Breakout trading statistics, from a real book
Almost every breakout statistic you’ll find online is unsourced. This one isn’t. Below is every closed signal our desk has published between Feb 24, 2026 and Oct 14, 2026 — 2,502 trades, winners and losers, sliced by when they fired, on what timeframe, and on which pair. Nothing is excluded for looking bad.
Two things the numbers say that most guides don’t
Win rate and profit point in opposite directions.
Our H1 signals have a win rate of 30.2% — and they produce +317.1R, more than any other timeframe we trade. The same pattern shows up by session: London/NY overlap has the highest win rate at 39.3%, while London wins less often and returns more per trade (+0.356R against +0.253R). If you pick a session or a timeframe by win rate, the data says you will pick the wrong one.
A losing majority is compatible with a profitable book.
34.9% of these 2,502 trades made money and the book is up +538.8R. That works because every signal defines its loss before entry — a losing trade costs exactly 1R, and the winners are worth several. Any breakout “win rate” quoted without the payoff attached tells you almost nothing.
By session
Bucketed by the hour the signal fired, in UTC. Boundaries are listed in the methodology.
| Bucket | Signals | Win rate | Avg R | Total R |
|---|---|---|---|---|
| New York afternoon | 709 | 38.6% | +0.025R | +18.0R |
| Asian | 708 | 30.2% | +0.260R | +183.8R |
| London | 607 | 32.5% | +0.356R | +216.1R |
| London/NY overlap | 478 | 39.3% | +0.253R | +120.9R |
By timeframe
| Bucket | Signals | Win rate | Avg R | Total R |
|---|---|---|---|---|
| M15 | 1567 | 34.3% | +0.089R | +139.9R |
| H1 | 569 | 30.2% | +0.557R | +317.1R |
| H4 | 323 | 44.9% | +0.232R | +74.8R |
| W | 17 | n < 30 | — | +3.7R |
| D | 16 | n < 30 | — | +6.8R |
| M | 10 | n < 30 | — | -3.5R |
By strategy
| Bucket | Signals | Win rate | Avg R | Total R |
|---|---|---|---|---|
| breakout | 1457 | 30.1% | +0.219R | +318.4R |
| mean reversion | 613 | 48.9% | +0.146R | +89.6R |
| ema 200 reclaim | 168 | 32.7% | +0.473R | +79.4R |
| asian range breakout | 146 | 20.5% | +0.304R | +44.4R |
| pivot extreme fade | 50 | 34.0% | -0.102R | -5.1R |
| liquidity sweep | 27 | n < 30 | — | -7.5R |
| weekend gap fill | 23 | n < 30 | — | +17.1R |
| noise breakout | 11 | n < 30 | — | +4.9R |
| news night | 4 | n < 30 | — | -2.1R |
| overnight reversal | 2 | n < 30 | — | -0.6R |
| index snapback | 1 | n < 30 | — | +0.2R |
Pair by session
Win rate with sample size beside it. Where a pair and session combination hasn’t produced 30 closed trades yet, we show the count instead of a rate — a win rate off a dozen trades is noise, and publishing it would make this whole page less useful.
| London/NY overlap | New York afternoon | Asian | London | |
|---|---|---|---|---|
| AUD/NZD | n=1 | n=1 | n=1 | n=2 |
| AUD/USD | 36.4%n=66 | 42.5%n=80 | 30.3%n=66 | 22.4%n=58 |
| EUR/JPY | n=18 | 47.6%n=42 | 36.5%n=63 | 37.1%n=62 |
| EUR/USD | 29.7%n=37 | 38.5%n=65 | 34.7%n=72 | 38.6%n=57 |
| GBP/JPY | n=23 | 53.7%n=41 | 24.5%n=53 | 40.0%n=60 |
| GBP/USD | n=26 | 34.0%n=47 | 33.3%n=51 | 23.4%n=64 |
| NZD/USD | 70.0%n=30 | n=25 | 53.7%n=41 | n=27 |
| US100 (Nasdaq 100) | n=28 | 41.5%n=41 | n=18 | n=18 |
| US2000 (Russell 2000) | — | n=1 | — | — |
| US30 (Dow) | 25.5%n=47 | 30.2%n=63 | 17.6%n=34 | 16.1%n=31 |
| US500 (S&P 500) | 44.4%n=36 | 37.5%n=48 | 23.7%n=38 | n=15 |
| USD/CAD | 37.8%n=37 | 34.9%n=63 | 45.7%n=46 | 39.5%n=43 |
| USD/CHF | 32.5%n=40 | 47.6%n=63 | 27.3%n=44 | 42.6%n=47 |
| USD/JPY | 51.6%n=31 | 58.1%n=43 | 29.3%n=75 | 34.0%n=50 |
| WTI/USD (Crude Oil) | n=24 | 22.0%n=41 | 21.6%n=51 | 28.6%n=42 |
| XAG/USD (Silver) | n=26 | n=26 | 21.9%n=32 | n=23 |
| XAU/USD (Gold) | n=8 | n=19 | n=23 | n=8 |
Methodology
What’s counted. Every signal our desk published to subscribers that has since closed, between Feb 24, 2026 and Oct 14, 2026. Internal-only signals — ones we generated but never sent — are excluded, so these figures describe what subscribers actually received. Open positions are excluded entirely, because their outcome isn’t known yet.
When each signal was recorded. Live recording started on 13 April 2026: from that date a signal is written down when the rule fires, before the outcome is known. Everything in the book before it we rebuilt from price history, so those earlier rows were written after their trades had already finished. They are still counted here, because dropping them would be a worse kind of editing than admitting to them.
How R is calculated. R is the return as a multiple of the risk defined at entry. For a long, (exit − entry) ÷ (entry − stop); for a short, (entry − exit) ÷ (stop − entry). A trade stopped out is −1R. This is the same expression our own dashboard and weekly recap use — the numbers here cannot disagree with the numbers a subscriber sees.
A “win” is a closed trade with realized R above zero, not a trade that reached its target.
Sessions are bucketed by the UTC hour the signal fired, as a clean partition of the day: Asian 00:00–08:00, London 08:00–13:00, London/NY overlap 13:00–16:00, New York afternoon 16:00–24:00. These are wall-clock buckets, not exchange hours, and they do not shift with daylight saving.
The sample floor is 30 closed trades. Below it we publish the sample size and withhold the rate. At n=30 a 50% win rate still carries roughly a ±18 point confidence interval; quoting a rate off less than that would be dressing noise as a finding.
What this is not. It’s the record of one desk’s signals on17 instruments over Feb 24, 2026–Oct 14, 2026, not a general claim about breakout trading everywhere. It excludes spread, slippage and commission, which vary by broker. Past results don’t predict future ones.
Figures refresh automatically. Last updated Oct 15, 2026, 12:46 PM UTC.
Citing this page
Journalists, educators and researchers are welcome to quote these figures. No permission needed — a link back is all we ask, and it helps readers check the numbers themselves.
Share this dataset
2,502 closed signals, published winners and losers — win rate 34.9%, +539R total.
Read the analysis
The tables are the evidence. These walk through what they mean.
- How often breakouts actually fail
- The retest: why failed breaks die fast and winners pay slow
- The ascending triangle: rules, targets, and what the break really pays
- The calibration log: every rule change we've made, dated, losers included
- Which timeframe suits breakout trading
- The best time of day to trade
- Spotting a false breakout
- Live levels and per-instrument track records
We got receipts.
Every signal our desk fires, graded in R — winners and losers, no cherry-picking. One email, every Friday.
Published for information and education. Not investment advice. Trading involves risk, and past performance does not indicate future results.