Adaptive Stop-Loss & Position-Sizing Framework for Systematic Equity Strategies

Customer

Challenge

From Pattern Ideas to Defensible Risk Rules

Pattern detections produced mixed visual quality; the client asked for explicit stop-loss and exit frameworks to quantify hit-rate, drawdown and expectancy across very large universes.

• From 'nice charts' to defensible risk rules: Pattern detections produced mixed visual quality; explicit stop-loss and exit frameworks needed to quantify hit-rate, drawdown and expectancy
• Scale and liquidity: Research needed to run over 10k+ securities (US + European venues)
• Team-usable controls: Simple control panel ('mask') to enter weights/thresholds and apply the same risk logic consistently in live trading via IBKR
• Avoiding look-ahead & bias: Risk engine had to be compatible with signal timestamps, enforce 'detect-then-act' causality, and keep evaluation logic clean

Solution

Parametric Risk-Control Engine

Marsbridge fielded a compact squad—Quant Lead, Risk Engineer, Data Engineer, and Execution/IBKR Integrator—to (1) harden large-universe data, (2) codify multi-stage stops and trailing exits, (3) scale runs + reporting, and (4) prepare an IBKR-ready parametric entry/exit "mask."

Large-universe data & liquidity hygiene

Built a multi-exchange EOD warehouse (US + XETRA/PA/SW/MI/VX, etc.), with automated updates and PDF batch exports for review. Introduced a fast liquidity screen to filter out illiquid names.

Multi-stage stop-loss and trailing-exit model

Implemented ATR-based stop levels with configurable multipliers and dual-regime trailing exits that adapt to volatility conditions. Created signal-agnostic integration layer compatible with multiple pattern detectors.

Backtest harness & reporting

Vectorized runner to compute per-scenario statistics over large universes; improved detector speed by ~200× to enable full-universe sweeps and generate multi-hundred-page PDFs for analyst review.

IBKR execution mask (parametric UI)

Prepared a frontend mask and parameter spec for IBKR integration, allowing consistent application of risk rules in live trading without hand-rewriting orders.

Technologies & tools

Process

Data foundation—EOD ingest, multi-exchange coverage. Quality & liquidity pass—fast screening. Risk framework design—multi-stage stops, trailing exits. Signal-agnostic wiring—compatible with multiple detectors. Backtests & review—vectorized runs, PDF generation. IBKR mask prep—parametric UI for live trading.

1. Data foundation
2. Quality & liquidity pass
3. Risk framework design
4. Signal-agnostic wiring
5. Backtests & review
6. IBKR mask prep

Results

• 38k+ Securities: Full-universe coverage with fast execution
• 200× Faster: Vectorized detector performance improvement
• Multi-Stage: Adaptive stop-loss and trailing exit system

Delivered a reusable risk-control engine with multi-stage stops and trailing exits. Achieved end-to-end runs across ~38k equities with vectorized performance. Implemented liquidity realism through fast pre-filtering. Provided IBKR-ready parameter mask for live deployment.