Data & Methodology

An analytics tool you can't interrogate is a black box with charts. This section documents what the numbers are made of — and, just as deliberately, where their limits are.

The pipeline, end to end

1. Data — SPX 0DTE options market data at second-level (1-second) resolution, plus the official settlement prints. Coverage runs from May 2022 to the current data through date shown in the app. → The dataset & parameter grid
2. Simulation — our own backtest engine replays every strategy variant against that data, trade by trade: scheduled entries, strikes selected from the actually-published option chain, intraday stop monitoring, cash settlement at the official close. → How a backtest runs
3. Fill modeling — simulated fills are derived from the quoted market with conservative haircuts, and we continuously measure how our own live fills compare to the quoted market to keep the modeling honest. → Fills, stops & modeled P&L
4. The catalog — the grid of 200,000+ parameter combinations is re-scored weekly, and every metric you see is computed over your selected period on a standardized $100k account.
5. The caveats — hypothetical performance, selection bias at catalog scale, and what none of this can promise. → Known caveats & honest limits

The posture

Three commitments run through everything:

• Same terms as the app. Definitions here (win rate, MAR, Sharpe, periods) are exactly what the Workbench computes — no marketing math.
• Determinism. The same strategy configuration on the same dataset version produces identical results. When numbers change, it's because data arrived, not because a model drifted silently.
• Education, not advice. Everything in the Workbench is hypothetical backtest output for research and education. It is not a recommendation, and past results — simulated or real — do not indicate future results.