The Turtle Soup is one of ICT's most precisely named concepts. The original "Turtle Trading" system — developed by Richard Dennis and William Eckhardt in the 1980s — was built on buying breakouts above 20-day highs and selling breakouts below 20-day lows. It made fortunes for a generation of trend-following traders. Then it became famous. Then retail traders started trading it. Then the breakouts started failing.
Linda Bradford Raschke documented the failure pattern in the 1990s and named her fading strategy "Turtle Soup" — cooking the Turtles in their own system. ICT adopted and refined the concept into the institutional framework: these breakout failures are not random. They are engineered. The prior high exists as a liquidity pool — a dense cluster of buy stops from short sellers and breakout buy orders from trend traders. Institutions push price above that level specifically to trigger those orders, use the flood of buying as their selling counterparty, and then deliver price lower once the stop cluster is exhausted.
Understanding the Turtle Soup means understanding that breakouts above prior highs are not signals to buy. They are signals that an institution just got filled. The Turtle Soup is the trade on the other side of that fill.
The Name, the Origin, and What ICT Changed
To understand ICT's version of Turtle Soup, it helps to understand what it was adapted from and what was different in the adaptation.
The original Turtle Trading system (Dennis/Eckhardt, 1983) bought 20-day highs and sold 20-day lows — the simplest possible trend-following rule. It worked because trend-following worked in the 1980s commodity markets. Raschke's Turtle Soup strategy (1990s) recognised that as the original Turtle system became widely known, the 20-day breakout became a predictable retail entry point — and therefore a reliable liquidity pool for institutions to sweep before reversing. Raschke's entry was: wait for a 4-day high or low to be breached, then fade the breach within 2 bars.
ICT's adaptation shifted the framework from a mechanical N-day rule to an institutional order flow context. In the ICT version:
The target level is not specifically a 20-day or 4-day extreme — it is any prior session high or low that has been respected for a minimum of 2 sessions, creating a meaningful stop cluster. The larger the cluster (more sessions, more touches of the level), the higher the probability of the Turtle Soup reversal. The entry is refined from Raschke's "2-bar reversal" to an MSS + FVG sequence on the lower timeframe — more precise, tighter stops, better R:R. And crucially, the Turtle Soup is placed within the AMD framework — it is only valid as the manipulation phase of AMD, not as a standalone pattern.
What the ICT Turtle Soup Setup Is
The ICT Turtle Soup is a reversal setup that occurs when price sweeps above a prior multi-session high (taking buy-side liquidity) or below a prior multi-session low (taking sell-side liquidity) before reversing sharply in the opposite direction.
It has four specific components:
1 — A prior high or low that has held for a minimum of 2 sessions. A random 30-minute swing high from the prior session is not a Turtle Soup target. The level must be a meaningful structural high or low — a PDH (prior day high), an equal high from two or more sessions, a prior week high, or a multi-session consolidation ceiling. The minimum standard is that the level was set at least 2 complete trading sessions ago and has not been breached since. More sessions = larger stop cluster = stronger Turtle Soup potential.
2 — A sweep beyond the level (the wick). Price extends beyond the level with a wick — taking all the stops and breakout orders stacked above (for a BSL sweep) or below (for an SSL sweep). The sweep typically extends 10–30 points on NQ, 5–15 points on ES, or 10–30 pips on forex pairs. The key characteristic: the wick extends beyond the level but the candle body does not close beyond it. A candle that closes beyond the level is not a Turtle Soup — it may be a genuine breakout, or the manipulation is still in progress.
3 — A body close back inside the range (the 2-bar rule). The sweep candle's body closes back inside the prior range. This is the first confirmation that the sweep is complete and the level held. Without this close-back, you do not have a Turtle Soup — you have a potential breakout in progress.
4 — Reversal confirmation (MSS on 5M chart). After the body closes back inside the range, a market structure shift on the 5-minute chart confirms the reversal. A bearish MSS for a BSL sweep, bullish MSS for an SSL sweep. The FVG formed during the MSS candle is the precise entry trigger.
The Minimum Lookback Rule — Why It Matters
The lookback rule is what separates a Turtle Soup from a random wick beyond a level. Raschke's original formulation used 4 bars (a 4-day high/low on the daily chart). ICT's modern application uses a minimum of 2 prior sessions, with higher probability as the session count increases.
The reason is mechanical. A level that was set yesterday and has not been tested creates a small stop cluster — perhaps the stops from traders who entered near that level in the prior session. A level that has been respected for 5 sessions creates a massive cluster: stops from traders who entered at the level in all 5 sessions, plus breakout orders from technical traders watching the "resistance," plus any algorithmic orders triggered by the multi-session pattern. The larger the cluster, the more orders institutions need to absorb, the more decisively they reverse once filled, and the cleaner the Turtle Soup reversal.
In practice, the most powerful Turtle Soups occur at:
Equal highs or equal lows with 3+ touches. When a prior high has been approached and rejected 3 or more times within the accumulation range, the stop cluster above it is enormous. Every rejected approach adds more trapped shorts and more breakout buy orders waiting for the "confirmed" breakout. The 4th or 5th approach often produces the Turtle Soup sweep — a decisive wick above followed by the sharpest possible reversal.
