Turtle soup trading: fading the breakout that fails

What turtle soup is, and where the name comes from

In the 1980s a group of trainee traders known as the Turtles became famous for a simple breakout system: buy a new 20-day high, sell a new 20-day low. In the mid-1990s Linda Raschke and Larry Connors published a strategy that did the opposite, fading those 20-day breakouts when they failed, and called it Turtle Soup, because it fed on the Turtle traders' losing entries.

The logic is about stop orders. Obvious highs and lows collect two kinds of orders just beyond them: stops from traders positioned against the break, and entry orders from breakout traders. When price pushes through, those orders fill. If nobody else wants to keep pushing, the move runs out of fuel and reverses, trapping the breakout traders, whose exits then add to the reversal.

The ICT and smart money concepts communities adopted the name for the same event, which they also call a liquidity sweep or stop run. Their explanation is that large players deliberately push price through the level to fill their orders against the crowd. That may or may not be what happens, and a retail chart cannot tell you; what you can see is that some breakouts fail quickly, and that a failed breakout gives a precise entry and stop. That is the part worth trading.

The original rules, on the daily chart

The long version, as usually taught from the original:

1. Today, price trades to a new 20-day low.
2. The previous 20-day low was made at least four trading days earlier. This matters: it means the old low had time to become an obvious level with orders around it.
3. Place a buy stop a little above that previous 20-day low. It only fills if price reverses back up through it, the same day or the next.
4. Once filled, place the protective stop just below today's new low.

The short version is the mirror image: a new 20-day high, a previous 20-day high at least four days old, a sell stop just below that old high, and a protective stop above the new high.

In the chart above, EUR/USD set a 20-day low at 1.0760, rallied to 1.0812 and drifted back. Eighteen days later it broke the old low, dipped to 1.0742, then reversed. The buy stop at 1.0760 filled as price climbed back through the old low, the protective stop sat at 1.0738 below the new low, and the target was 1.0810, just under the swing high, a little over twice the risk.

Entry, stop, target and the intraday version

Entry. Use a stop order at the old level, not a limit order at the new low. That way price has to prove the break failed before you are in. Trying to catch the low with a limit order turns turtle soup into simple bottom-picking, and on the days the breakout is real you are filled at once and stopped soon after.

Stop. Just beyond the new extreme, plus the spread. That price is where the breakout is clearly working again. Because the new low is usually close to the old one, the stop is small relative to a daily move, which is the main attraction of the strategy. Size it with the position size calculator like any other trade.

Target. The nearest swing high inside the range, or twice the risk, whichever is nearer and still worth the trade. Connors and Raschke favoured short holds with a trailing stop: once the trade is in profit, move the stop to breakeven, then trail it below each new higher low. Failed breakouts often snap back fast, then stall.

Turtle Soup Plus One. A variant waits a day: the new 20-day low closes below the old one, and the entry is a buy stop at the old low on the next day. It catches breakouts that looked real at the close and then failed, at the cost of a slightly wider stop.

The intraday version. ICT traders apply the same idea to the previous day's high or low, or an obvious intraday swing, during the London or New York session. They usually add a confirmation step: after the sweep, wait for a lower timeframe change of character before entering. The ICT trading strategy covers that model in full.

When turtle soup fails

Turtle soup is a bet against breakouts, so it fails when breakouts work, and breakouts work best in strong trends and on news. A 20-day low in a market that has been falling for months is not a trap; it is the trend doing what it does.

In this GBP/USD chart the falling 20-day average shows a steady downtrend. Price breaks the 1.2700 low, dips to 1.2690 and pokes back above, triggering the buy stop. The bounce lasts two days and the trend resumes, through the protective stop at 1.2685 and on to new lows. Nothing about the level was wrong; the context was.

Filters that help:

• Context. Prefer ranges and trends that are visibly losing momentum (smaller swings, a flattening average). Skip fades against a steep, persistent trend.
• News. Do not fade a break caused by a central bank decision or a major data surprise. Those moves reflect new information, not an exhausted push.
• The speed of the reversal. The best failed breakouts reverse within the same candle or the next. A break that holds below the level for several days is not failing.

Be honest about the trade-off, too. Because the stop is tight, a normal wobble back below the level stops you out, and some trades that eventually work will cost you a loss first. That is the price of small, defined risk, and a backtest will show you how often you pay it.

How to backtest turtle soup

The original rules are mechanical, which makes turtle soup one of the easiest strategies here to test honestly.

1. Pick a market and the daily chart, and mark every new 20-day low and high over two or three years.
2. Keep only those where the previous 20-day extreme was at least four days earlier.
3. For each, check whether price came back through the old level that day or the next. If it did, record the entry, the stop below the new extreme and the outcome under your exit rule. If it did not, record a non-trade.
4. Measure results in units of risk after the spread, and collect at least fifty trades.

Then add one filter at a time, for example skipping trades against a falling 50-day average, and see whether the average result per trade improves on the same data. The academy lesson on Judas swings and false breakouts shows the intraday side of the same idea, and the drawdown calculator shows what the inevitable runs of small losses do to an account.