Minimum
Trade CSV
Canonical core trade fields only
Canonical Upload Specification
Strategy Robustness Lab Docs
Accepted upload formats, supported fields, runtime assumptions, and what each artifact unlocks.
The lab can analyze multiple artifact classes, but richer uploads unlock deeper diagnostics and stronger interpretation fidelity. Treat this page as the canonical upload and runtime specification for Strategy Robustness Lab.
Recommended
Rich trade CSV
Add MAE/MFE + risk + R fields
Richest accepted
bundle_v1 ZIP
trades + metadata + assumptions + params + optional OHLCV
Runtime assumptions
Benchmark + account size + risk %
Set before running analysis
Accepted Upload Types
Trade CSV
Extension: .csv
Closed trades table resolved through canonical field aliases.
Unlocks: overview, distribution, monte carlo, ruin, report (execution/regimes may be limited).
For: fastest baseline validation intake.
Structured bundle (bundle_v1)
Extension: .zip
ZIP with required manifest.json + trades.csv; optional metadata/assumptions/params/OHLCV files.
Unlocks: richer eligibility and context-aware diagnostics.
For: institutional upload workflows and reproducibility.
Research bundle profile
Container: bundle_v1 with artifact_type: research_bundle
Not a separate parser route; this is a manifest-level artifact profile under bundle_v1.
Unlocks: full diagnostic availability when artifact richness is classified as research_complete.
For: teams supplying full contextual artifacts.
Trade CSV Specification
Core required fields
| Field | Required | Meaning | Example | Accepted aliases |
|---|---|---|---|---|
| symbol | Required | Instrument identifier. | BTCUSDT | symbol, ticker, instrument, asset |
| side | Required | Trade direction (normalizes to long/short). | long | side, direction, position_side, trade_side |
| entry_time | Required | Entry timestamp (UTC ISO after normalization). | 2026-01-05T10:00:00Z | entry_time, entry time, entry, opened_at, open_time |
| exit_time | Required | Exit timestamp (UTC ISO after normalization). | 2026-01-05T14:30:00Z | exit_time, exit time, exit, closed_at, close_time |
| entry_price | Required | Entry price. | 43125.50 | entry_price, entry price, open_price, buy_price |
| exit_price | Required | Exit price. | 43520.00 | exit_price, exit price, close_price, sell_price |
| quantity | Required | Position size (> 0). | 0.50 | quantity, qty, size, position_size |
Optional fields — economics
| Field | Required | Meaning | Example | Accepted aliases |
|---|---|---|---|---|
| pnl | Optional | Absolute P&L for trade. | 197.25 | pnl, profit, net_pnl, realized_pnl |
| pnl_pct | Optional | Trade return in percent units. | 0.92 | pnl_pct, return_pct, pnl_percent, roi |
| fees | Optional | Commissions/fees. | 6.75 | fees, fee, commission, cost |
| duration_seconds | Optional | Trade duration in seconds. | 16200 | duration_seconds |
Optional fields — excursion
| Field | Required | Meaning | Example | Accepted aliases |
|---|---|---|---|---|
| mae | Optional | Max adverse excursion. | -120.00 | mae, mae_price, max_adverse_excursion |
| mfe | Optional | Max favorable excursion. | 320.00 | mfe, mfe_price, max_favorable_excursion |
Optional fields — risk
| Field | Required | Meaning | Example | Accepted aliases |
|---|---|---|---|---|
| risk_amount | Optional | Risk capital tied to trade. | 1000 | risk_amount, risk, risk_amt |
| stop_distance | Optional | Entry-to-stop distance. | 80 | stop_distance, entry_stop_distance, stop_dist |
| r_multiple | Optional | Net/generic R multiple. | 1.75 | r_multiple, r, r_value |
| r_multiple_net | Optional | Net R multiple. | 1.62 | r_multiple_net |
| r_multiple_gross | Optional | Gross R multiple. | 1.70 | r_multiple_gross |
Optional fields — metadata
| Field | Required | Meaning | Example | Accepted aliases |
|---|---|---|---|---|
| trade_id | Optional | Unique trade identifier. | T-2026-00142 | trade_id, id, ticket, order_id |
| strategy_name | Optional | Strategy label. | mean_revert_v4 | strategy_name, strategy, system |
| timeframe | Optional | Execution timeframe. | 1h | timeframe, tf, interval |
| market | Optional | Market or asset class hint. | crypto | market, asset_class |
| exchange | Optional | Execution venue. | binance | exchange, venue |
| notes | Optional | Free-form notes. | post-FOMC continuation | notes |
| entry_reason | Optional | Entry rationale text. | breakout confirmation | entry_reason |
| exit_reason | Optional | Exit rationale text. | target hit | exit_reason |
Side normalization accepts long aliases (buy, long, b, bull) and short aliases (sell, short, s, bear).
