"Buy the dip in a trend" is the most-repeated advice in trading, and one of the least examined. We took a strict, mechanical definition of a EURUSD pullback — a counter-move inside a confirmed H4 trend — and measured what happened to every one across 16 years of data: 3,341 events.

The headline: the dip itself is a coin flip

Outcome	Count	Share
Continuation (trend resumes)	1,480	44.3%
Reversal (prior swing breaks first)	1,556	46.6%
Unresolved after 24h	305	9.1%

Of the pullbacks that resolve within a trading day, 48.7% continue and 51.3% reverse. That is the number that should reset your expectations: buying a dip because it is a dip has no edge on EURUSD. The trend does not owe you a bounce. The edge — and it is a large one — is in refusing the low-quality dips.

Depth is the first filter, and it's decisive

The single most powerful trait is how far the pullback retraced the move that preceded it:

Shallow pullbacks (retracing less than 50% of the prior leg) continue 75.2% of the time.
Deep pullbacks (50% or more) continue only 37.4% of the time.

A typical pullback in a live trend gives back roughly a quarter to six-tenths of the prior leg. Stay in the shallow end of that range and you are buying continuations; chase the deep retracements and you are, more often than not, catching reversals. Depth alone nearly doubles your hit rate.

Speed is the second filter — and it's counterintuitive

How fast the pullback moves matters as much as how far:

Slow, grinding pullbacks (bottom quartile of velocity) continue 67.3% of the time.
Sharp, fast pullbacks (top quartile) continue just 27.8% of the time.

This runs against instinct — a fast flush feels like a shakeout you should buy. In the data it is usually the first leg of a genuine reversal. A slow, reluctant drift lower is the market catching its breath; a violent drop is the market changing its mind.

The EMA20 rule is worse than a coin flip

Here is the finding that should change a lot of charts. The most-taught continuation rule is "wait for price to pull back to the 20 EMA, then buy." On EURUSD, that rule is actively harmful:

Pullbacks that reached the H1 EMA20 continued only 38.5% of the time.
Pullbacks that bottomed before reaching it continued 71.0% of the time.

The reason is mechanical: by the time price has traveled all the way to the EMA20, the pullback is no longer shallow — it has reached medium depth, which we already know is reversal territory. The EMA20 isn't a magic buy zone; it's the line where "shallow" becomes "deep." The best continuations bounce before they ever get there. Waiting for the textbook entry systematically selects for the worse trades.

Timing: where the pullback bottoms tells you what it is

The clock adds a final, independent filter. A pullback that puts in its low during the London open (around 07:00 UTC) continues about 68% of the time. A pullback bottoming during the New York overlap peak (15:00–16:00 UTC) continues only 24–25% of the time — a bottom there is far more often a real reversal than a buyable dip. Liquidity and the time of day decide whether a dip is a pause or a turn.

What this means

The exact entry, stop, and target parameters stay in the engine — a published edge is a dead edge. But the transferable lessons are unusually concrete, and they hold on more than just EURUSD:

The dip is not the signal; its character is. Shallow, slow, early-bottoming pullbacks are continuations. Deep, fast, EMA-reaching ones are reversals dressed as dips.
Question the textbook. "Buy the pullback to the 20 EMA" is a losing filter here precisely because it waits for depth that predicts failure.
Read speed, not just depth. A slow grind is a pause. A sharp flush is a warning.

The 48.7% base rate is what makes most dip-buying frustrating. The traits above are what turn it into an edge — and the edge is in the no, not the yes.