Turtle Soup Trading: Raschke Rules, ICT Version and a Test

A market drops below its lowest low of the last 20 days, breakout traders sell, and shortly afterward price is back above that level. Turtle Soup trades exactly this failed breakout, in the opposite direction. The setup comes from Linda Bradford Raschke, who published it with Laurence Connors in the 1995 book Street Smarts. Today the name turns up mostly in ICT teaching, the followers of the Inner Circle Trader Michael Huddleston, and there it follows different rules.

This guide puts the original rules, Turtle Soup Plus One and the ICT version side by side. It also follows one SPY trade that worked and one that failed, and it reports a 25-year test on the S&P 500 and the Nasdaq 100. In that test, every version we tried had a negative average return per trade. This article follows our editorial policy.

Turtle Soup in 30 seconds

• Meaning: Turtle Soup trades a failed breakout of a 20-day high or low in the opposite direction.
• Origin: Raschke and Connors described it in Street Smarts (1995) as a counter to the breakout rules of the Turtle traders.
• Original rules: a new 20-day low, an old low at least four days back, a buy stop just above the old low, and a stop below the new low.
• ICT version: a sweep of an obvious high or low, with an entry on a short timeframe, often through a broken fair value gap.
• Our test: on SPY and QQQ from 2001 to 2026, every tested version had a negative average return per trade, and the median trade was a full loss.

What is Turtle Soup?

Turtle Soup is a trading strategy that takes a failed breakout of a 20-day high or low in the opposite direction, once price comes back through the old extreme.

The idea is a counter to classic breakout trading. Breakout traders buy above an obvious high and sell below an obvious low. If the breakout fails, they are wrong at once and have to get out. Turtle Soup bets that this reversal gains momentum. It is the same basic pattern as a false breakout, which our guide to breakout trading describes, but with fixed rules.

Turtle Soup is a countertrend setup. You trade against the last move, not with it, and you expect price to return into its previous range. It is meant for a short holding period of a few days, which makes it a form of swing trading rather than a long-term position.

Where the name comes from

The name is a jab at the famous Turtle traders. In the 1980s, Richard Dennis and William Eckhardt taught a group of beginners a trend-following system that, among other things, bought a new 20-day high. How that system worked is in our guide to Turtle trading.

A trend-following system lives with many false signals. It loses a little on most breakouts and earns its money on the few big trends. Raschke and Connors turned that logic around and traded exactly the failed 20-day breakouts that cost trend followers money. The turtles became the soup. More about Raschke’s work and her other setups is in our portrait of Linda Raschke.

The original rules from Raschke and Connors

These are the buy rules from chapter 4 of Street Smarts, in a simplified summary. For a short trade, everything is mirrored at a 20-day high. The details were checked against the book’s chapters by our reviewer; we have not read the full book ourselves, so treat this as a careful summary, not a quotation. The book shows the setup mostly on daily charts, but it uses the same logic on other timeframes, including 10-minute charts with 20-bar extremes. The entry offsets below are the daily rules. For intraday charts, the book uses one tick beyond the previous 20-bar extreme instead, for both Turtle Soup and Plus One.

1. New low: today, price trades below the lowest low of the last 20 trading days (or the last 20 bars on another timeframe).
2. Old low with distance: that previous 20-day low must be at least four trading days old.
3. Entry (daily): a buy stop 5 to 10 ticks above the previous 20-day low, good for that day only. It fills if price comes back up to it.
4. Stop: one tick below today’s low.
5. Exit and re-entry: the book trails the stop as the trade moves and gives no fixed target. If the trade is stopped out on the first or second day, the book allows a re-entry.

The age filter is meant to exclude fresh levels. A low that formed yesterday has had little time to be noticed. Which orders really sit below a low, a chart cannot show. The idea is only that an older, more visible low attracted more breakout trades. Some summaries online give three days instead of four; in the book, three sessions belong to the Plus One variant.

