Skip to content

BenYahMin/DeltaFlow

Repository files navigation

DeltaFlow: Real-Time ARIMA-GARCH Quantitative Inference Pipeline

DeltaFlow is an enterprise-grade automated quantitative analytics platform designed to ingest high-frequency market data, model statistical distribution variations via an integrated ARIMA-GARCH forecasting framework, and route risk-adjusted structural entry signals downstream to local databases and messaging layers.


🛰️ System Architecture Overview

The system is engineered as an interconnected array of decoupled micro-modules, ensuring high cohesion and preventing state corruption across calculation loops:

  • predictive_model.py: The mathematical core. Calculates asset log returns, runs unit-root stationarity testing, and optimizes joint $ARIMA(1, 0, 1) - GARCH(1, 1)$ architectures to generate out-of-sample forward expectation returns and instantaneous volatility fields.
  • signal_generator.py: Risk-management node. Normalizes distribution output data maps using custom rolling Z-score parameters ($1.25z$ threshold barriers) to output position directives (BUY, HOLD, SELL) alongside defensive bracket margins.
  • database_manager.py: A transactional storage interface that safely logs multi-column model matrices into a persistent local SQLite data warehouse (deltaflow_metrics.db) without leaving hanging file handles.
  • notifier.py: Network communication client. Formats real-time analytical run snapshots into stylized markdown logs and routes them downstream via Telegram API sockets.
  • run_pipeline.py: Sequential orchestrator that fits, validates, logs, and broadcasts a single end-to-end execution pass.
  • worker_loop.py: An insulated automated daemon thread that synchronizes execution loops precisely with historical candle boundaries to eliminate chronological clock drift.
  • dashboard_ui.py: A desktop terminal front-end written in CustomTkinter that isolates the worker daemon thread while providing real-time data table auditing.
  • test_deltaflow.py: Self-contained Quality Assurance (QA) suite executing synthetic matrix validation tests across the pipeline.

🛠️ Production Installation & Requirements

Ensure you are using Python 3.11+ (Fully verified on Python 3.14). Install the mathematical modeling packages and modern graphical window frameworks:

pip install pandas numpy arch statsmodels customtkinter requests
🚀 Execution & Operational Modes
DeltaFlow supports multiple operational workflows depending on your environment needs. Navigate into your workspace directory first:

PowerShell
cd C:\Users\benja\Documents\Main\deltaflow
1. Launch the Visual Monitoring Dashboard (Recommended)
Spawns a dark-mode graphical user interface to monitor the engine parameters visually. It starts the background worker loop automatically and updates tables live every 3 seconds:

PowerShell
python dashboard_ui.py
2. Run Headless Continuous Core Daemon
Launches an autonomous command-line process that sleeps and self-corrects until the closing tick of the next hourly market candle:

PowerShell
python worker_loop.py
3. Run a Single Test-Trace Pass
Orchestrates a single discrete data-ingestion, estimation, persistence, and messaging run sequence:

PowerShell
python run_pipeline.py
4. Execute the System Test Suite
Runs the internal unit tests to validate system integrity and verify math convergence loops are performing perfectly:

PowerShell
python test_deltaflow.py
📋 Sample Run Metric Record Output
When running, the execution pipeline generates clean, highly detailed, scannable structural summary diagnostics:

Plaintext
============================================================
🚀 EXECUTING DELTAFLOW REAL-TIME INFERENCE PIPELINE
============================================================
2026-06-28 14:41:01 [INFO] DeltaFlow.Orchestrator: Market Data Loaded | Spot: $60,302.01
2026-06-28 14:41:01 [INFO] DeltaFlow.PredictiveModel: Stationarity metrics updated: Stationary | p-val: 0.0000
2026-06-28 14:41:01 [INFO] DeltaFlow.PredictiveModel: ARIMA optimization complete. Structural AIC: 779.73
2026-06-28 14:41:02 [INFO] DeltaFlow.PredictiveModel: GARCH optimization complete. Inst. Volatility: 0.003886
2026-06-28 14:41:02 [INFO] DeltaFlow.SignalGenerator: Signal Matrix Computed | Action: HOLD (Z-Score: -0.08)
2026-06-28 14:41:02 [INFO] DeltaFlow.Notifier: Telemetric execution alert successfully routed to Telegram client.

==================================================
📋 PIPELINE RUN RESULTS SUMMARY
==================================================
Target Timestamp:     2026-06-28 11:00:00+00:00
Signal Z-Score:       -0.0828
Position Directive:   👉 HOLD 👈
Market Vol Regime:    NORMAL_VOLATILITY
==================================================

About

Real-time quant pipeline utilizing a joint ARIMA-GARCH framework to model conditional mean and asset volatility. Features precise candle-boundary synchronization to eliminate clock drift, a CustomTkinter desktop UI monitor, SQLite persistence, and automated Telegram messaging alerts.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages