The model buys after a close below the lower band with RSI below 30 and sells after a close above the upper band with RSI above 70. Entry occurs at the next hourly open. The target is the current middle band, the stop is 1.5 ATR, and the round-trip cost is 1.2 pips.
Oversold is a condition, not a promise
Bollinger Bands measure where price sits relative to recent volatility. RSI adds a momentum condition, but neither indicator proves that price must reverse.
EUR/GBP can oscillate for long periods, yet a policy surprise or one-sided trend can keep price outside a band long enough to hit an ATR stop.
Reproducible H1 model
Long setup
Close below the lower 20-period band and RSI 14 below 30; enter at the next open.
Short setup
Close above the upper band and RSI above 70; enter at the next open.
Exit
Take profit at the live middle band. Place the stop 1.5 ATR from entry; if both are touched in one candle, count the stop first.
Position and costs
Only one trade may be open. Each trade risks 1% of initial capital and pays 1.2 pips round trip.
Signals, momentum and the equity curve
On the first chart, blue bands surround the yellow middle line. Entry markers sit beyond a band; exit markers show where the target or stop resolved the trade. The RSI panel confirms the 30/70 condition.
EUR/GBP · H1
2024-09-24 — 2026-09-23 · UTC
Start with a single marker. Check whether the signal candle closed outside a band, then verify RSI, and only then trace the next-open entry to its middle-band target or ATR stop.
RSI 14
30 / 50 / 70
$10,000 · Final value
Trades: 298
Costs reversed a small gross edge
Before modelled costs, the signals produced a small positive pip total. With 298 round trips at 1.2 pips each, costs exceeded that gross edge and the account finished lower.
The near-50% win rate was not enough. Stop distances, average win, average loss and trading frequency determine the outcome; the indicator names do not.
Trades: 298
Maximum drawdown: -21.8%
What can change the conclusion
Feed and spread
Hourly candles and executable spreads differ between brokers, especially around rollover and news.
Parameter selection
Changing 20/2, RSI thresholds or ATR distance after seeing the answer creates overfitting.
Market regime
A range-friendly period can favour mean reversion; a persistent trend can produce consecutive stops.
A better next test
- Repeat the same rules on independent broker data.
- Separate London, New York and quiet-session results.
- Test a spread filter without changing the original benchmark.
- Reserve the latest period as an untouched forward sample.
A plausible idea failed after costs
That is a valuable result. The bands and RSI generated visually convincing reversals, but the fixed rules traded too often for the edge available in this sample. A serious strategy article should show that failure instead of hiding it.
Data and methodology
Yahoo Finance · EURGBP=X · H1 · UTC · 2024-09-24 — 2026-09-23
Rules fixed before calculation; next-bar execution and stated costs included