UDT-Sweep MAPE Adaptive Hybrid

Family: hybrid · Regime: high_vol · Complexity: high · Asset classes: FX, Equities, Crypto · Timeframes: M5, M15

Thesis

Institutional participants utilize liquidity sweeps to fill large orders; by entering when this behavior is detected (via Daily Sweep/FVG) during high-complexity regimes (UDT Array logic), and scaling size based on prediction error (MAPE), a quantitative edge can be harvested from the subsequent volatility expansion.

Components

• User Type Array Scalar Tree Test Fixture (regime) — Acts as a complexity filter; higher static offsets (e.g., >9000) represent regimes where data persistence and institutional 'nested' orders are theorized to be active.
• Unsupported ta.macd Fast Length (Test Case) (direction) — The ultra-fast (1.5 length) EMA crossover serves as a high-frequency momentum trigger, filtering for immediate price pressure following a sweep.
• The Daily Sweep (entry) — Identifies institutional liquidity raids and Fair Value Gaps (FVG) specifically within the high-volume New York window.
• Bollinger Bands (Standard) (exit) — Provides a volatility-adjusted mean reversion target for exiting positions.
• Mean Absolute Percentage Error (MAPE) (risk) — Calculates the 'unpredictability' of the current price action against a baseline EMA; used to scale position size inversely to model error.
• Volumatic Support/Resistance Levels [BigBeluga] (volatility_filter) — Filters out trades that are entering directly into high-volume supply or demand zones to prevent 'buying the top' of a range.

Known failure conditions

• MAPE remains consistently above 5% on major pairs, indicating the underlying EMA model is failing to track price.
• Daily Sweep bias fails to provide directionality during low-volatility summer months.
• StaticOffset regime gate (9000+) is never triggered due to lack of market 'depth' or data complexity.

Explore the full interactive blueprint with parameter ranges and evidence on WOBR StrategyVerse, or generate this strategy as an MT4/MT5 Expert Advisor with QuantMogul AI Engine (free download).