Skip to content
Rejected

Bollinger Band Continuation (Public-Rule Model)

Volatility expansion trend continuation

Rejected: 71 locked-holdout trades lost money, and double costs did not rescue the result.

Category
Trend
Window
Locked holdout: 2024-01-01–2026-06-19
Instruments
EURUSD, GBPUSD, AUDUSD, USDJPY
Timeframe
D1 from real M5 bars
Tested
2026-07-19

Pending validation

A scaled equity curve appears here once this strategy clears the data needed to compute one honestly. We don't show a curve we can't stand behind.

See what the gates require →
0.55Profit factor
-1.03Sharpe
-6.0%Max drawdown
32.4%Win rate
71Trades
1/6Gates passed
Not runPlacebo

How it works

  1. The bet

    What market behavior this strategy is wagering on.

    A close outside an expanding Bollinger Band may signal persistent directional pressure rather than an immediate reversal.

  2. How it decides

    What makes it enter, size, and exit a position.

    The clean-room model trades a directional close outside a 20-day, two-standard-deviation band when bandwidth is expanding, then manages risk with ATR-based stops.

  3. How it can break

    The regimes and failure modes that turn the edge negative.

    Band breaks often exhaust or reverse; daily execution and trading costs can consume weak continuation.

Market context

Live chart

Context only · not backtest evidence

Chart powered by TradingView. Live prices can differ from the point-in-time dataset used in the published test.

Result

This operationalization failed the locked holdout: PF 0.548, Sharpe -1.028, and 71 trades. Its 10,000-sample block-bootstrap Sharpe interval was entirely below zero (-1.976 to -0.057). The result is evidence against this specific rule set, not a claim that every Bollinger continuation method fails.

Public Bollinger rules describe band tags and “walking the band,” but do not specify a complete executable strategy. The daily timeframe, bandwidth-expansion test, and ATR exits are original clean-room research choices.

Evidence boundary

Real Dukascopy BID M5 data were resampled to Europe/Athens D1. Historical ask, variable spread, swap, slippage, and tick ordering are absent. No local MT4 compiler or Strategy Tester was available, so the EA remains demo-only and uncompiled.

Frequently asked

Is this an MT4-validated strategy?

No. It is an indicative Python replication on bid M5-derived daily bars; MT4 compilation and Strategy Tester evidence are unavailable.

Methodology: 11-gate validation — pre-registered spec, 6-gate battery, real market data. Full reproducible report: backtests/expert_pullback_public_models/REPORT.md in the source repository.Author: Validated Research Team (Methodology v1.0 — 11-gate validation). Backtests are not investment advice.