What Is the Turtle Soup Setup? (ICT and Raschke)
Turtle soup is a setup that trades against a failed break of a 20-day high or low: price pushes past an old extreme, then closes back inside it. It was named as a fade of the Turtle traders' 20-day breakout, and ICT uses the name for the same raid on an old level.
Turtle soup is a trade against a failed breakout of a 20-day high or low: price runs past an old extreme, catches the orders waiting there, and closes back inside the range. The name comes from the book Street Smarts, written by Linda Bradford Raschke with Laurence A. Connors, and it was built as a fade of the Turtle traders’ 20-day breakout.
ICT borrowed the name for the same event. In that material it describes price raiding an old high or low before reversing, which is close to what the original setup was built to catch.
How it works
Start with the Turtle rule it fades. In the Turtle trading system, System 1 bought when price exceeded the high of the preceding 20 days and sold short below the 20-day low. A rule that simple is followed by many traders, so their orders gather just beyond those levels.
Turtle soup waits for the break to fail. The long version needs three things:
- Today’s low goes below the lowest low of the prior 20 sessions.
- That prior 20-day low was set at least four sessions earlier. A low made yesterday is part of the current leg down, not an old level anyone has been watching.
- Price gets back above the old low.
The short version mirrors it: a new 20-day high over an old high, then a return below it.
The book’s entry, as quoted in a 2018 StockCharts article by Julius de Kempenaer, is a buy stop placed slightly above the old 20-day low once the market has fallen below it, and, if that fills, a good-till-canceled stop just below that day’s low. This page measures a daily-close version of step 3, because the order of prices inside a daily bar cannot be known from the bar alone.
Why the four-session rule matters
Most breaks of a 20-day low are not setups. SPY broke its prior 20-day low on 743 days between 1993 and 2026. Only 227 of those breaks came against a low that was at least four sessions old.
The rule separates a raid from a trend. In a steady decline, every day makes a new 20-day low and the previous one was set yesterday. Fading each of those means standing in front of the move. The age rule keeps only the breaks of a level that had time to become obvious, which is where the resting orders are assumed to be.
It is the same idea as inducement run backward: an old, visible level gets taken, and the question is whether price can stay beyond it.
A worked example
26 June 2026, SPY daily. The lowest low of the prior 20 sessions was $722.59, set on 9 June, 12 sessions earlier. That clears the four-session rule.
- The day opened at $728.95, traded as low as $716.58 and closed at $728.99.
- The low went $6.01 under the old low ($722.59 - $716.58).
- The close finished $6.40 back above it ($728.99 - $722.59).
That is a long turtle soup under this page’s rule. A stop under the day’s low sits at $716.58, which is $12.41 below the close, and that distance is what the position size has to be built around.
What followed. The lowest low of the next 10 sessions was $732.09, so the day’s low was never retested. Ten sessions later, on 13 July, SPY closed at $749.17, up 2.77% from $728.99.
That is one favorable case. The next section counts every case, and the “When it fails” section shows three in a row that went the other way.
The original data
The rule. SPY daily bars, 29 January 1993 to 25 September 2026, 8,472 sessions. A long setup is a day whose low is below the prior 20-session low, where that low was set four or more sessions earlier, and whose close is back above it. A day with the same break that closes at or below the old low is counted as “held below”. Short setups mirror this with highs. Every case is in the turtle soup event file.
How often it forms. Of the 227 qualifying breaks of an old 20-day low, 89 (39.2%) closed back above it, about 2.6 a year. Of 376 qualifying breaks of an old 20-day high, 161 (42.8%) closed back below, about 4.8 a year.
Long setups against any day. SPY closed higher 5 sessions later after 47 of the 89, 52.8%, against 57.9% of all days (4,888 of 8,447). Ten sessions later it was higher after 51 of 89, 57.3%, against 60.6% (5,118 of 8,442). Both are within chance on a two-sided binomial test (p = 0.34 and p = 0.52). The breaks that stayed below, 138 of them, were higher 5 sessions later 63.0% of the time and 10 sessions later 64.5%.
Short setups against any day. SPY was lower 10 sessions after 60 of the 161, 37.3%, against 39.2% of all days (3,307 of 8,442), again within chance (p = 0.69).
The stop under the day’s low. Within 10 sessions, the low of a long setup day was traded through in 61 of 89 cases, 68.5%. For comparison, any day’s low was taken out within 10 sessions 76.7% of the time. The setup day’s low held a little more often, but a stop placed there was still hit in about two cases out of three.
Coverage is scarce. In the 24,971-video study behind this site, 3 titles name turtle soup, from 3 channels, at a median of 23,976 views, and the most-viewed of them has 31,547.
When it fails
The soup gets souped
March 2026 shows the failure in sequence. On 3 March SPY traded to $669.66, under the old low of $675.78 from 17 February, and closed back above it at $680.33: a long setup. That old low was itself the low of a 17 February setup, now taken out. On 9 March it traded to $662.39, taking out the 3 March low and any stop under it. That day closed back at $678.27, so it was a new setup of its own, against the 3 March low. On 13 March the low went to $661.36 and the close stayed below at $662.29. Ten sessions after 9 March, SPY closed at $655.38, down 3.37%.
Each setup’s low became the next obvious level, which is the sweep of a sweep pattern with a 20-day ruler on it.
It is a trend, not a raid
Breaks that held below the old low were followed by more up days than turtle soups were. On an index that rises over time, the next two weeks after any sharp dip have often been positive, so the setup has to beat a high bar.
The daily close hides the path
A daily bar cannot show whether the reclaim came at 10 a.m. or in the last minute. A buy stop just above the old low can fill, get stopped under the new low and leave the daily bar looking like a clean setup.
The lookback is arbitrary
Twenty sessions was chosen to fade one particular rule. Change it to 15 or 30 and a different set of days qualifies, which leaves room to find a turtle soup behind almost any reversal after the fact.
Related
Turtle trading is the breakout system this setup was built to fade, with the published rules.
A liquidity sweep is the general event, and turtle soup is one version of it with a fixed lookback and an age rule.
A false breakout describes the same failure from the breakout trader’s side, including what it costs to keep taking range breaks.
The 20-day high and low are the outer lines of Donchian channels, which is the quickest way to see the level on a chart.
If you trade this, put the stop under the day’s low and size the position for that distance before you look at the target. The setup exists because stops sit just beyond obvious levels, so assume yours can be taken the same way.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.