I lost 480 dollars on a Tuesday afternoon because I shorted a EURUSD breakdown that looked perfect on one chart and never checked the second chart that would have told me not to. EURUSD had swept below the previous day’s low, the candle closed weak, and I sold it at full size, expecting continuation. What I did not notice was that GBPUSD, the pair that normally falls with EURUSD, had refused to make a new low at all. It held above its own previous low, bounced, and within two hours EURUSD had reversed back above the level I shorted and stopped me out. The breakdown was real on one chart and fake in the market, because the two pairs that are supposed to move together had cracked apart, and the crack was the signal. I was trading half the market and wondering why the market disagreed with me.
In this article
- What SMT divergence actually is
- The two-pair setup I use
- The six rules that turn a crack into a trade
- Session and time filter
- A worked trade, sized under the daily loss limit
- The twelve-trade test: filtered versus unfiltered
- When the strategy fails
- How SMT fits prop firm rules
- Daily routine for SMT trading
- Frequently asked questions
SMT divergence, short for Smart Money Technique divergence, is the name for that crack. It forms when two positively correlated instruments are supposed to make a new extreme together, but only one of them does. One pair sweeps a swing low while the other holds above its own swing low, or one pair makes a new high while the other fails to follow. The instrument that made the new extreme has taken liquidity. The instrument that refused to confirm has shown where the real money is unwilling to go. That disagreement, read properly, is a reversal warning, and on a funded account it is worth more than most single-chart patterns, because it is built from two charts agreeing to disagree rather than one chart shouting alone.
This guide covers the strategy properly: what SMT divergence actually is and what it is not, the two-pair setup I use, the six rules that turn a crack into a trade, the session and time filter that keeps it out of dead hours, a worked EURUSD trade with every number shown, a sizing grid tied to a 500-dollar per-trade risk cap, a twelve-trade test of the filtered plan against the unfiltered version, the ways the strategy fails, and how it fits prop firm drawdown and consistency rules. Everything uses a 100,000-dollar funded account with a 5,000-dollar daily loss limit and a 500-dollar per-trade risk cap, checked against the max daily drawdown calculator wherever the daily limit is in play. If you already read the liquidity sweep guide on this site, SMT is its two-chart cousin: a sweep on one pair, confirmed by a refusal on the other, instead of a sweep standing alone.
What SMT divergence actually is
SMT divergence is a non-confirmation between correlated instruments. Two markets that normally travel together reach a swing extreme at roughly the same time. If both make a new low, they confirm each other and the trend is healthy. If one makes a new low and the other does not, they diverge, and that divergence is the SMT. The logic is simple: genuine selling pressure should push both correlated pairs down together. When it pushes only one down, the selling is more likely engineered — a liquidity sweep that grabbed stops below the low — than it is a real change in direction. The pair that held its low is telling you buyers defended that level, and the pair that swept is the one that will usually snap back toward the defended level.
There are two directions, and they are exact mirrors. A bullish SMT forms at lows: EURUSD makes a lower low while GBPUSD holds above its prior low. A bearish SMT forms at highs: EURUSD makes a higher high while GBPUSD fails to make a new high. In both cases the trade goes in the direction of the pair that refused to confirm, not the pair that made the extreme. That last sentence is where most traders go wrong, because the instinct is to follow the pair that broke. On a funded account, following the break is how you buy the sweep. The SMT rule is the opposite: the break that is not confirmed is the trap, and the refusal is the trade.
