Read backtest results¶
Backtest.run, run_inputs, and runtime.run return a pandas Series. Start with summary statistics, then inspect trades and the equity curve to understand the run.
Read common fields¶
Continue from the first backtest:
print(result[["Return [%]", "# Trades", "Win Rate [%]", "Max. Drawdown [%]"]])
trades = result["_trades"]
equity = result["_equity_curve"]
print(trades[["EntryTime", "ExitTime", "Size", "PnL"]].tail())
print(equity.tail())
| Statistic | Interpretation |
|---|---|
Return [%] |
Account return in percentage points; 5 means 5%. Includes open holdings valued at termination. |
Equity Final [$], Equity Peak [$] |
Final and maximum account equity in settlement-currency units. |
# Trades |
Closed trade fragments, including partial exits; not submitted orders or necessarily distinct lots. |
# Closed Lots |
Count of fully closed lots. |
Win Rate [%] |
Winning share of closed trade fragments in percentage points. |
Max. Drawdown [%] |
Largest equity decline from a previous peak; a negative percentage. |
Sharpe Ratio, Sortino Ratio, Calmar Ratio |
Risk-adjusted statistics; undefined inputs may produce NaN. |
Commissions [$] |
Actual charged commissions. |
Start, End, Duration |
Recorded input period. Inspect report execution information for early termination. |
Inspect execution tables¶
| Result key | What you learn |
|---|---|
_equity_curve |
Equity/drawdown path indexed by market bars. |
_trades |
Closed fragments with entry/exit times, prices, signed Size, net PnL, Commission, and ReturnPct. |
_open_trades |
Lots still open at termination, remaining entry fees, and unrealized profit/loss. |
_orders |
All requests and latest states, including canceled/rejected/expired requests. |
_order_events |
Status and trigger transitions. |
_fills |
Each actual execution: order ID, time, side, quantity, price, and fee. |
_order_sizing |
Original size mode/budget and the quantity frozen at submission. |
_trade_allocations |
How fills opened or closed particular lots. |
_positions, _account_curve |
Exposure and balance/equity/margin observations. |
_final_account |
Final account values. |
_strategy |
Completed strategy instance for inspecting your own attributes. |
_trades.ReturnPct is a fraction: 0.05 means 5%, unlike summary keys ending in [%]. A run can show nonzero return with zero closed trades when it still holds a position. Use end_policy="close_at_last_close" if you want final holdings liquidated rather than simply valued.
Save a report¶
import json
from openbacktesting.portfolio import export_result
report = export_result(result)
text = json.dumps(report, allow_nan=False)
The Portfolio module provides a complete plain-value report independent of later edits or runs. Its metric objects include value, unit, method/status information; inspect status before treating a missing value as zero. Portfolio fractions and native percentage statistics use different units.
Native underscore-prefixed tables are useful in Python. For saved/application reports, use Portfolio export. For visual data, use the separate plotting export.
Interpret limitations¶
OHLC cannot establish the true intrabar path or market liquidity. A backtest report reflects the configured execution model. Profit ratios may be undefined without losses, and calendar statistics need sufficient history. Read metric status and trading rules before comparing results.