing, not that it moved less.
On the warning about correlations changing: in my experience it happens predictably on UK data mornings. On UK CPI or a Bank of England day, cable trades its own story and a divergence against EURUSD mostly reflects the release, not weakness in the dollar move. I switch the check off on those days and on euro-specific data days for the same reason.
Also compare structure, not size. GBPUSD normally moves more pips than EURUSD, so "failed to make a higher high" means it stayed below the earlier swing, not that it moved less.
On the warning about correlations changing: in my experience it happens predictably on UK data mornings. On UK CPI or a Bank of England day, cable trades its own story and a divergence against EURUSD mostly reflects the release, not weakness in the dollar move. I switch the check off on those days and on euro-specific data days for the same reason.
-
LondonNewsTrader
- Posts: 421
- Joined: Mon Sep 21, 2026 9:30 am
Re: SMT Divergence, Using Correlated Pairs as a Lie Detector
Before reading SMT as a trap, I check whether one of the two currencies had its own news. EURUSD and GBPUSD share the dollar leg, so when they disagree the difference is the euro or the pound. If UK CPI came out at 7:00 London time and was soft, GBPUSD failing to make a new high while EURUSD does is not hidden weakness, it is a different driver. The divergence then has nothing to do with liquidity and the reversal you expect may never come.
The same thing happens with ECB speakers, German data or BoE minutes. On those days the two pairs come apart for a while, and the cause is already on the calendar.
Where I've found it more useful is on US data. If both pairs react to the same dollar number and one of them refuses to follow through, that is a cleaner signal, because both had the same input and only one responded.
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
Re: SMT Divergence, Using Correlated Pairs as a Lie Detector
LondonNewsTrader's US data case is the only version of SMT I still trade, and the detail that matters most to me is which of the two pairs you actually sell. I short the one that failed to make the new high, because it is already showing the weakness and its stop can sit above a lower swing.
The chart is a made up M5 afternoon after a 13:30 US release, with EURUSD sweeping its 12:50 high of 1.0872 at 13:45 while GBPUSD stalls at 1.2731, below its own 1.2740 high. Entry on cable comes on the M5 break of the 13:35 low at 1.2712, with the stop at 1.2734 just above the lower high.
That gives a 22 pip risk on GBPUSD instead of the 30 or so you would need above the swept EURUSD high. Same idea, better location.
I would loosen PropScalpDesk's same candle rule slightly on M1, where the two feeds can print their extremes one bar apart. On M5 I hold it strictly.
One case I skip is when both pairs are already more than 40 pips into the move on the day. By then a failure to make a new high is often plain exhaustion rather than a trap.
The chart is a made up M5 afternoon after a 13:30 US release, with EURUSD sweeping its 12:50 high of 1.0872 at 13:45 while GBPUSD stalls at 1.2731, below its own 1.2740 high. Entry on cable comes on the M5 break of the 13:35 low at 1.2712, with the stop at 1.2734 just above the lower high.
That gives a 22 pip risk on GBPUSD instead of the 30 or so you would need above the swept EURUSD high. Same idea, better location.
I would loosen PropScalpDesk's same candle rule slightly on M1, where the two feeds can print their extremes one bar apart. On M5 I hold it strictly.
One case I skip is when both pairs are already more than 40 pips into the move on the day. By then a failure to make a new high is often plain exhaustion rather than a trap.
-
FXS Prop Desk
- Posts: 1691
- Joined: Wed Oct 07, 2026 9:52 am
Re: SMT Divergence, Using Correlated Pairs as a Lie Detector
One practical requirement for SMT is that both charts use the same time reference and the same session definition, since different brokers or data sources may show slightly different highs. Comparing on the same platform and the same broker feed avoids false divergences caused by data differences rather than market behaviour.
Keeping a log of SMT signals and outcomes for a month also shows quickly whether the divergence adds value on your pairs, or whether it mostly looks convincing in hindsight.