Turtle Soup Plus One: the entry one day later

Turtle Soup Plus One gives the breakout one more day. Instead of entering on the day of the new low, it waits to see whether the break is confirmed by the close, and enters the next day if it then fails. In its long form:

1. Day one: a new 20-day low, the previous low at least three sessions old, and a close at or below that previous low.
2. Day two: a buy stop at the previous 20-day low on daily charts (one tick above the old 20-bar low intraday), which fills if price comes back above it. If it does not fill on day two, the order is cancelled.
3. Stop: one tick below the lower of the two lows, day one or day two.
4. Exit: take partial profits within two to six bars and trail a stop on the rest.

These are the book’s rules from chapter 5; our test replaces two of them. Our stop sits one cent below the low of day one, because the low of day two is not known yet at the moment of entry on daily bars. And instead of partial profits and a trailing stop, the test uses one fixed exit, described below. Both are simplifications that make the rule countable, not part of the original.

The advantage is that you can plan it. The order is set before the session opens, so you do not have to watch for a reversal during the day. The drawback: the stop is often farther away, because the new low has more room to the entry. Which version works better is a question for a test, not for a rule of thumb.

Turtle Soup in ICT: a sweep instead of the 20-day rule

In ICT teaching, Turtle Soup is no longer a fixed rulebook but a name for a pattern. It means a liquidity sweep: price runs briefly above an obvious high or below an obvious low and comes straight back. The entry then comes on a short timeframe, often through an inverse fair value gap, a broken three-candle gap, or through a market structure shift. The wider vocabulary is in our guides to ICT trading and Smart Money Concepts.

Feature	Original Turtle Soup	Turtle Soup Plus One	ICT usage
Level	20-bar low or high, at least 4 bars old	20-bar low or high, at least 3 bars old	Any obvious high or low
Timeframe	Mostly daily; also intraday, e.g. 10-minute bars	Mostly daily; also intraday	Often minute charts
Confirmation	Price trades back through the old level	Close beyond the old level on day one	Sweep, then a broken gap or structure
Entry, daily	Stop order 5 to 10 ticks beyond the old level, same day	Stop order at the old level, next day	Often at a gap or its midpoint
Entry, intraday	Stop order 1 tick beyond the old 20-bar extreme	Stop order 1 tick beyond the old 20-bar extreme	Same as above
Rules	Fixed	Fixed	Discretion

Not every sweep is a Turtle Soup, and the original was not an ICT idea. A sweep of a two-day-old high does not meet Raschke’s age rule, and a sweep of just any obvious level skips the 20-bar rule. Trading on a short timeframe alone does not rule out the original, because the book applies it to intraday charts too. The original enters with a stop order just beyond the old level, 5 to 10 ticks on daily charts and one tick intraday, not on a gap. Our IFVG guide shows how one scalper uses the ICT version in Nasdaq futures. Apart from the basic idea, it has little in common with the 20-day rule.

Two SPY examples: one worked, one failed

Both examples follow the test version of the rule described below, so they were not picked by eye. We took the last 2R winner and the last full loser of that rule on SPY daily bars, in a sample ending October 16, 2026. The test version waits for a close back above the old low and buys the next open, which is later than the original stop order. Two cases prove nothing about the next one.

The setup worked: September 2026

This trade did what the idea promises. The numbers below match the markers in the chart, and all prices are in US dollars.

1. Old low: the lowest low of the prior 20 days was 760.52, set on August 4.
2. New low, close back above: on September 1, SPY traded down to 759.48 and closed at 761.78, above the old low.
3. Entry: the buy filled at the September 2 open of 762.45, with a stop at 759.47 and a target at 768.41, twice the risk.
4. Exit: on September 3, the high of 774.03 passed the target.

The risk was small because the signal bar barely broke the old low. The stop sat only 2.98 below the entry, about 0.4% of the price. That made the 2R target close enough to reach in one day. Under the original rule, the buy stop would have filled on September 1 itself, a few ticks above 760.52, at a better price. Daily bars cannot show the order of intraday moves, which is why our test enters at the next open.

The setup failed: March 2026

The last full loser looked just as valid on the day of the signal. On March 9, SPY traded down to 662.39, below the old 20-day low of 669.66 from March 3, and closed far above it at 678.27. The buy filled at the March 10 open of 677.72.

