Methodology.

Where the data comes from

All charts and backtests use 1-minute OHLCV bars from Databento, covering 10 years of continuous futures (NQ and GC) from 2016 through 2026. Bars are forward-filled across illiquid sessions (Asian hours on GC, holidays) to keep timestamps consistent for event-driven studies.

We do not use tick data — we explicitly trade off resolution for cost and storage. For sub-minute precision (e.g. spread captures around an 8:30 release) the numbers here are a fair approximation, not a microsecond-accurate replay.

How backtests are run

Every study runs through the same Python engine: a setup is encoded as a rule-based entry trigger, with optional confirmation filters, a fixed stop, and a target-management variant. The exact logic of each setup lives on its own page — this is a generic methodology note, not a description of any single edge.

For every study we explore a small grid of parameter variants and surface the one with the highest profit factor on the full sample. Trade-by-trade logs are stored as JSON, so the weekday and year breakdowns you see are recomputed from the underlying ledger — not from a pre-rounded summary.

Backtests are AI-assisted: the engine is human-written, the orchestration and the per-study analysis prose are drafted with the help of LLMs and then reviewed before publishing.

What this is not

• Not financial advice. Every page here is a research note. Sample sizes are often small (N<30 on rare events) and backtests overstate edge if you ignore slippage, fills, and the mental cost of holding losers.
• Past performance ≠ future results. A setup that printed +320 pts over 10 years can spend its next 18 months in drawdown without breaking statistical assumptions.
• No live signals, no alerts. Nothing on this site tells you what to do at 08:30 ET. The Calendar tab shows historically profitable setups for each weekday; reading it is not the same as taking a trade.
• Re-verify before risking capital. If a study matters for your decision, replay it yourself in TradingView or your own backtester before sizing into it.

Have a question on a specific study or want the raw trade log? Reach out on X (@jacktradesnq).