TempusOnePS is a systematic quantitative trading ecosystem architected to bridge cutting-edge alpha discovery with ultra-reliable, low-latency live execution. Designed primarily for the Vietnamese securities and index futures market (VN30F1M) with multi-asset extensibility, the platform provides an industrial research and operations pipeline spanning raw data ingestion, automated feature engineering, machine learning target labeling, multi-agent AI research, realistic slippage-aware backtesting, broker API orchestration, and real-time operations telemetry.

• Separation of Concerns: Strict decoupling between market data pipelines, mathematical feature profiling, signal calculation, risk modeling, and order routing.
• Reproducibility & Statistical Rigor: Comprehensive tracking of dataset snapshots, deterministic feature generation, multi-barrier labeling, and walk-forward validation to mitigate data leakage and overfitting.
• Execution Resilience: Fault-tolerant connection pooling with Vietnamese broker APIs (DNSE, Entrade X) featuring asynchronous order state reconciliation, risk boundary checks, and automated failover.
• Extensible Modularity: Each component operates as an autonomous, version-controlled library or microservice while sharing normalized tabular and event protocols.

The following diagram illustrates the complete dataflow and operational pipeline across all 17 component repositories of the TempusOnePS ecosystem:

flowchart TD
    subgraph Data ["1. Data Ingestion & Analytics"]
        D1["vn-stock-data<br/>(Feeds & Historical Archives)"]
        D2["vnstock-analytics-v2<br/>(VN30F1M Futures Analytics)"]
    end

    subgraph FeatureEngineering ["2. Feature Engineering & Profiling"]
        FE1["autofcholv<br/>(Automated Feature Extraction)"]
        FE2["stock-price-patterns<br/>(40+ Candlestick Patterns)"]
        FE3["labelohlcv<br/>(Systematic Target Labeling)"]
        FE4["fl-data-profiling<br/>(Statistical Profiling & Pruning)"]
    end

    subgraph Alpha ["3. Alpha Research & Strategy"]
        A1["trading-ideas<br/>(Hypotheses & Concept Incubator)"]
        A2["ai-agent-reseacher<br/>(AI-driven Strategy Search)"]
        A3["signalx<br/>(Quant Signal Generation)"]
        A4["strategies<br/>(Production & Experimental Algos)"]
    end

    subgraph Validation ["4. Backtesting & Rules"]
        V1["tempusonebt<br/>(High-Performance Simulation)"]
        V2["trading-rules<br/>(Risk & Execution Constraints)"]
        V3["easy-visualize<br/>(Feature/Label Visual Dashboards)"]
    end

    subgraph Execution ["5. Execution & Operations"]
        E1["syxtrade<br/>(Core Execution Engine)"]
        E2["vnbrokers<br/>(DNSE & Entrade X Broker SDK)"]
        E3["tempusoneps_monitoring<br/>(Real-Time Telemetry Client)"]
        E4["aianalytics-docker<br/>(Containerized Stack)"]
    end

    Data --> FeatureEngineering
    FeatureEngineering --> Alpha
    Alpha --> Validation
    Validation --> Execution

The ecosystem comprises 17 core functional repositories across 5 operational layers (alongside organizational repository governance):

Every trading algorithm at TempusOnePS navigates a rigorous 5-stage quantitative pipeline before running with live capital:

[ Data Ingestion ] ➔ [ Feature Engineering ] ➔ [ Alpha & Signal Modeling ] ➔ [ Backtesting & Validation ] ➔ [ Live Execution & Monitoring ]

• Historical Data: Market feeds and tick-level trades are pulled and verified via vn-stock-data, converting raw tick records into partitioned Parquet databases.
• Contract Analytics: vnstock-analytics-v2 continuously tracks contract settlement cycles, term structures, and liquidity dynamics for VN30F1M index futures.

• Vectorized Generation: Raw OHLCV time-series are fed into autofcholv to calculate over 100 technical and statistical indicators.
• Geometric Patterns: stock-price-patterns scans for price formation triggers across multiple bar intervals.
• Target Labeling: labelohlcv generates robust training targets using dynamic triple-barrier methods and fixed-horizon returns.
• Dimensionality Reduction & Pruning: fl-data-profiling audits the feature set using 27 diagnostic modules to eliminate low Information Coefficient (IC) features and multicollinear redundancies.

