Backtest — running and reading results

Strategies saved from Builder get run against real data on this page — the engine runs asynchronously (status moves pending → running → done, the system polls automatically every 3 seconds).

Run a backtest

• Start a run — The Run Backtest button on the toolbar → pick a strategy, then set up the form
• Date range — The data range being tested — Free plan is capped at ~2 years; for indicators with a long lookback (e.g. EMA200) the engine adds warmup automatically, no need to pad it yourself
• TF — Locked to the strategy's timeframe (the tf value set on the OHLCV node)
• Initial Cash ($) — Starting capital (default $300)
• Spread ($) — Cost per trade (default $0.15) — very important: results run at spread 0 look unrealistically good, always set it to match your actual broker
• Max Open Positions — Maximum number of trades open at the same time — not a cap on total trades over the whole run

Open and read results

• Open a result as a tab — Drag a result from the left sidebar onto the chart area — multiple results can be open at once, tabs can be dragged to reorder, and each tab can view a different TF
• Trade markers — ▲ entry long (cyan) / ▼ entry short (red) / ✕ exit (yellow) + Entry/SL/TP lines — click a marker to see trade detail
• Stats panel — Win rate · Profit factor · Max drawdown · Total PnL $ · trade count · Avg RR + starting capital → final
• Auto-grade — PASS badge when WR ≥ 60% and PF ≥ 1.2 and DD ≤ 20% — missing any one = FAIL
• Equity curve — Cumulative capital curve across the test range + underwater drawdown below it (the deepest point is Max DD)
• Trade list — A table of every trade (entry/exit time, price, lot, PnL $) — click a row to highlight its marker on the chart; filter by win/loss/month (Free plan sees the latest ~200 trades)
• Import / Export — Import = bring an external backtest result (JSON) in to visualize; Export = download a report (the detailed version is Pro)

Strategy status: draft (never run yet) → tested (has results) → live (passed criteria + manually confirmed) — any graph edit always requires re-testing.