Strategy Library

ICT

Turtle soup

A failed-breakout reversal from Laurence Connors and Linda Bradford Raschke's book Street Smarts (1995), named as a joke about the Turtles' 20-day breakout system. When price makes a new 20-day low and the old low was at least four sessions earlier, a buy stop just above the old low fills you only if the breakdown fails. ICT uses the same name more loosely for any sweep of an old high or low that reverses.

IntermediateIndex futures, stocks, forex, goldAny; the original rule uses daily barsH1, D1
Use this strategyPractice it

Example 01

Turtle soup on a chart

XAUUSD · D1Failed lownew low, fail, turn

Old 20-day low: 2,385.40Buy stop: 2,386.40Stop: 2,370.40Target, range middle: 2,410.02Sample candles, not market data

The rules

Setup
  • Price makes a new 20-day low (mirror it for highs)
  • The previous 20-day low was made at least 4 sessions earlier
Entry
  • A buy stop a little above the old 20-day low, so you are filled only if price comes back above it
Stop
  • Just under the new low
Target
  • The sources leave targets open: set your own, such as the middle of the range or a fixed R
Manage
  • Cancel the order if price does not come back above the old low that day

Common mistakes

  • Fading before it fails. Selling a breakout that is still going is not turtle soup. Wait for the return inside.
  • Calling it an ICT invention. The rule is Connors and Raschke's; ICT borrowed the name.
  • Any sweep counts. The original rule has a 20-day level and a 4-session gap. Say which version you trade.

When it's off

  • No close back inside the range means the breakout has not failed: no trade
  • Skip it when the old low is fewer than 4 sessions old

Words to know

Turtle Soup
Connors and Raschke's failed-breakout reversal (Street Smarts, 1995): price makes a new 20-day low while the prior 20-day low is at least 4 sessions old, and a buy stop just above that old low fills you only if the breakdown fails (mirror for highs). ICT uses the name loosely for any sweep of an obvious high or low; that version is defined as a liquidity sweep.
Liquidity sweep
Price trades beyond an obvious high or low, fills the orders resting there, and closes back inside. Academy's test is a wick beyond the level with the same candle's body closing back inside; a body close beyond is a real break, and a sweep alone reverses nothing. ICT's loose use of "turtle soup" means this.
False breakout
Price breaks beyond a level but cannot stay there and comes back inside, trapping breakout traders. It is only known after the return, never at the breakout candle's close.
Failed breakout
A break of an old high or low that comes straight back inside the range.

When you make it yours, I'll ask what the rules above leave open. Start with one of these:

  • Do you trade the original rule (20-day level, 4 sessions old) or the looser ICT version?
  • Which timeframe do you use, and how many bars make your lookback?
  • Do you enter on a stop order back inside, or on a candle close back inside?
Ask Edge in DT Terminal

From DT Academy

What testing exists

The authors published the rule with their own examples. No independent test was found in the sources consulted.

Before you trust a backtest

  • Many professional traders risk under 1% of their account on one trade idea, and the common guideline is 1 to 2%.
  • The more versions of a rule you try on the same data, the more likely a good backtest is luck. Decide your rules before you test.
  • Include spread and commission, and practice on dates you have not looked at, before you trust a result.

Sources