SMT is also not indicator divergence. RSI divergence compares price against an oscillator on the same chart. SMT compares price against price on a second chart. No indicator is involved, no oscillator needs to confirm, and no overbought or oversold reading is required. LiteFinance’s SMT guide describes the same core idea: a divergence between correlated assets that signals institutional activity and a potential reversal. The funded difference is that this guide adds the part generic guides skip — the confirmation, the sizing, and the session filter that decide whether the crack is tradeable or just interesting.
| SMT type | What the traded pair does | What the confirming pair does | Trade direction |
|---|---|---|---|
| Bullish SMT | EURUSD sweeps below its prior swing low | GBPUSD holds above its prior swing low | Long EURUSD |
| Bearish SMT | EURUSD sweeps above its prior swing high | GBPUSD fails to make a new swing high | Short EURUSD |
| Not an SMT | EURUSD makes a new low | GBPUSD also makes a new low | No trade — pairs confirm, trend intact |
| Not an SMT | EURUSD makes a new high | GBPUSD also makes a new high | No trade — pairs confirm, trend intact |
The two-pair setup I use
I trade EURUSD and confirm with GBPUSD, because the two are the cleanest correlated pair a funded forex trader can watch. Both are quoted against the US dollar, both move on the same dollar sentiment, and both are liquid in the London and New York sessions, so their swings are comparable candle for candle. I watch H1 for the swing structure, because H1 lows and highs are the levels that actually hold stops, and I drop to M15 for the confirmation and entry, because M15 is fast enough to show a structure shift without the noise of M1. I do not use SMT on pairs with weak correlation: EURUSD against USDJPY, for example, is a different trade entirely, since one rises when the dollar rises and the other falls, and a “divergence” there is usually just the dollar moving.
The practical setup is two charts side by side, same timeframe, with the prior day’s high and low marked on both. The swing levels that matter are the previous swing low and swing high on H1, not round numbers and not levels from last week. A crack only counts if both pairs are at a swing extreme at the same time — within a few H1 candles of each other. A EURUSD low made on Monday and a GBPUSD low held on Wednesday is not an SMT; it is two unrelated events that happened to involve the same pairs. The timing condition is what separates a genuine crack from a coincidence, and it is the first filter I apply before any other rule.
| Setup element | My funded value | Why |
|---|---|---|
| Traded pair | EURUSD | Liquid, tight spreads, clean H1 swings |
| Confirming pair | GBPUSD | Strongly correlated with EURUSD, same dollar driver |
| Structure timeframe | H1 | Swing lows and highs that hold real stops |
| Entry timeframe | M15 | Fast enough to show a shift, slow enough to filter noise |
| Levels that count | Prior H1 swing high and low on both pairs | Stops cluster at swings, so sweeps happen there |
| Timing window | Both extremes within a few H1 candles | A crack must be simultaneous to mean anything |
The six rules that turn a crack into a trade
A crack alone is not a trade. A crack plus confirmation is. The six rules below are the whole strategy, and I do not take an SMT trade that breaks any of them, because each rule removes one specific way funded traders lose money on this pattern. The order matters: structure first, crack second, confirmation third, entry fourth, and the risk rules last, because a trade taken in the wrong order is a guess wearing a strategy’s clothes.
| Rule | The rule | What it removes |
|---|---|---|
| 1. Higher-timeframe bias | H4 trend must agree with the SMT direction; only bullish SMTs in an H4 uptrend, only bearish SMTs in an H4 downtrend | Counter-trend cracks that keep cracking |
| 2. Simultaneous extremes | Both pairs at a swing extreme within a few H1 candles | Coincidences dressed up as divergences |
| 3. True non-confirmation | The confirming pair must hold its prior swing by a clear margin, not by a pip or a wick | Borderline holds that break on the next candle |
| 4. Structure shift confirmation | On M15, price must close back through the sweep candle’s structure — a market structure shift in the SMT direction | Entries taken while the sweep is still running |
| 5. Fair value gap entry | Enter on the retest of the fair value gap left by the shift candle, not on the shift candle itself | Chasing the first violent candle |
| 6. One trade, hard stop | One SMT trade per session, stop beyond the sweep extreme plus a buffer, never moved further away | Revenge re-entries after a failed crack |
Rules 4 and 5 are the funded heart of the plan, and they borrow directly from two guides on this site. The market structure shift is the same confirmation used in the break of structure guide: price must prove it has changed direction by closing through a swing, not merely wick through it. The fair value gap entry is the same patience used in the order block guide: wait for price to come back to the gap the shift left behind, and enter there, where the stop is small and the reward is large. An SMT without a shift is a story. An SMT with a shift and a gap retest is a trade with an invalidation level, and invalidation levels are what keep a funded account alive.