• Ideation: Trading concepts originate in trading-ideas notebooks or are proposed autonomously by ai-agent-reseacher.
• Signal Composition: Validated concepts are encoded into modular signal generators using signalx.
• Strategy Implementation: Concrete trading strategies (e.g. 5m VN30F1M breakout with adaptive trailing stops) are coded in strategies.

• Realistic Backtesting: Strategies are simulated in tempusonebt with accurate commission schedules, bid-ask spread slippage, and latency penalties.
• Visual Diagnostics: Feature importance, strategy tear-sheets, drawdowns, and trade distributions are audited in easy-visualize.
• Constraint Enforcement: trading-rules verifies that risk parameters (max lot size, daily loss limit, mandatory close-out before ATC) are strictly defined.

• Container Deployment: Service containers are orchestrated using aianalytics-docker.
• Order Routing: syxtrade boots the validated strategy, receiving live feed ticks and routing execution orders via vnbrokers (DNSE / Entrade X).
• Telemetry & Risk Control: The trading desk observes positions, margin utilization, and broker heartbeats in real-time using tempusoneps_monitoring.

• Operating System: Linux (Ubuntu 22.04+ or Debian 12 recommended) or macOS
• Python: Version 3.10 or 3.11
• Container Engine: Docker Engine 24.0+ & Docker Compose v2
• Version Control: Git 2.34+ configured with SSH keys for GitHub access

Clone the core development workspace or individual repositories:

# Clone the parent repository containing the ecosystem
git clone git@github.com:tempusoneps/.github.git tempusoneps-ecosystem
cd tempusoneps-ecosystem

# Setup a dedicated virtual environment
python3 -m venv .venv
source .venv/bin/activate

# Upgrade packaging tools
pip install --upgrade pip setuptools wheel

# 1. Load market candles from vn-stock-data
from vnstock_data import load_futures_bars
bars = load_futures_bars(symbol="VN30F1M", timeframe="1m", start="2026-01-01")

# 2. Extract features using autofcholv & pattern recognition
from autofcholv import extract_features
from ohlcpattern import CandlestickPatternDetector

features = extract_features(bars)
patterns = CandlestickPatternDetector(bars).detect_all()

# 3. Formulate signal using signalx
from signalx import CompositeSignal, MomentumRule

signal_engine = CompositeSignal(rules=[MomentumRule(threshold=1.5)])
signals = signal_engine.generate(features, patterns)

# 4. Simulate in tempusonebt
from tempusonebt import Engine, SlippageModel

bt = Engine(data=bars, signals=signals, slippage=SlippageModel.VN30F1M())
results = bt.run()
print(f"Sharpe Ratio: {results.sharpe_ratio:.2f}, Max Drawdown: {results.max_drawdown:.2%}")

Store API credentials and brokerage tokens securely in .env files (never commit credentials):

# Brokerage API Configuration (vnbrokers)
DNSE_API_KEY=your_dnse_api_key
DNSE_SECRET_KEY=your_dnse_secret_key
ENTRADE_API_TOKEN=your_entrade_token

# Realtime Telemetry (syxtrade-monitoring)
TELEMETRY_PORT=8080
LOG_LEVEL=INFO

To maintain institutional code quality and operational stability, all repositories adhere to the following standards:

1. Strict Type Annotations: All Python modules must utilize Python typing constructs (mypy --strict compliant).
2. Deterministic Code: Random seeds must be explicitly configurable across feature extraction and model training.
3. No Lookahead Bias: Feature transformations must strictly apply past information only (expanding or rolling lookback windows without forward leak).
4. Mandatory Slippage & Fees: Backtests must incorporate exchange transaction fees, brokerage fees, and realistic spread crossing slippage.
5. Fail-Safe Live Execution: Strategies running under syxtrade must implement idempotency keys on orders and handle network disconnection gracefully.

The source code, models, and documentation within the TempusOnePS ecosystem are proprietary and confidential. Unauthorized copying, distribution, modification, or commercial exploitation is strictly prohibited without explicit written consent from the TempusOnePS team.

For security vulnerability reports or access inquiries, contact the administrative team via GitHub Organization Inquiries.