Dogonomics - Financial Assistant Project

Development Notes - Updated 8/7/2025

A modern financial assistant that helps investors make informed bond investment decisions and stock decisions by analyzing market trends and news sentiment.

Backend Architecture

Core Technology Stack

Language: Go (Golang)
- Fast, easy concurrency
- Excellent API and HTTP support
- Direct integration with financial APIs
Database: PostgreSQL/SQLite for trend analysis and data storage
ML Pipeline: Python PyTorch → ONNX → Go ONNX Runtime

APIs and External Services

API / Tool	Purpose	Priority
Finnhub	Primary financial data and company info	High
Polygon.io	Real-time stock data and bond prices	High
EODHD	Financial news for sentiment analysis	High

Primary Stack: Finnhub and Polygon.io are the most reliable and well-documented for financial data. EODHD provides quality news data specifically for sentiment analysis.

Machine Learning Pipeline

Sentiment Analysis Model

FinBERT: Deep learning model trained on financial texts
Pipeline: PyTorch → ONNX → Go ONNX Runtime
Advantage: Domain-specific financial sentiment analysis vs. generic VADER model

@misc{yang2020finbert,
  title={FinBERT: A Pretrained Language Model for Financial Communications},
  author={Yi Yang and Mark Christopher Siy UY and Allen Huang},
  year={2020},
  eprint={2006.08097},
  archivePrefix={arXiv},
}

Prediction Algorithms

Algorithm	Purpose	Use Case
Linear Regression	Continuous yield prediction	Baseline model (+2.3% yield)
Logistic Regression	Binary classification	“70% chance bond will go up”
Random Forest	Non-linear patterns	Handles missing data well
XGBoost/LightGBM	Gradient boosting	High accuracy on structured data
ARIMA	Time series forecasting	Future price prediction
LSTM Neural Networks	Deep learning sequences	Long-term trend analysis

Go Backend Structure

Core Data Models*

package DogonomicsProcessing

import (
    "time"
    "github.com/MadebyDaris/dogonomics/sentAnalysis"
)

// StockDetailData - Main data structure for frontend integration
type StockDetailData struct {
    CompanyName         string                  `json:"companyName"`
    Description         string                  `json:"description"`
    CurrentPrice        float64                 `json:"currentPrice"`
    ChangePercentage    float64                 `json:"changePercentage"`
    Exchange            string                  `json:"exchange"`
    Symbol              string                  `json:"symbol"`
    AssetType           string                  `json:"assetType"`
    EBITDA              string                  `json:"ebitda"`
    PERatio             float64                 `json:"peRatio"`
    EPS                 float64                 `json:"eps"`
    AboutDescription    string                  `json:"aboutDescription"`
    ChartData           []ChartDataPoint        `json:"chartData"`
    TechnicalIndicators []TechnicalIndicator    `json:"technicalIndicators"`
    SentimentData       []ChartDataPoint        `json:"sentimentData"`
    News                []sentAnalysis.NewsItem `json:"news"`
    AnalyticsData       []ChartDataPoint        `json:"analyticsData"`
}

type ChartDataPoint struct {
    Timestamp time.Time `json:"timestamp"`
    Open      float64   `json:"open"`
    High      float64   `json:"high"`
    Low       float64   `json:"low"`
    Close     float64   `json:"close"`
    Volume    int64     `json:"volume"`
}

type TechnicalIndicator struct {
    Name      string    `json:"name"`
    Value     float64   `json:"value"`
    Signal    string    `json:"signal"` // "BUY", "SELL", "HOLD"
    Timestamp time.Time `json:"timestamp"`
}

type CompanyProfile struct {
    Country          string  `json:"country"`
    Currency         string  `json:"currency"`
    Exchange         string  `json:"exchange"`
    Ipo              string  `json:"ipo"`
    MarketCap        float64 `json:"marketCapitalization"`
    Name             string  `json:"name"`
    Phone            string  `json:"phone"`
    ShareOutstanding float64 `json:"shareOutstanding"`
    Ticker           string  `json:"ticker"`
    WebURL           string  `json:"weburl"`
    Logo             string  `json:"logo"`
    FinnhubIndustry  string  `json:"finnhubIndustry"`
}

Backend Architecture Modules

1. Data Processing Module (`/DogonomicsProcessing`)

Purpose: Handle all financial data fetching, processing, and storage

API integration (Finnhub, Polygon.io, EODHD)
Data transformation and validation
Database operations (PostgreSQL/SQLite)
Real-time data streaming
Technical indicator calculations

2. Sentiment Analysis Module (`/sentAnalysis`)

Purpose: Dedicated ML pipeline for news sentiment analysis

FinBERT model integration via ONNX Runtime
News data preprocessing
Sentiment scoring and aggregation
Model performance monitoring
Batch processing for historical analysis

package sentAnalysis

type NewsItem struct {
    ID          string    `json:"id"`
    Title       string    `json:"title"`
    Content     string    `json:"content"`
    Source      string    `json:"source"`
    Timestamp   time.Time `json:"timestamp"`
    Symbol      string    `json:"symbol"`
    Sentiment   float64   `json:"sentiment"`    // -1.0 to 1.0
    Confidence  float64   `json:"confidence"`   // 0.0 to 1.0
}

type SentimentBatch struct {
    Items   []NewsItem `json:"items"`
    Results []float64  `json:"results"`
}