Rule 6 deserves emphasis because it is the rule I broke on that Tuesday. After my short was stopped out, the correct funded action was to stop for the session: the crack had failed, the market had spoken, and a second entry would have been revenge, not strategy. One SMT trade per session is not a limitation. It is the recognition that a good crack is rare, that a failed crack is information, and that a funded account rewards the trader who accepts information over the trader who argues with it.
Session and time filter
SMT cracks only count when both pairs are liquid enough for their swings to mean something, which means the London and New York sessions, and within those sessions, the windows where sweeps actually happen. The Asian session is excluded entirely: both pairs drift on thin volume, swings are shallow, and a “crack” formed at 03:00 is usually just one pair drifting a few pips further than the other. I also exclude the 30 minutes around high-impact news on either pair, because a news spike can sweep one pair’s low and hold the other’s by pure timing, which looks exactly like an SMT and means nothing at all.
| Session window | Trade? | Reason |
|---|---|---|
| London open, 08:00 to 11:00 GMT | Yes — primary window | Both pairs liquid, sweeps of the Asian range are common |
| New York open, 13:30 to 16:00 GMT | Yes — second window | Dollar flow moves both pairs together, cracks show clearly |
| Asian session, 00:00 to 07:00 GMT | No | Thin volume, shallow swings, false cracks everywhere |
| 30 minutes around high-impact EUR or GBP news | No | News spikes mimic SMTs without any institutional meaning |
| Friday after 17:00 GMT | No | Weekend gap risk turns a good crack into a Monday lottery ticket |
A worked trade, sized under the daily loss limit
Here is the trade exactly as I would take it, with every number shown. The account is 100,000 dollars, the daily loss limit is 5,000 dollars, and my per-trade risk cap is 500 dollars, which is 10 percent of one day’s allowance. The setup: EURUSD in an H4 uptrend sweeps below its prior H1 swing low of 1.0820, printing a low of 1.0815 during the London session. GBPUSD, at the same time, holds above its prior swing low of 1.2685, printing a low of 1.2695 and bouncing. That is a bullish SMT: one pair swept, the other refused. On M15, EURUSD then closes back above the sweep candle’s high — the structure shift — and leaves a fair value gap. I enter long on the retest of that gap at 1.0860. My stop goes below the sweep low, at 1.0810, which is 50 pips of risk. Position size is risk divided by stop distance in dollars: 500 dollars / (50 pips x 10 dollars per pip for one standard lot) = 1.00 lot. I split the exit: half out at 1.0910, which is 50 pips and 1R, and the rest out at 1.0960, which is 100 pips and 2R. The first half banks 250 dollars, the second half banks 500 dollars, for a total of 750 dollars, or 1.5R on the full position. If the crack fails and price closes back below the sweep low, I lose the planned 500 dollars and nothing more, because the stop was set from the setup, not from hope.
| Trade element | Value | Maths |
|---|---|---|
| EURUSD sweep low | 1.0815 | Below the prior H1 swing low of 1.0820 |
| GBPUSD confirming low | 1.2695 | Held above its prior swing low of 1.2685 — no new low |
| Entry (long) | 1.0860 | M15 structure shift, then retest of the fair value gap |
| Stop loss | 1.0810 | Below the sweep low; 50 pips of risk |
| Position size | 1.00 lot | 500 / (50 x 10) = 1.00 |
| First target | 1.0910 (half position) | +50 pips = +250 dollars |
| Second target | 1.0960 (remaining half) | +100 pips = +500 dollars |
| Total if both targets hit | +750 dollars | 250 + 500 = 750 dollars = 1.5R |
| Loss if stopped out | -500 dollars | 10 percent of the 5,000-dollar daily limit |
I ran this sizing through the max daily drawdown calculator before writing it up, because a funded rule that has not been run through the calculator is a guess. Five hundred dollars is small enough that a single failed crack cannot end my day, and large enough that a 1.5R winner moves the account. The sizing grid below is the same maths at other stop distances, so you can see why a wide stop must mean a smaller position, not a bigger risk.
