Skip to main content

Understanding the output

TradeJS output is meant for analysis and comparison, not for promises about future behavior.

Backtest Output​

A backtest usually produces:

  • console progress and summary lines
  • per-test statistics in Redis
  • result artifacts for orders/positions when available
  • optional AI or ML dataset chunks
  • result indexes that the app can display

The exact shape depends on strategy config, connector, cache mode, and enabled AI/ML flags.

Common Metrics​

  • orders: number of closed simulated orders.
  • wins: orders with positive result.
  • losses: orders with negative result.
  • winRate: wins divided by closed orders.
  • netProfit: historical net result for the run.
  • maxDrawdown: largest historical peak-to-trough decline.
  • riskRewardRatio: relationship between average reward and risk when available.

Signal Output​

A signal describes a strategy decision at a candle timestamp. It commonly includes:

  • strategy
  • symbol
  • interval
  • direction
  • timestamp
  • prices.currentPrice
  • prices.takeProfitPrice
  • prices.stopLossPrice
  • figures for chart inspection
  • indicators and additionalIndicators for context

AI/ML Output​

When enabled, AI/ML layers may add:

  • AI analysis records such as analysis:<symbol>:<signalId>
  • ML feature rows
  • approval or rejection metadata
  • quality and threshold values

These are research and gating artifacts. They should be validated on historical data and monitored in runtime.

How To Read Results​

Use results to answer practical questions:

  • Did the strategy trade when expected?
  • Are entries and exits explainable from the candle data?
  • Do metrics remain stable across symbols and periods?
  • Are fees, slippage, and fill timing assumptions realistic?
  • Does the same logic behave consistently in backtest and runtime?

Next: Backtesting caveats.