Skip to content

Glossary

Quick reference for trading and platform terminology used throughout TradAI documentation.


Trading Terms

Term Definition
Backtest Historical simulation of a trading strategy using past market data to evaluate performance
Drawdown Peak-to-trough decline in portfolio value, measuring downside risk
Equity Curve Graph showing portfolio value over time during a backtest or live trading
OHLCV Open, High, Low, Close, Volume - standard candlestick data format
Pair Trading pair notation (e.g., BTC/USDT:USDT for Bitcoin perpetual futures)
Sharpe Ratio Risk-adjusted return metric; higher values indicate better risk-adjusted performance
Slippage Difference between expected and actual execution price
Stake Amount Capital allocated per trade
Stop Loss Order to sell when price drops to a specified level to limit losses
Take Profit Order to sell when price rises to a specified level to lock in gains
Timeframe Data interval for candles (e.g., 1m, 5m, 1h, 1d)
Walk-Forward Backtesting method that periodically retrains models on rolling windows
Win Rate Percentage of trades that are profitable

Platform Terms

Term Definition
ArcticDB High-performance time-series database used for storing OHLCV data on S3
Backend Service Main API gateway (port 8000) that orchestrates platform operations
CCXT Cryptocurrency exchange trading library used for unified exchange connectivity
Data Collection Service (port 8002) responsible for fetching and storing market data
DynamoDB AWS NoSQL database used for job tracking, workflow state, and status updates
ECS Task AWS container instance running a service or backtest job
Fargate AWS serverless container compute engine for running ECS tasks without managing servers
Freqtrade Open-source trading bot framework that TradAI extends
FreqAI Freqtrade's machine learning module for ML-based strategies
LoggerMixin TradAI composition pattern for structured logging (use instead of print())
MLflow ML experiment tracking service (port 5000 container / 5001 host) for logging models and metrics
Protocol Python structural typing interface (PEP 544) used for dependency inversion
Pulumi Infrastructure as Code framework (Python) used for deploying AWS resources
Pydantic Data validation library using Python type hints; all TradAI entities use BaseModel with frozen=True
Step Function AWS workflow orchestration for multi-step operations like backtesting
Strategy Service Service (port 8003) that executes strategies and backtests
tradai-common Shared library with base classes, entities, and AWS utilities
tradai-data Library for market data fetching and ArcticDB storage
tradai-strategy Library with strategy framework, validation, and metadata

Operational Terms

Term Definition
ADR Architecture Decision Record — formal documentation of a key design choice with context, decision, and consequences
DLQ Dead Letter Queue — SQS queue that receives messages that failed processing after max retries
ECS Fargate Spot Discounted (~70% savings) AWS container compute using spare capacity; used for backtests, not live trading
FIFO Queue First-In-First-Out SQS queue with exactly-once processing and message ordering guarantees
job_id Unique identifier for a backtest or training job; primary key in DynamoDB workflow-state table; linked to mlflow_run_id for experiment tracking
ResiliencePolicy Unified retry + circuit breaker decorator combining RetryConfig and CircuitBreakerConfig
trace_id End-to-end correlation ID propagated from CLI → Backend → SQS → Step Functions → ECS → DynamoDB
WorkflowState DynamoDB table (tradai-workflow-state-{env}) storing backtest job lifecycle state

Configuration Terms

Term Definition
Dry Run Simulation mode with paper trading (no real money)
Dry Run Wallet Simulated starting capital for backtests
Hyperopt Hyperparameter optimization to find best strategy settings
Stake Currency Base currency for trading (typically USDT)
Strategy Config YAML/JSON configuration defining strategy parameters
Timerange Date range for backtesting (format: YYYYMMDD-YYYYMMDD)

Metrics

Metric Description Good Value
Sharpe Ratio Risk-adjusted returns > 1.0
Sortino Ratio Downside risk-adjusted returns > 1.5
Max Drawdown Largest peak-to-trough drop < 20%
Win Rate Profitable trade percentage > 50%
Profit Factor Gross profit / Gross loss > 1.5
Expectancy Average profit per trade > 0
CAGR Compound Annual Growth Rate Varies

See Also