Backend Implementation Status

Completed

Core Go data structures and types
Package structure (DogonomicsProcessing, sentAnalysis)
Basic API endpoint framework
SQLite/PostgreSQL integration setup

TODO

Complete sentiment analysis module
Implement prediction algorithms
Add caching layer for API responses
Set up monitoring and logging
API rate limiting and error handling
Unit and integration tests
FinBERT ONNX Runtime integration in /sentAnalysis
Real-time data pipeline optimization
Technical indicator calculations

Data Flow Architecture

Financial APIs (Finnhub/Polygon/EODHD)
            ↓
    Go Backend (/DogonomicsProcessing)
            ↓
    Data Processing & Storage (PostgreSQL/SQLite)
            ↓
    Sentiment Analysis (/sentAnalysis + ONNX Runtime)
            ↓
    Aggregated Results → Frontend API

Note: All timestamps from financial APIs are in UTC. Convert to Eastern Time (ET) for market-hour alignment.

Frontend Architecture

Technology Stack

Framework: Flutter
State Management: Provider/Riverpod (TBD)
Charts: fl_chart ^0.64.0
Authentication: Firebase Auth
Database: Firebase Firestore

UI Design System

Key Components

StocksPage

Features:

Interactive price charts with timeframe selector (1D, 1W, 1M, 1Y)
Three-tab navigation: Overview, Analytics, News Analysis
Real-time price updates
Technical indicators (RSI, MACD, Moving Average)

Tabs Structure:

Overview: Key metrics, company information
Analytics: Performance charts, technical indicators
News Analysis: Sentiment analysis, recent news with sentiment indicators

Authentication System

Firebase-powered login/registration
Portfolio synchronization across devices
Real-time data updates

Portfolio Management

Real-time portfolio tracking
Add/remove stocks with confirmation dialogs
Multi-device synchronization via Firebase

Frontend Implementation Status

Completed

Basic UI framework and navigation
Dark theme implementation
Stock detail pages with charts
Firebase authentication integration
Portfolio management system

In Progress

Real-time chart updates
News sentiment visualization
Performance analytics dashboard

TODO

Push notifications for price alerts
Offline data caching
Advanced charting features
User preferences and settings
Portfolio performance analytics

Firebase Database Structure

users/{userId}/
├── email: "user@example.com"
├── username: "username"
├── createdAt: timestamp
└── portfolio: [
    {
      symbol: "AAPL",
      name: "Apple Inc",
      code: "ETF",
      price: 150.25,
      change: 2.5
    }
  ]

Deployment Architecture

┌───────────────────┐     ┌──────────────────────┐
│   Flutter App     │ <-> │   Golang REST API    │
│   (Mobile/Web)    │     │ /DogonomicsProcessing │
└───────────────────┘     └────────┬─────────────┘
                                   │
     ┌─────────────────────────────┴────────────────────────────┐
     │              Financial APIs                              │
     │         (Finnhub, Polygon.io, EODHD)                    │
     └──────────────────────────────────────────────────────────┘
                                   │
                ┌──────────────────┴────────────────────┐
                │     Backend Modules                   │
                │  Data Processing + Sentiment Analysis │
                │    (PostgreSQL/SQLite + ONNX)         │
                └───────────────────────────────────────┘

Development Notes

Why Go Over Rust?

Data Sources Strategy

Use Bond ETFs (AGG, BND) as proxies for real-time bond prices
Track credit spreads and yield curves via FRED and ICE BofA
Implement sentiment analysis on financial news for market insights

Next Development Priorities

Complete ONNX integration in Go backend
Implement real-time data streaming
Add advanced portfolio analytics
Deploy production infrastructure
Add comprehensive testing suite

Developed by Daris Idrene
Website | GitHub

Dogonomics devlog n°1 Enhanced UI

Dogonomics - Financial Assistant Project

Backend Architecture

Core Technology Stack

APIs and External Services

Machine Learning Pipeline

Sentiment Analysis Model

Prediction Algorithms

Go Backend Structure

Core Data Models*

Backend Architecture Modules

1. Data Processing Module (/DogonomicsProcessing)

2. Sentiment Analysis Module (/sentAnalysis)

Backend Implementation Status

Completed

TODO

Data Flow Architecture

Frontend Architecture

Technology Stack

UI Design System

Key Components

StocksPage

Authentication System

Portfolio Management

Frontend Implementation Status

Completed

In Progress

TODO

Firebase Database Structure

Deployment Architecture

Development Notes

Why Go Over Rust?

Data Sources Strategy

Next Development Priorities

1. Data Processing Module (`/DogonomicsProcessing`)

2. Sentiment Analysis Module (`/sentAnalysis`)