The PDH and PDL (prior day high and low). On a daily AMD cycle, the PDH is the natural Turtle Soup target for a bearish day — the Judas Swing sweeps the PDH (which has held since the prior day, satisfying the minimum lookback) before bearish distribution begins.
Prior week high and low. On weekly AMD cycles, the prior week's high and low are the primary Turtle Soup targets. A Monday Judas that sweeps the prior week's high before Tuesday's bearish weekly distribution is a textbook weekly Turtle Soup.
Valid vs Invalid Turtle Soup — The Four Filters
The single most important filter is the bias alignment. On a bearish day, a sweep above a prior high is valid — it is the manipulation phase sweeping buy-side liquidity before bearish distribution. But on a bullish day, a sweep above a prior high is not a Turtle Soup — it is price delivering to the buy-side target in the distribution phase. Entering short into a bullish day's distribution is fading the institutional move, not riding it. The daily bias determines whether the sweep is manipulation (Turtle Soup) or distribution (continuation). Never apply Turtle Soup in the direction of the daily bias.
The Entry Mechanics — 2-Bar Rule vs FVG Precision
There are two entry approaches for the Turtle Soup, representing different precision levels. Both are valid; the FVG version gives tighter risk and better R:R.
The 2-bar reversal rule (Raschke's original, simplified): When the sweep candle's body closes back inside the prior range, that is bar 1 of the reversal signal. On the open of the next candle (bar 2), enter in the reversal direction — short for a bearish Turtle Soup, long for a bullish one. Stop goes above the wick high of the sweep candle. This is the quick, mechanical entry. The limitation: entering at bar 2 typically means entering during a fast-moving candle after a large displacement, which often gives a worse price than the FVG version.
The FVG/MSS version (ICT 2022 Model precision): Wait for the sweep candle to close back inside the range (same first signal). Then switch to the 5-minute chart and wait for the market structure shift — a prior 5M swing low broken for a bearish Turtle Soup. The displacement candle that creates the MSS will also create a bearish FVG. On the brief retrace into that FVG (typically 2–5 candles), place a limit order at the 50% CE. Stop above the Turtle Soup wick high. This version gives a tighter stop (entering deeper in the reversal, not at the top of it) and a better average entry price.
The recommendation: use the 2-bar rule as a trigger awareness signal. When you see the sweep candle close back inside the range, switch to 5-minute and begin watching for the MSS. Use the FVG entry once the MSS fires. If no clean FVG forms within 3–5 candles after the MSS (rare), the 2-bar market order entry is acceptable as a fallback.
Turtle Soup, Judas Swing, and AMD — The Same Mechanism
This is the insight that no competing article on Turtle Soup explains: in the ICT framework, Turtle Soup, Judas Swing, and the AMD manipulation phase are all describing the same institutional mechanism from different perspectives.
| Name | What it describes | When both labels apply | The key distinction |
|---|---|---|---|
| Turtle Soup | A reversal trade fading a sweep of a prior multi-session high/low | When the Judas Swing sweeps a level that held for 2+ sessions | Describes the setup structure — what level was swept |
| Judas Swing | The manipulation phase of AMD — a deceptive move before true distribution | When the Judas sweeps a prior multi-session high or low | Describes the AMD phase — where in the cycle this sits |
| AMD Manipulation | The second phase of the Accumulation–Manipulation–Distribution cycle | Every time any Judas or Turtle Soup occurs | Describes the macro cycle — the broadest label |
| Silver Bullet | A specific 10–11 AM ET entry window using the AMD/Judas sequence | When a Turtle Soup fires during the Silver Bullet window on NQ/ES | Describes the timing window — the most specific label |
On a bearish NQ day, the pre-market range forms the accumulation. At 9:30–10:00 AM, the Judas Swing sweeps above the pre-market high (which has held since the pre-market session — satisfying the Turtle Soup lookback minimum). The sweep wick forms, the body closes back inside, the MSS fires. This is simultaneously: a Turtle Soup (the structural label), a Judas Swing (the AMD phase label), the AMD manipulation phase (the cycle label), and if it resolves during the 10–11 AM window, a Silver Bullet (the timing label).
All four labels are correct. They describe the same candle from four different angles of the ICT framework. Understanding this convergence is what separates traders who see isolated concepts from those who read the market as a single integrated system.
Turtle Soup on NQ and ES — The Modern Application
The Turtle Soup is most commonly applied to NQ and ES in the current ICT community, for the same reason these markets dominate ICT trading generally: clean structure, consistent kill zone behaviour, and the NQ/ES SMT pair for confirmation.
The most reliable NQ Turtle Soup scenario: the pre-market session (7:00–9:30 AM ET) forms a range. At 9:30 AM open, NQ spikes above the pre-market high with a wick — taking all the buy stops from traders who were short from below the pre-market high, plus breakout buyers who entered on the "open above the high." ES simultaneously makes a smaller new high or fails to confirm the NQ extension — SMT divergence confirmed. NQ body closes back below the pre-market high. MSS fires. Bearish FVG forms. Entry at 50% CE.