The wide signal bar made the stop expensive. The stop sat one cent below the low of the signal day, at 662.38, about 2.3% below the entry. The target at 708.40 never came close. On March 13, the low of 661.36 hit the stop, a full loss of one R. A strong reversal day can look like a clean Turtle Soup and still leave you with a stop that is too far away to be comfortable, or too close to survive the next leg down.

Turtle Soup test: 25 years of SPY and QQQ

Whether Turtle Soup holds up as a fixed rule can be counted. We used daily bars of SPY (S&P 500) and QQQ (Nasdaq 100) from January 2, 2001 to October 16, 2026, from Yahoo Finance, adjusted for splits but not for dividends. We tested long and short setups and measured each trade in percent and in R, where 1R is the distance from entry to stop.

• Turtle Soup, test version: a new 20-day low, the old low at least four days old, and a close back above the old low. Buy at the next open; stop one cent below the low of the signal day. If the next day opens below that stop, the trade counts as closed at the open, with a result of zero.
• Turtle Soup Plus One: a new 20-day low with a close at or below the old low. On the next day, a buy stop exactly at the old low, valid for that day only, as in the book. Stop one cent below the low of day one (our adaptation; the book uses the lower of both days’ lows).
• Exit for both: a target at twice the risk (2R), the stop, or the close of the 20th trading day after entry.

The test deliberately differs from the original at the entry. The original rule enters with a stop order during the day. Daily bars do not show whether price fell first and rose later, so a fill could not be checked honestly. We discarded four earlier versions of this count for that reason: one estimated fills from the close, one bought at a close that was not yet tradable, and two handled Plus One stops or entry-day cases wrongly. The current version uses only prices that are known at the moment of entry. We re-ran it on November 7, 2026 and got identical results.

Version	SPY trades	SPY average	QQQ trades	QQQ average
Turtle Soup (old low 4+ days)	198	−0.13% / −0.12R	192	−0.37% / −0.19R
Plus One (old low 3+ sessions, book)	181	−0.20% to −0.04%	168	−0.50% to −0.33%
Plus One (old low 4+ sessions)	137	−0.28% to −0.09%	127	−0.45% to −0.30%

Every version had a negative average return in percent. For Turtle Soup the median trade on both ETFs was a full loss of −1R, and more than half of all trades ended at the stop. On SPY, 82 of 198 trades were stopped on the entry day itself, 33 of them right at the open. On QQQ it was 82 of 192, with 30 at the open.

For Plus One we report a range instead of one number. The buy stop often fills during the day, and on many of those days price also touched the stop. In the four-session version, that happened on 64 of the 92 SPY trades filled during the day, and on 61 of 87 in QQQ. A daily bar does not show which came first. The unfavorable reading counts these as stops. The favorable reading first counts a target touch, which must have come after the fill, then a close beyond the stop. Even the favorable reading stayed negative in percent; in R it reached +0.17R on SPY in the four-session version, with a 95% range from −0.11R to +0.45R.

Other settings did not change the picture. With three instead of four days of distance, and with a fixed exit after five days instead of the 2R target, the average return was negative in every combination. As a rough reference, without comparable rules, buying on any day and selling five days later averaged +0.17% on SPY and +0.24% on QQQ over the same period. The long Turtle Soup trades with a five-day exit stayed below that.

How we counted the details. One long and one short position per market can run at the same time, but not two in the same direction. From the entry day on, the stop applies; if stop and target are touched on the same day, the stop counts. If price opens beyond the stop or target, the open counts. Fees and slippage are not deducted. The 95% ranges for mean R use a normal approximation and assume independent trades. For Turtle Soup they mostly include zero; in the unfavorable Plus One readings, most lie entirely below zero.

What the test does not answer. A discretionary trader picks setups and manages exits, and that cannot be written as one fixed rule. The test says nothing about the ICT version on minute charts. It shows only that the tested versions had a negative average return on these two ETFs. If you want to check your own version, our guide to backtesting shows how to start without fooling yourself.

Turtle Soup, fakeout and liquidity sweep: the difference

The three terms describe the same event from three angles. A fakeout is the general name for a breakout that fails at once. A liquidity sweep is the SMC reading of the same sequence: the break is said to have triggered stop orders that served larger traders as the other side. Turtle Soup, in contrast, is a trading rule that takes this failed breakout under fixed conditions.