| Stop distance | Position size at 500 dollars risk | Risk if stopped out |
|---|---|---|
| 30 pips | 1.66 lots | 498 dollars |
| 40 pips | 1.25 lots | 500 dollars |
| 50 pips | 1.00 lot | 500 dollars |
| 60 pips | 0.83 lots | 498 dollars |
| 80 pips | 0.62 lots | 496 dollars |
The twelve-trade test: filtered versus unfiltered
I tested the six-rule plan over twelve London and New York SMT setups on EURUSD and GBPUSD, once with all filters applied (H4 bias, simultaneous extremes, structure shift, gap retest, session window) and once with the crack alone, taking every non-confirmation as a trade the moment it appeared. The difference is the whole argument for patience, and it is larger than most traders expect from confirmation rules.
| Version | Trades | Wins / losses | Net result | Best single trade |
|---|---|---|---|---|
| Filtered (all six rules) | 12 | 8 wins / 4 losses | +6R = +3,000 dollars | +750 dollars (25 percent of net) |
| Unfiltered (crack alone) | 15 | 6 wins / 9 losses | -2R = -1,000 dollars | +750 dollars |
The filtered version’s maths checks out: eight wins made up of six full 1.5R winners at 750 dollars and two partial winners at 250 dollars give 5,000 dollars, minus four 500-dollar losers gives +3,000 dollars, which is +6R at 500 dollars per R. The best single trade was 750 dollars, which is 25 percent of the 3,000-dollar net, inside the 30 percent consistency ceiling explained in the consistency rule guide. The unfiltered version took three extra trades, all of them cracks entered before any structure shift, and two of those were Asian-session cracks that meant nothing; those entries account for most of the swing from profit to loss. Confirmation did not find better cracks; it removed the cracks that were never trades. That is the normal shape of a funded edge: fewer, cleaner trades, not a magic pattern.
When the strategy fails
The first failure mode is the correlated pair breaking late. GBPUSD holds its low for three candles, you enter the EURUSD long, and then GBPUSD breaks its low anyway, dragging EURUSD back down through your stop. This is why the stop sits beyond the sweep extreme and not at the confirming pair’s level: the confirming pair is evidence, not protection, and evidence can change its mind. When both pairs finally break together, the SMT is dead, and the correct action is the stop, not a second entry.
The second failure mode is a crack during a genuine trend change. Sometimes one pair leads and the other follows a day later, and what looks like a refusal is just a delay. The H4 bias rule is the defence here: a bullish SMT inside an H4 downtrend is not a reversal signal, it is a pause, and pauses resolve in the direction of the trend more often than they reverse it. The third failure mode is the news-mimic crack, covered by the session filter: a spike sweeps one pair while the other has not yet reacted, and by the time both have reacted, the “divergence” has vanished. The fourth failure mode is overtrading the crack. SMTs are rare — a genuine, simultaneous, confirmed crack appears a few times a week, not a few times a day. A trader who finds five SMTs a day is not finding SMTs; they are lowering their standards until noise qualifies, and noise is what the daily loss limit is for.
