Introduction - CryptoView Pro

Overview

CryptoView Pro is an advanced cryptocurrency forecasting system developed by Data Scientist Julian E. Coronado Gil. It leverages multiple machine learning models to provide accurate price predictions with confidence intervals, combining the strengths of different algorithms for short-term, medium-term, and long-term forecasts. The platform integrates real-time data collection from cryptocurrency exchanges, sophisticated technical analysis, and automated alert systems to provide a comprehensive trading intelligence solution.

CryptoView Pro is designed for educational and research purposes. Cryptocurrency markets are highly volatile, and predictions should not be used as the sole basis for investment decisions.

Architecture

CryptoView Pro uses a modular architecture with four distinct layers:

Data Layer

Collects real-time OHLCV (Open, High, Low, Close, Volume) data from cryptocurrency exchanges using the CCXT library. Supports multiple exchanges including Kraken, Binance, and more.

from data.collectors import CryptoDataCollector

# Initialize collector for Kraken exchange
collector = CryptoDataCollector('kraken')

# Fetch historical data
df = collector.fetch_ohlcv('BTC/USDT', timeframe='1h', limit=1000)

Model Layer

Implements three specialized machine learning models:

XGBoost: Optimized for short-term predictions (1-72 hours)
Prophet: Best for medium to long-term forecasts (1 week - 1 month)
Hybrid Ensemble: Intelligently combines both models based on prediction horizon

Analysis Layer

Generates technical indicators and trading signals:

RSI (Relative Strength Index)
MACD (Moving Average Convergence Divergence)
Bollinger Bands
Multiple EMAs (9, 21, 50, 200 periods)
Custom momentum and volatility indicators

UI Layer

Interactive Streamlit dashboard providing:

Real-time price monitoring
Interactive charting with Plotly
Model comparison and backtesting
Alert configuration and monitoring

Machine Learning Models

XGBoost Model

The XGBoost model is a gradient boosting algorithm optimized for cryptocurrency time series. It excels at capturing short-term patterns and non-linear relationships.

Best Use Cases

Intraday trading (1-24 hours)
High-frequency predictions
Volatile market conditions
Feature-rich datasets

Key Features Created:

Returns across multiple timeframes (1h, 4h, 24h, 7d)
Moving averages (7, 14, 30, 50 periods)
Volatility measures (rolling standard deviation)
Volume ratios and momentum indicators
OHLC price ratios
Bollinger Band positions

Prophet Model

Meta’s Prophet model is designed for time series with strong seasonal patterns and multiple seasonal cycles. It’s particularly effective for longer-term cryptocurrency trends.

Best Use Cases

Weekly to monthly forecasts
Trend analysis
Seasonal pattern detection
Confidence interval estimation

Configuration:

from models.prophet_model import ProphetCryptoPredictor

predictor = ProphetCryptoPredictor(
    changepoint_prior_scale=0.5,  # Flexibility for trend changes
    seasonality_prior_scale=10,    # Strength of seasonality
    interval_width=0.95            # Confidence interval width
)

Hybrid Ensemble Model

The Hybrid model intelligently combines XGBoost and Prophet predictions based on the forecast horizon, providing optimal accuracy across all timeframes.

Short-Term (≤72h)

Uses 100% XGBoost for fast, accurate predictions

Medium-Term (3-7d)

Blends both models with dynamic weighting

Long-Term (>7d)

Relies primarily on Prophet for trend analysis

Ensemble Weighting

XGBoost weight = max(0, 1 - periods/168)

Technical Indicators

CryptoView Pro includes a comprehensive suite of technical indicators for market analysis:

RSI
MACD
Bollinger Bands
EMAs

Relative Strength Index measures momentum and identifies overbought/oversold conditions.

from utils.indicators import TechnicalIndicators

# Calculate RSI with 14-period window
df['rsi'] = TechnicalIndicators.calculate_rsi(df['close'], periods=14)

# Trading signals
# RSI > 70: Overbought (potential sell)
# RSI < 30: Oversold (potential buy)

Moving Average Convergence Divergence identifies trend changes and momentum shifts.

# Calculate MACD with standard settings
macd, signal, histogram = TechnicalIndicators.calculate_macd(
    df['close'],
    fast=12,
    slow=26,
    signal=9
)

# Bullish crossover: MACD crosses above signal line
# Bearish crossover: MACD crosses below signal line

Bollinger Bands show price volatility and potential reversal points.

