Reading the result

A json result has six parts. Amounts are in coin unless the field name says usd; returns and rates are decimals (0.089 is 8.9%).

Field	What it holds
`strategy`	A plain-English description of what ran, generated from the spec rather than its name.
`stats`	Performance statistics for the whole run.
`equity`	The equity curve, one point per decision time.
`positions`	One row per structure opened, with its P&L and why it closed.
`trades`	One row per leg fill, with prices, greeks and IV at entry and exit.
`warnings`, `diagnostics`	Data-quality and sample-size flags.

strategy

Check this first: it confirms the engine understood your spec.

{
  "headline": "Short call",
  "legs": [{ "sentence": "Short 20 delta call, front monthly (last Friday) at least 20 days out" }],
  "lifecycle": [
    "enter whenever flat",
    "close the position when the structure's premium falls -75% below entry",
    "close the position when the call leg's delta rises above 0.45",
    "close the position when days to expiry falls below 2"
  ],
  "sizing": "1 contracts, fixed",
  "costs": "filled at mark, 0.03% taker fee, capped at 12.5% of premium"
}

stats

number

Return over the run, and annualised.

number

Annualised volatility of daily returns, and the usual risk-adjusted ratios.

object

max_drawdown, avg_drawdown, longest_underwater_days, and the dates the worst drawdown started (max_drawdown_start) and bottomed (max_drawdown_trough).

object

Per-leg statistics: count, winners, losers, win_rate, avg_pnl, avg_win, avg_loss, best, worst, profit_factor, avg_holding_days and total_fees.

object

The same statistics per position. For multi-leg strategies this is the figure that matters: a condor whose short legs win and long legs lose is one winning position, not a 50% win rate.

object

Return per calendar year, keyed by year.

mixed

The span actually reported. With trim_to_first_trade on, start is the first trade, not your requested start.

equity

{ "timestamp": "2026-05-01T08:00:00+00:00", "equity": 0.9997, "equity_usd": 77088.7,
  "cash": 1.0111, "positions_value": -0.0114, "spot": 77111.9, "open_positions": 1 }

equity = cash + positions_value. equity_usd is equity × spot.

positions

One per structure, with the rule that closed it in reason, or expiry and end_of_run:

{ "position": "BTC 20-delta covered call-00001", "entry_time": "2026-05-01T08:00:00",
  "exit_time": "2026-05-18T08:00:00", "legs": 1, "expiry": "2026-05-29T08:00:00",
  "net_premium": 0.0114, "fees": 0.000588, "pnl": 0.008513, "reason": "take_profit_75" }

trades

One per leg, with the instrument actually traded:

{ "instrument": "BTC-29MAY26-84000-C", "type": "C", "strike": 84000, "quantity": -1.0,
  "entry_price": 0.0114, "exit_price": 0.0023, "pnl": 0.008513, "fees": 0.000588,
  "entry_dte": 28.0, "exit_dte": 11.0, "holding_days": 17.0,
  "entry_delta": 0.212, "exit_delta": 0.084, "entry_iv": 0.358, "exit_iv": 0.354,
  "reason": "take_profit_75", "position": "BTC 20-delta covered call-00001", "tags": ["call"] }

Prices are in coin per contract, as Deribit quotes inverse options.

warnings and diagnostics

Treat these as part of the answer.

You might see	What it means
`only 14 positions: ratio estimates are dominated by noise below ~30`	Too few positions for Sharpe, win rate and the other ratios to be reliable. Widen the window.
`multi-leg strategy ...: per-position win rate is 85.2% against 45.4% per leg`	Read `structures`, not `trades`, for win rate.
`40 bar-marks were carried forward because the contract had no row`	Some contracts had no mark at a decision time, so the last mark was reused. A handful is normal; many means the strategy trades illiquid strikes.

diagnostics counts the same things as numbers: fills_total, fills_degraded and fills_degraded_pct (fills priced from a fallback), stale_marks, structure_unresolved (times the structure couldn’t be built, such as a missing strike or expiry) and bars_before_first_trade.

A run with fills_degraded_pct above a few percent, or many structure_unresolved, is telling you the market couldn’t have supported the strategy as specified. Move strikes closer to the money or choose a more liquid expiry.

Backtesting

Reference

strategy

stats

equity

positions

trades

warnings and diagnostics

​strategy

​stats

​equity

​positions

​trades

​warnings and diagnostics

strategy

stats

equity

positions

trades

warnings and diagnostics