What Each Field Unlocks
| Input richness | Primary unlocks |
|---|---|
| Core trade fields | Overview, Distribution, Monte Carlo, Ruin, Report available. |
| MAE/MFE | Improved excursion interpretation and richer distribution context. |
| risk_amount / stop_distance / R-multiples | Stronger execution-quality framing and risk translation. |
| Bundle assumptions.json | Execution diagnostic can move from limited to available. |
| Bundle params.json | Stability can move from limited/unavailable toward available. |
| Bundle ohlcv.csv / ohlcv.parquet | Regimes can move from limited/unavailable toward available. |
| Runtime account_size + risk_per_trade_pct | Sizing-aware survivability interpretation for ruin diagnostics. |
| Benchmark selection | Benchmark-relative comparison diagnostics when benchmark data is available. |
Runtime Assumptions You Can Set
Benchmark
Optional but recommended for relative performance framing. Used to configure benchmark comparison diagnostics.
Account size
Optional positive number. Used for capital translation and survivability framing in risk diagnostics.
Risk per trade %
Optional positive percent. Used in sizing-aware ruin/survivability interpretation.
Runtime assumptions enrich interpretation; they do not replace missing trade-level fields.
Benchmark Selection
- • Supported benchmark IDs: BTC, SPY, XAUUSD, DXY.
- • Modes: Auto (asset-class inference), Manual (explicit benchmark ID), None (disable benchmark comparison).
- • Auto maps: crypto→BTC, equities→SPY, metals→XAUUSD, macro/fx→DXY. Unknown detection keeps benchmark disabled.
- • Engine comparison uses daily frequency, intersection window alignment, and normalization basis 100_at_first_common_timestamp.
- • If selected benchmark dataset is missing, benchmark is automatically disabled for that run.
Example Rich Trade CSV
symbol,side,entry_time,exit_time,entry_price,exit_price,quantity,pnl,pnl_pct,fees,mae,mfe,risk_amount,stop_distance,r_multiple,strategy_name,timeframe,market,exchange,trade_id\nBTCUSDT,long,2026-01-05T10:00:00Z,2026-01-05T14:30:00Z,43125.5,43520.0,0.5,197.25,0.92,6.75,-120.0,320.0,1000,80,1.75,momentum_v2,1h,crypto,binance,T-2026-001\nETHUSDT,short,2026-01-07T09:15:00Z,2026-01-07T12:45:00Z,2520.0,2478.0,2.0,84.0,1.67,5.8,-48.0,112.0,900,45,1.40,momentum_v2,1h,crypto,binance,T-2026-002
How to Get the Best Analysis
- • Keep timestamps UTC-consistent and machine-parseable.
- • Avoid mixed conventions across rows (symbols, side labels, number formats).
- • Include MAE/MFE whenever available.
- • Include risk_amount and/or R-multiples for stronger risk diagnostics.
- • Set benchmark, account size, and risk per trade % for decision-grade survivability interpretation.
- • Keep one strategy per upload unless your bundle is explicitly structured for multi-run parameter analysis.
Current Limits
- • Upload intake currently accepts only .csv and .zip, up to 10MB.
- • Artifact kinds are currently limited to trade CSV and bundle_v1.
- • Parameter Stability is unavailable for trade-only uploads and requires parameter metadata/bundle context.
- • Regimes diagnostics require OHLCV/regime context for full availability.
- • Execution diagnostics can remain limited without richer assumptions/context artifacts.
Ready to run?
Use this specification to package your next upload, then return to the lab intake.