# Calculate Bollinger Bands
upper, middle, lower = TechnicalIndicators.calculate_bollinger_bands(
    df['close'],
    period=20,
    std_dev=2
)

# Price near upper band: Potentially overbought
# Price near lower band: Potentially oversold

Exponential Moving Averages smooth price data and identify trends.

# Calculate multiple EMAs
for period in [9, 21, 50, 200]:
    df[f'ema_{period}'] = TechnicalIndicators.calculate_ema(
        df['close'],
        period
    )

# Golden Cross: EMA 50 crosses above EMA 200 (bullish)
# Death Cross: EMA 50 crosses below EMA 200 (bearish)

Real-Time Alerts

CryptoView Pro includes a sophisticated alert system with Telegram integration for instant notifications.

Alert Types

Price Alerts

Triggers when price changes exceed threshold

Default: ±5% movement
Customizable per asset

RSI Alerts

Notifies on overbought/oversold conditions

Overbought: RSI > 70
Oversold: RSI < 30

Volume Alerts

Detects unusual trading activity

Threshold: 2x average volume
Indicates market interest

MACD Signals

Identifies momentum changes

Bullish/bearish crossovers
Trend reversals

Configuration

Alerts are configured in config/settings.py:

ALERT_THRESHOLDS = {
    'price_change_pct': 5.0,      # 5% price movement
    'rsi_overbought': 70,          # RSI upper threshold
    'rsi_oversold': 30,            # RSI lower threshold
    'volume_spike': 2.0            # 2x volume increase
}

# Telegram credentials (set in .env file)
TELEGRAM_BOT_TOKEN = os.getenv('TELEGRAM_BOT_TOKEN', '')
TELEGRAM_CHAT_ID = os.getenv('TELEGRAM_CHAT_ID', '')

Always store API keys and bot tokens in environment variables, never commit them to version control.

Supported Assets

CryptoView Pro supports 12 major cryptocurrencies by default:

AVAILABLE_CRYPTOS = [
    'BTC/USDT', 'ETH/USDT', 'BNB/USDT', 'SOL/USDT',
    'ADA/USDT', 'XRP/USDT', 'DOGE/USDT', 'DOT/USDT',
    'MATIC/USDT', 'AVAX/USDT', 'LINK/USDT', 'UNI/USDT'
]

Timeframes

Multiple timeframes for different trading strategies:

1m: Scalping and micro-trends
5m: Short-term momentum
15m: Intraday patterns
1h: Standard analysis (default)
4h: Swing trading
1d: Position trading

Backtesting

CryptoView Pro includes comprehensive backtesting capabilities to evaluate model performance on historical data:

from utils.backtesting import Backtester

# Initialize backtester
backtester = Backtester(model=predictor)

# Run backtest on historical data
results = backtester.run(
    df=historical_data,
    prediction_horizon=24,  # 24-hour predictions
    n_splits=5             # 5-fold time series cross-validation
)

# Metrics included:
# - RMSE, MAE, MAPE
# - Directional accuracy
# - Sharpe ratio
# - Maximum drawdown

Performance Characteristics

Data Points

Processes 500-2000 historical candles for training

Update Frequency

Real-time updates every 30 seconds (configurable)

Prediction Speed

Generates forecasts in 2-5 seconds

Memory Usage

~200-500 MB typical footprint

Cache Duration

60-second TTL for API responses

API Rate Limits

CCXT rate limiting enabled

Next Steps

Installation

Install all dependencies and set up your environment

Quickstart

Get your first predictions running in 5 minutes

Model Selection

Learn which model to use for different scenarios

Configuration

Configure exchanges, alerts, and model parameters

Get Started

Core Concepts

ML Models

Features

Configuration

​Overview

​Architecture

​Machine Learning Models

​XGBoost Model

Best Use Cases

​Prophet Model

Best Use Cases

​Hybrid Ensemble Model

Short-Term (≤72h)

Medium-Term (3-7d)

Long-Term (>7d)

Ensemble Weighting

​Technical Indicators

​Real-Time Alerts

​Alert Types

Price Alerts

RSI Alerts

Volume Alerts

MACD Signals

​Configuration

​Supported Assets

​Timeframes

​Backtesting

​Performance Characteristics

Data Points

Update Frequency

Prediction Speed

Memory Usage

Cache Duration

API Rate Limits

​Next Steps

Installation

Quickstart

Model Selection

Configuration

Build docs developers (and LLMs) love

Overview

Architecture

Machine Learning Models

XGBoost Model

Prophet Model

Hybrid Ensemble Model

Technical Indicators

Real-Time Alerts

Alert Types

Configuration

Supported Assets

Timeframes

Backtesting

Performance Characteristics

Next Steps