API Access for Researchers — All data and models from this series are available via the API Gateway. Get your API key →

📖 New: Beyond the Benchmark: What AI Looks Like When It Actually Works · Free API Access →

Research API Access: All data and tools referenced in this research series are available via the Stabilarity API Gateway — free for researchers. Request your personal API key →

🔬 HPF-P Portfolio Optimizer

Decision Readiness Index (DRI) · Decision Readiness Level (DRL) · ML-augmented multi-objective optimisation for Ukrainian pharmaceutical portfolios

5-Group DRL Classification Fourier ML Forecast 500-Path Monte Carlo CVaR / MVO Optimisation 12-Month Prediction

🏭 Select Company Portfolio

Or upload your own CSV + metadata below

📂 Input Data

Portfolio CSV (SKU × Month revenue + quantity)

📊

Drop CSV or click to browse

Metadata JSON (SKU attributes, market data)

📋

Drop JSON or click to browse

🔒 Deterministic

ℹ️ How HPF-P Works — Methodology & DRL Groups ▾ expand

DRI Formula

      DRI = 0.25·R1 + 0.25·R2 + 0.20·R3 + 0.15·R4 + 0.15·R5
    

Dimension	Weight	What it measures
R1 — Data Quality	25%	Completeness, consistency, temporal coverage of SKU data
R2 — Market Observability	25%	Market share visibility, competitive data availability
R3 — Forecast Reliability	20%	ML forecast confidence, MAPE, seasonal stability
R4 — Model Applicability	15%	Portfolio diversity enabling optimisation methods
R5 — Environmental Entropy	15%	Geopolitical / regulatory disruption risk (Ukraine context)

DRL Groups — Decision Method by Readiness

DRL	DRI Range	Method	Action
DRL-1	DRI < 0.25	Abstain	Flag for data collection; do not optimise
DRL-2	0.25 – 0.45	Rule-Based	Heuristic rules (cap growth, floor decline)
DRL-3	0.45 – 0.65	LP	Linear programming within safety bounds
DRL-4	0.65 – 0.80	CVaR / MVO	Conditional Value-at-Risk mean-variance optimisation
DRL-5	DRI ≥ 0.80	ML-Augmented	Multi-objective optimisation with ML signal integration

📊 Portfolio Summary

📈 Revenue Trajectory — Historical + 12-Month Predictions

Historical (actual)

No rebalance · median

No rebalance · P10–P90

HPF rebalanced · median

HPF rebalanced · P10–P90

Bands = P10–P90 of 500 Monte Carlo paths · Forecast horizon: 12 months · Currency: UAH

💰 Economic Impact — Baseline vs HPF Rebalanced

Return Metrics

Risk Metrics

⚖️ Portfolio Weights — Before vs After Rebalancing

Weight Allocation — All SKUs

DRL Group Distribution

DRI Readiness Score per SKU

Weight Reallocation Delta

🔍 Per-SKU Forecast — Historical vs Predicted (Baseline vs HPF Contribution)

Each chart shows SKU absolute revenue in UAH. ■ Historical actual revenue · ╌ No rebalance ML forecast (weight unchanged) · ━ HPF rebalanced forecast scaled by weight change ratio (×rebalance factor). If HPF raises a SKU’s weight 2× the rebalance line sits 2× higher — showing proportional portfolio gain. Bands = P10–P90 confidence range.

All SKUs

DRL-1 Abstain

DRL-2 Rule

DRL-3 LP

DRL-4 CVaR

DRL-5 ML

📋 SKU Detail Table

Version: 1.0.0 | Last Updated: 2026-03-09 | Dependencies: HPF-P API (/hpf-api/) | Related Research: HPF Project

📡 API Dependencies

HPF-P Portfolio Optimizer API — /hpf-api/
Status: checking…
Version: –
Key endpoints: /api/health, /api/sample, /api/analyze
Algorithm: 6-module HPF pipeline — DRI/DRL scoring, multi-strategy optimization, Monte Carlo simulation
API Documentation ↗

📋 Release Notes

v1.0.0 (2026-03-09)
• Full 6-module HPF pipeline implementation
• DRI/DRL decision-readiness framework
• 7 optimization strategies with auto-selection
• 500-path Monte Carlo time-series simulation
• 12-month revenue forecasting with confidence bands
• 5 predefined Ukrainian pharma company scenarios
• Deterministic/Stochastic toggle
• Economic metrics: Sharpe, Sortino, VaR, CVaR, Max Drawdown

v0.9.0 (2026-03-03)
• Initial release with core HPF algorithm