Keeping a log of SMT signals and outcomes for a month also shows quickly whether the divergence adds value on your pairs, or whether it mostly looks convincing in hindsight.
FXS Official Team | The Idea Lab & Strategy
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
Re: SMT Divergence, Using Correlated Pairs as a Lie Detector
FXS Prop Desk's same feed point goes one level further, because most platforms draw candles from the bid. A buy stop above a high triggers on the ask, so on a wide spread minute the level you think was swept may have been hit only on the ask while the bid high stayed intact.
That matters most around the US releases where I still use SMT. Spread widens for a few seconds, and one pair can print a fresh high on the ask while its bid looks like a lower high.
On Fairman's suggestion of using the inverted dollar index, I would be careful with EURUSD in particular. The euro is close to 58 percent of DXY, so a divergence between those two is rare almost by construction, and GBPUSD or USDCHF gives a more independent second opinion.
For the month of logging FXS Prop Desk suggests, I would write down every divergence that met the rules, including the ones not traded. Say a month throws up 25 and only the 9 you took get recorded, then the log mostly measures your own filter.
PropScalpDesk, do you switch the check off for the full UK data morning or only the first hour after the release? Cable sometimes settles back into step with the euro well before lunch.
That matters most around the US releases where I still use SMT. Spread widens for a few seconds, and one pair can print a fresh high on the ask while its bid looks like a lower high.
On Fairman's suggestion of using the inverted dollar index, I would be careful with EURUSD in particular. The euro is close to 58 percent of DXY, so a divergence between those two is rare almost by construction, and GBPUSD or USDCHF gives a more independent second opinion.
For the month of logging FXS Prop Desk suggests, I would write down every divergence that met the rules, including the ones not traded. Say a month throws up 25 and only the 9 you took get recorded, then the log mostly measures your own filter.
PropScalpDesk, do you switch the check off for the full UK data morning or only the first hour after the release? Cable sometimes settles back into step with the euro well before lunch.
Re: SMT Divergence, Using Correlated Pairs as a Lie Detector
The calendar check before reading a divergence is where I've been lazy. I'd see GBPUSD fail at its high while EURUSD printed a new one and call it a lie detector, when UK CPI had come out at 7:00 and the pound was just reacting to its own number. That divergence wasn't hiding anything, there was a plain reason for it.
I like your suggestion about US data. Both pairs get the same dollar input, so if the dollar number is strong and EURUSD falls 25 pips but GBPUSD only 8, that is information about the pound, perhaps because buyers are defending it. A made-up trade: after the number, GBPUSD refuses to make a new low while EURUSD does, so I'd look for a long on GBPUSD with a stop under the failed low, 9 pips, at 0.4 lots, 36 dollars at risk, and a target at the pre-release level.
For checking, I put the two charts side by side with the calendar open. If the divergence follows a release only one of the pairs has, I don't count it. If it follows a common dollar number, I do.
I have also seen the divergence work less well on a day with several overlapping events, like ECB speakers in the morning and US data in the afternoon, since both pairs have two sets of drivers by then.
Do you use a third pair, like EURGBP, to confirm which currency is driving the gap? And how long do you wait after the release before trusting a divergence?
I like your suggestion about US data. Both pairs get the same dollar input, so if the dollar number is strong and EURUSD falls 25 pips but GBPUSD only 8, that is information about the pound, perhaps because buyers are defending it. A made-up trade: after the number, GBPUSD refuses to make a new low while EURUSD does, so I'd look for a long on GBPUSD with a stop under the failed low, 9 pips, at 0.4 lots, 36 dollars at risk, and a target at the pre-release level.
For checking, I put the two charts side by side with the calendar open. If the divergence follows a release only one of the pairs has, I don't count it. If it follows a common dollar number, I do.
I have also seen the divergence work less well on a day with several overlapping events, like ECB speakers in the morning and US data in the afternoon, since both pairs have two sets of drivers by then.
Do you use a third pair, like EURGBP, to confirm which currency is driving the gap? And how long do you wait after the release before trusting a divergence?
It’s Fairman 