Skip to the research
Runs · Measured

How Long Do Winning Streaks Actually Last?

A run of higher closes feels like a trend. Across 1,595 GBP/USD days there were 332 such runs, and the distribution looks a great deal like what repeatedly flipping a coin produces — which is the point.
332runs observed
50%are a single day
42runs of 4 or more
1,595days measured
How this was measured. Computed from 1,898,746 one-minute GBPUSD candles covering 2021-01-03 to 2026-04-30 — 1,595 trading days. Range means high minus low over the stated period, grouped by real clock time in UTC rather than by bar count, so an hour is an hour even across a weekend. Figures are in pips. The underlying numbers are published as JSON at https://fxabsolute.com/stats/GBPUSD.json.

Run lengths

Consecutive up daysTimes observedShare of runsA coin flip would give
116750.3%50.0%
28525.6%25.0%
33811.4%12.5%
4257.5%6.3%
592.7%3.1%
661.8%1.6%
710.3%0.8%
8 or more10.3%0.4%

The last column is the comparison that matters. A fair coin produces a run of exactly n heads with probability 1/2ⁿ, and the observed shares track it closely. Long runs feel significant because they are memorable, not because they are rare relative to chance.

What this rules out

It does not rule out trends. Daily direction is a crude measure and a market can trend steadily while alternating closes. What it does rule out is treating a four-day run as information — at 7.5% of runs against chance's 6.3%, it is close to what randomness delivers.

The same measure across nine pairs

PairRunsSingle-day runsRuns of 4+Share
GBP/USD33250.3%4212.7%
EUR/USD34150.7%3510.3%
USD/JPY34146.0%4814.1%
GBP/JPY33041.2%4914.8%
EUR/JPY32545.2%4915.1%
AUD/USD36049.4%328.9%
AUD/CAD35650.8%3911.0%
EUR/GBP33951.6%3610.6%
USD/CAD33647.9%4714.0%

Chance puts single-day runs at 50% and runs of four or more at 6.3%. Every pair lands close to both, which is the finding: nine independent markets, one distribution, and it is the distribution of a coin.

Why streaks feel meaningful anyway

Because a run of four is memorable and the ninety-six ordinary sequences around it are not. The same asymmetry makes a losing streak feel like a broken strategy when it is an ordinary draw — at a 50% win rate, a run of five losses appears roughly once every thirty-two sequences, which is to say regularly.

The practical consequence is about position sizing rather than entries. A plan has to survive the streaks the distribution guarantees, and the distribution above says how long those are. Sizing that assumes a bad run stops at three is sizing for a market that does not exist.

Where sequence information does live

Not in direction, but in width. The same data shows a top-quartile range day is followed by another 38.9% of the time against a 25% base rate — a real and repeatable dependency. Volatility remembers what happened yesterday; direction does not.

FXAbsolute free backtester

Test the finding on the chart—not just on the page

Open FXAbsolute and replay EUR/JPY candle by candle. Turn the observation into explicit entry, stop, target and exit rules, then measure the result yourself.

Free in the browser · no download · no credit card · no sign-in required to start

Questions

How often does forex close higher several days in a row?

Runs of consecutive higher closes on GBP/USD follow roughly the distribution a coin flip produces — about half of all runs are a single day, and each additional day roughly halves the frequency.

Is a long winning streak meaningful?

Rarely on its own. Streak lengths in this sample track chance closely, so a run is weak evidence of anything beyond having happened.

Does this mean markets are random?

No. It means daily close direction, taken alone, carries little sequential information. Range and volatility do cluster measurably — direction is the part that resists.