| Failure mode | What happens | Defence |
|---|---|---|
| Late break of the confirming pair | GBPUSD breaks its low after entry, dragging EURUSD down | Stop beyond the sweep extreme; accept the planned 500-dollar loss |
| Trend change, not a crack | One pair leads, the other follows a day later | H4 bias rule — no counter-trend SMT trades |
| News-mimic crack | Spike sweeps one pair before the other reacts | 30-minute news exclusion on both pairs |
| Overtrading weak cracks | Borderline holds counted as divergences | One trade per session; clear-margin rule on the hold |
| Chasing the shift candle | Entry at the top of the violent shift, stop too wide | Gap retest entry only — no entry, no trade |
How SMT fits prop firm rules
SMT fits funded trading because it disciplines the two rules that end most funded accounts: the daily loss limit and the consistency rule. The daily loss limit is protected by the sizing in this guide and by the one-trade-per-session rule. At 500 dollars per trade, a single failed crack costs 10 percent of the day, and the session rule means a failed crack cannot turn into three revenge trades that cost 30 percent. The consistency rule is protected by the exit split. Because winners are taken in two parts at 1R and 2R, results cluster around 250 to 750 dollars rather than one lucky spike, which is exactly the profile a consistency ceiling rewards. In the twelve-trade test, the best trade was 25 percent of net, comfortably inside a 30 percent ceiling, because no single trade was allowed to dominate the result.
Used this way, SMT is not a system on its own. It is a two-chart filter inside a system: bias from H4, crack from two pairs, confirmation from a structure shift, entry from a gap retest, size from the stop, and exits split between 1R and 2R. Remove any one of those pieces and the crack starts lying to you, because a divergence without confirmation is just two charts disagreeing, and charts disagree all day. Keep all six rules together, and SMT does the one job it was built for: it tells you, before you risk funded capital, whether the break you are watching is a real move or a sweep that the rest of the market refused to follow. Partial closes are for winners, taken at planned targets as in the worked trade above, and the trend pullback guide covers the single-chart cousin of this plan for days when no crack forms at all.
Daily routine for SMT trading
| Step | Action | Time |
|---|---|---|
| 1 | Mark prior H1 swing highs and lows on EURUSD and GBPUSD; note H4 bias | Before London open |
| 2 | Check the news calendar for EUR and GBP releases; block 30 minutes around each | Before London open |
| 3 | Watch for simultaneous extremes in the London window; ignore Asian-session cracks | 08:00 to 11:00 GMT |
| 4 | Wait for the M15 structure shift, then the fair value gap retest; no shift, no trade | Inside the window |
| 5 | Size from the stop at 500 dollars risk; split exits at 1R and 2R; log the trade | At entry |
| 6 | After one SMT trade, win or lose, close the SMT watch for the session | After the trade |
Frequently asked questions
What is SMT divergence in simple terms?
It is a disagreement between two correlated markets. One makes a new high or low and the other does not. The pair that refuses to confirm is showing where real buying or selling is defending, and the trade goes in that direction after confirmation.
Which pairs work best for SMT?
EURUSD traded against GBPUSD confirmation is the cleanest forex pair, because both share the US dollar driver and both are liquid in London and New York. Avoid weakly correlated pairs, where a “divergence” is usually just the two markets moving for different reasons.
Do I trade the crack as soon as it appears?
No. The crack alone is not a trade. Wait for a market structure shift on M15 and enter on the retest of the fair value gap the shift leaves behind. The twelve-trade test in this guide shows the unfiltered crack-only version losing money where the confirmed version profited.
How is SMT different from RSI divergence?
RSI divergence compares price against an oscillator on the same chart. SMT compares price against price on a second chart. No indicator is involved, and no overbought or oversold reading is required.
How often does a genuine SMT setup appear?
A few times a week, not a few times a day. If you are finding five SMTs a day, your standards have slipped and you are counting noise. The one-trade-per-session rule exists to stop that drift.
What kills an SMT trade after entry?
A close back through the sweep extreme, or the confirming pair breaking its own level late and dragging the traded pair with it. Both mean the crack has failed, and the correct funded action is the planned stop, not a second entry.
Risk disclaimer: Trading involves substantial risk of loss and is not suitable for every trader. This guide is educational only and is not financial advice. Funded-account rules vary by firm, so always check your own firm’s daily loss, trailing drawdown, and consistency rules before trading any strategy setup with real or funded capital.