In practice the order matters. First you see a break, then you check whether price comes back, and only then do you decide on an entry. Whether stops were really triggered, the chart does not show. Below a low, triggered sell stops add selling; above a high, triggered buy stops add buying. You only see that price could not hold the level, and our liquidity sweep guide tests that pattern on its own.

Common Turtle Soup mistakes

Most mistakes come from counting every break of a low as a setup. These four cost the most money, and each has a simple fix:

• A level that is too young: a low from yesterday had little time to be noticed. The original asks for several days of distance.
• No return: without a return above the old low, it is not a failed breakout but a successful one.
• A stop that is too tight: a stop just below the day’s low is often hit the next day. In our test, more than half of all trades ended at the stop.
• Against a strong trend: shorts at new highs in a rising market risk that the trend confirms the breakout instead of rejecting it. We did not test a trend filter here.

Pros and cons of Turtle Soup

Whether Turtle Soup suits you depends on how you handle many small losses. These strengths and weaknesses apply to any fixed version of the rule.

Pros:

• A clear level: the old 20-day extreme is an unambiguous reference. Entry and stop follow from the chosen version.
• Plannable risk: the planned risk follows from the stop distance and position size, as our guide to position sizing shows. Opening gaps can make a loss larger.
• Testable: a fully defined version can be counted on any market.

Cons:

• No demonstrated edge: in our SPY and QQQ test, every version had a negative average return per trade.
• Many stops: the typical trade ended with a full loss.
• Details differ between sources: even the minimum age of the old low is given differently in summaries.

Conclusion: a good idea, negative as a fixed rule

Turtle Soup describes a real market situation: breakouts fail, and then breakout traders are on the wrong side. That is a useful way to look at a chart, especially at obvious 20-day extremes. As a fixed rule on daily bars, though, the pattern had a negative average return in our 25-year test on the S&P 500 and the Nasdaq 100.

If you want to work with Turtle Soup, use it as a building block, not as a system. Combine it with the higher-timeframe trend and with a clear idea of where price should go after the failed breakout. Then test your exact version before you risk money.

This guide is education, not investment advice. Trading futures, options or leveraged products can lose more money than you expect, and past examples do not predict future results. Decide your risk per trade before you enter.

Frequently asked questions about Turtle Soup

What is Turtle Soup in trading?

Turtle Soup is a strategy that trades a failed breakout of a 20-day high or low in the opposite direction. It comes from Linda Bradford Raschke and Laurence Connors and was published in 1995 in Street Smarts. In ICT teaching, the name now stands for a liquidity sweep followed by a reversal.

Why is it called Turtle Soup?

The name refers to the Turtle traders, who bought new 20-day highs under Richard Dennis’s rules. Turtle Soup trades exactly those failed breakouts the other way, so the turtles become the soup.

What is Turtle Soup Plus One?

Turtle Soup Plus One is the version with an entry one day later. On day one, price makes a new 20-day low and closes at or below the old low, which must be at least three sessions old. On day two, a buy stop at the old low triggers the entry if price comes back above it. The stop goes one tick below the lower of the two lows.

What is the difference between Turtle Soup and ICT Turtle Soup?

The Raschke and Connors original is a fixed rule set built on 20-bar extremes, shown mostly on daily charts and also on intraday charts. The ICT version uses any obvious high or low, is usually traded on minute charts and enters through a broken fair value gap or a structure break. Not every sweep meets the original rules, but a short timeframe alone does not rule them out.

Does Turtle Soup work?

In our test on SPY and QQQ from 2001 to 2026, every tested version had a negative average return per trade. We tested Turtle Soup and Turtle Soup Plus One, with a 2R target or a five-day exit. Discretionary trading with your own judgment cannot be captured this way, and the test says nothing about intraday charts.

Is Turtle Soup the same as a fakeout?

Turtle Soup trades a fakeout, but it is not the same thing. A fakeout only describes the failed breakout. Turtle Soup also defines which level counts, when you enter and where your stop goes.