The timing is critical: on NQ, the 9:30 AM Turtle Soup sweep typically completes by 9:40–9:45 AM. The MSS and FVG entry are available from 9:40–9:55 AM. The Silver Bullet window opens at 10:00 AM. If the entry opportunity falls between 9:40 and 10:00 AM, wait for the window. If the FVG is still intact at 10:00 AM when the window opens, take the entry then. If price has already moved beyond the FVG before 10:00 AM, the entry is gone — do not chase.
The weekly Turtle Soup on NQ follows the same logic scaled up: Friday closes establish the prior week's high and low. Monday and Tuesday accumulate around the weekly EQ. On Wednesday, the weekly Judas Swing sweeps the prior week's high (for a bearish week) — this is a weekly Turtle Soup. The sweep of a 3-day-old level (Thursday–Friday prior week, Monday–Tuesday current week) satisfies the lookback rule emphatically. Wednesday's Turtle Soup short is one of the highest-probability weekly setups in the ICT framework.
Full Walkthrough — NQ Turtle Soup Short
Pre-session analysis (6:00 AM ET): NQ daily bias bearish — weekly premium at 21,380 vs weekly EQ 21,040. Draw on liquidity: PDL at 20,890. Pre-market range forming: pre-market high 21,418, pre-market low 21,344. The pre-market high at 21,418 is 10 points above the PDH at 21,408 from Thursday — creating a combined BSL cluster (stops from Thursday's failed high attempt + stops from pre-market short sellers + breakout buy orders above 21,418). This is a Turtle Soup setup in waiting. The level has been held since Thursday (2 sessions ago). Minimum lookback satisfied.
9:30 AM ET — Turtle Soup fires: NQ opens at 21,390 and immediately spikes to 21,452 — 34 points above pre-market high. ES simultaneously reaches 5,642 from a prior high of 5,638 — only 4 points above versus NQ's 34 points. SMT divergence confirmed. NQ body closes at 21,404 — back below the pre-market high. Turtle Soup wick confirmed.
9:36 AM — MSS: 1M swing low at 21,388 (formed at 9:33 AM) broken. Bearish MSS confirmed. Bearish FVG forms between 21,404 and 21,436 during the MSS candle.
Entry (10:03 AM ET — kill zone open): At 10:03 AM, NQ retraces to 21,418, entering the FVG zone. Limit short at 50% CE: (21,404 + 21,436) / 2 = 21,420. Filled 10:04 AM.
Stop: Above Turtle Soup wick at 21,452 — buffer to 21,458. Distance: 38 points.
T1 (IRL): Equal lows at 21,190 — 230 points, 6.1R. Hit 11:32 AM.
T2 (ERL): PDL at 20,890 — 530 points, 13.9R. Hit the following session.
Common Turtle Soup Mistakes
Trading Turtle Soup without a daily bias. A sweep above a prior high when the daily bias is bullish is not a Turtle Soup — it is price delivering to the buy-side target. The most common Turtle Soup mistake is shorting every sweep of a prior high regardless of the day's directional bias. The bias determines whether the sweep is manipulation (Turtle Soup) or distribution (continuation). Never fade the distribution direction.
Using levels that are too fresh. A high set three hours ago does not have a meaningful stop cluster above it. A high set in the prior session, which other traders have been watching for 8+ hours and adding to their bias, has a genuine cluster. The 2-session minimum is not arbitrary — it is the time required for enough traders to act on the level for the cluster to be worth sweeping. Single-session wicks above new highs are different from multi-session consolidation ceilings.
Entering on the sweep candle itself. The sweep candle is the Judas candle — by definition, you should not be entering in the sweep direction, and you should not be entering in the reversal direction until the body closes back inside the range. Entering short as soon as a wick appears above the prior high means entering before the manipulation is complete. Wait for the body close.
Missing the AMD context. A Turtle Soup without AMD context is just a random counter-trend entry. Placing the Turtle Soup within the AMD sequence — confirming that the sweep is the manipulation phase, not a random wick — is what elevates it from a 50/50 trade to a high-probability institutional setup. Always ask: does this sweep fit the expected AMD sequence for today? Is this the manipulation phase I was looking for, at the time I expected it?
Frequently Asked Questions
What is ICT Turtle Soup?
What is the minimum lookback for a valid ICT Turtle Soup?
How is ICT Turtle Soup different from a regular liquidity sweep?
What is the difference between ICT Turtle Soup and the Judas Swing?
What is the entry rule for the ICT Turtle Soup?
1 — The level must have held for 2+ sessions (no fresh levels). 2 — The wick sweeps the level but the body closes back inside (wait for the body close). 3 — The sweep must be against the daily bias — fading the manipulation, not the distribution. 4 — Enter on the MSS + FVG retrace, not on the sweep candle. Everything else is refinement.