Double Top Pattern: Rules and False Signals

A double top pattern is a reversal pattern that forms when price rallies to a high, pulls back, rallies a second time to roughly the same high, and then turns down. It signals that buyers tried twice to push price higher and failed both times. The pattern isn't confirmed until price breaks below the low of the pullback between the two peaks, called the neckline. Most traders who lose money on this pattern act on the shape alone and skip that last step.
What is a double top pattern?
Picture ES rallying from 4900 to 5040, pulling back to 4980, then rallying again to 5045 before rolling over. That's the raw shape: two peaks at nearly the same price, separated by one pullback. The peaks don't need to match to the tick. A difference of a few points on ES is normal and doesn't invalidate the pattern.
The pullback low between the two peaks is the neckline. In this example that's 4980. The pattern is still just a shape at this point. It becomes a signal only when price closes below that neckline after the second peak.
The rules that define a valid double top
Four conditions need to be present. First, a clear prior uptrend, since a double top is a reversal pattern and you can't reverse a trend that wasn't there. Second, two peaks at approximately the same level, generally within 1% to 2% of each other on an index future like ES. Third, a visible pullback between the two peaks deep enough to look like an actual retracement, not just a one-bar dip. Fourth, a close below the neckline on the move down from the second peak.
That fourth condition is where most traders stop reading the pattern correctly. A wick that pokes below the neckline intrabar and closes back above it does not confirm anything. You need a completed candle that closes below the level, on whatever timeframe you're trading it.
A double top isn't confirmed by the shape on the chart. It's confirmed by a closed candle below the neckline.
Why the neckline break fails so often
The neckline on a double top sits at an obvious, widely watched level. Every trader looking at the same chart sees the same 4980. Stop orders from traders who bought the second peak cluster just below that number, and buy orders from traders anticipating the breakdown cluster there too.
That concentration of orders is exactly why the first touch of the neckline often produces a sharp poke through the level, a quick flush of stops, and then a snap back above it. Price prints below 4980, triggers the resting sell stops, absorbs them, and reverses. Anyone who shorted on the first break is now underwater on a pattern that looked textbook five minutes earlier.
The confirmation that avoids the false break
The fix is simple to state and hard to sit through: wait for the close, not the touch. If you're trading the daily chart, that means waiting for the daily candle to close below the neckline, not just print a low below it. If you're trading a 5 minute chart intraday, wait for a 5 minute close.
A stronger version adds a retest. After the close below the neckline, wait for price to rally back up toward that same level and get rejected there, turning old support into new resistance. On the ES example, that would mean a close below 4980 followed by a bounce back to 4980-4985 that fails and rolls back down. That retest costs you some of the move, but it removes most of the single-tick fake-outs.
A volume filter helps on instruments where you have reliable volume, though ES volume is less informative than it is on individual stocks since so much of it is algorithmic and spread across venues. If you do use it, look for the breakdown candle to trade on volume at or above the recent average. A breakdown on thin volume is more likely to be the kind that reverses within a session.
Measuring the target after confirmation
Once the neckline break is confirmed, the standard measured target is the height of the pattern projected down from the neckline. If the peaks sit at 5040 and the neckline sits at 4980, the pattern height is 60 points. Projected from the neckline, that gives a target near 4920.
Treat that number as a reference, not a promise. Price frequently stalls before reaching the full measured move, and it sometimes overshoots it. The measured target tells you what the pattern implies, not what the market is obligated to do.