Research Output

Models

Validated, automated strategies built from hypothesis-driven research. Logic and results are shared publicly. Code is not.

MODEL 01 Live Testing

Structural Displacement

Institutional Order Flow — Forex

EURUSD GBPUSD AUDUSD USDCAD USDCHF NZDUSD
Win Rate
27–35%varies by spread
Intentionally sub-50% — edge is in RR asymmetry
Avg Win RR
3.0–4.5R
Winning trades return 3–4.5× amount risked
Expectancy
+0.31R avg
+0.25R to +0.37R per trade across portfolio
Sample Size
26ktrades
At 0.0 pip spread across 15-year dataset
Profitable Years
15/ 15
Gross level — including 2024 & 2025 OOS
Risk Per Trade
0.35%
Fixed % of current account equity
CAGR — Fixed Equity
9–19% p.a.
Spread-dependent, non-compounding basis
CAGR — Compounding
14–87% p.a.
Theoretical maximum — requires equity management at scale
// NOTE All figures are derived from backtested and out-of-sample validated historical data. Live trading commenced recently. Live results will be published as they accumulate.
Strategy Logic

The edge

What it exploits

The strategy targets a specific and repeatable market behaviour — the tendency for price to initiate a directional displacement from a structurally significant level and follow through toward the next objective structural target.

Two conditions must align simultaneously for a signal to be valid. Structural location and displacement quality. Both must be present. Neither alone is sufficient.

Structural location

Entries are only considered at algorithmically identified structural reference points — price levels derived from objective market geometry that are meaningful to institutional participants. These are not manually drawn support and resistance lines. They are systematically defined zones based on specific price action characteristics.

Displacement quality

Not all moves from a structural level are tradeable. The strategy specifically identifies weak displacement initiations — the early phase of a directional institutional order flow imbalance captured at its initiation point. Entering early produces a structurally tight stop and a favourable risk-reward ratio to the next objective target.

Entry Conditions

Sequential filter pipeline

All conditions are applied in sequence. A trade is only entered when every filter passes. Any single failure rejects the signal entirely.

01
Signal Detection
A weak displacement initiation pattern is identified algorithmically on the M15 timeframe. The pattern must belong to a validated scenario category for that specific pair — validated across 13 years of development data.
02
Session Filter
All three major forex sessions are active across all six pairs. Asian 17:00–03:00, London 03:00–10:00, New York 10:00–17:00 server time.
03
Sequence Distribution Filter
The preceding market context is classified and must belong to one of three validated distribution types. Signals from unclassifiable contexts are automatically rejected.
04
Initiation Context Filter
The displacement initiation context must belong to a pair-specific validated set. For AUDUSD and EURUSD, one combined context type is excluded based on persistent underperformance.
05
Structural Geometry Filter
The trade's proportional stop loss size and initial risk-reward ratio must fall within a validated two-dimensional whitelist specific to each pair.
06
Pre-Entry Viability Filter
Before any order is placed, the system calculates the expected net dollar value of the trade after all execution costs — commission and live spread. If the expected value is insufficient, the trade is rejected. No order placed, no cost incurred.
07
Single Position Rule
Only one trade is open across all six pairs at any time. If a new signal fires while a position is active, the higher-quality trade is retained based on an empirical quality ranking.
Risk & Exits

Position management

Stop Loss

Hard stop placed at the structural invalidation level at entry. Fixed at entry — does not move under any circumstances. Sized so a full stop loss hit costs exactly 0.35% of current account equity.

Take Profit

Placed at the nearest objective structural target beyond entry. Dynamic — tracks evolving structure in real-time. A small fixed pip extension is applied per pair to account for minor structural overshoot.

No Time Exit

Trades remain open until stop loss or take profit is hit. No maximum hold period. Duration varies from minutes to multiple sessions.

Position Sizing

Risk per trade fixed at 0.35% of current account equity. Lot size calculated dynamically at each entry based on stop distance and pip value. Maximum leverage capped at 30x.

Maximum Viable Spread Per Pair

Pair Max Spread Session Performance Note
EURUSD 0.2 pip Strong across all sessions
GBPUSD 0.2 pip Strong across all sessions
AUDUSD 0.2 pip Spread-sensitive — viable near zero
USDCAD 0.3 pip Strong across all sessions
USDCHF 0.2 pip Spread-sensitive
NZDUSD 0.1 pip Most spread-sensitive pair in portfolio
Validation

How it was tested

2011 – 2023 Development Period

13 years of data used for all conditioning decisions. No validation data examined at any point during development. Strict separation maintained throughout.

→
2024 – 2025 Out-of-Sample Validation

2 years of completely unseen data examined only after all development decisions were finalised and locked. All six pairs showed positive results.

→
120,000 Runs Monte Carlo Simulation

Two Monte Carlo simulations across 10,000 iterations per spread level. Both confirmed zero probability of negative annualised return under any realistic trade sequencing.

// 2024 NOTE 2024 produced below-average results across most pairs — attributed to Bank of Japan policy normalisation disrupting carry trade dynamics. 2025 partial recovery across most pairs suggests this was a temporary regime rather than permanent structural change. This remains the primary ongoing monitoring concern.
Live Results

Performance

// PENDING

Live trading has recently commenced. Results will be published here as they accumulate. All historical figures are from the validated backtesting dataset.