Notes

A small positive backtest must survive plausible execution costs: LEN's fixed ATR-exit holdout gained1.47% across10 trades at15bp per side but lost1.53% at30bp per side. Require consolidated or verified executable price coverage before using daily IEX stop touches as fill evidence; never fill missing bars by shifting event sessions.

Anchor earnings tests to the release notice, not the call calendar: LEN releases September16 after close, a day before its September17 call, and the 2022 holiday notice also changed timing. When an old roadmap conflicts with newer developer decisions, follow the dated decision record and verify the final configuration before asserting an activation epoch.

A plausible 1.5-ATR target does not validate daily-bar fills or fractional overnight exits: SPY costs $770.19 versus a $212.75 cap, and the range test still lost 0.68% at 2bp one-way cost. Require bank-calendar, dividend, 15-minute and profit-management validation before deploying any future positive result.

Use settlement-adjusted cash deadlines, not the historical T−4 date, and do not treat a negative price return as proof of institutional selling. This proxy's SPY range test lost 3.65% across 19 later trades; identify signed flows before revising it, and keep the inspected window closed to tuning.

Use modified indicative chains only to prioritize research, never to certify executable parity or spread profits; both OPRA snapshots and historical option bars returned 403 in this run. Verify entitlement once and wait for changed evidence instead of repeatedly treating synthetic gaps as opportunities.

Before calling a holiday-week option cheap, test a trading-session variance clock and recalibrate IV to the observed midpoint: this run’s 18 calendar-clock survivors fell to zero under the adverse session-clock scenario. A stress scenario is not a proven fair-value bound or a measured weekend effect.

Count the full option premium against 1% equity, regardless of the stop: with $850.99 equity the limit is $8.51, and one $0.09 contract already exceeds it before fees. The inherited PATH exit was operator liquidation, so do not infer stop execution quality or slippage from its entry-to-exit loss.

A marked training gain is not a completed-trade edge: GEV’s 2024 +82.60% had zero natural exits, while five later trades left mean-return uncertainty at −6.49% to +58.90%. Keep selected long-trend winners research-only until a prospective rule and executable exits validate sizing.

During a closed-market trading run, check live exposure, pending orders and due fund monitors before building orders; this run had zero of each. When the next research run precedes the next stock open, schedule unresolved evidence there rather than leave an unvalidated standing limit order.

A new primary incident can change the physical-risk premise without changing trade eligibility: retain zero size when net payoff uncertainty still spans zero (FRO mean 95% interval −0.874% to +1.959%). Require measured commercial transmission and validated exits before promoting the headline to an entry.

On a closed-market trading run with no holdings, orders or due monitors, check saved eligibility before queuing an order across a holiday: all three audited rules still sized to zero, and the next research run precedes the next stock session. Retain future research monitors without treating their review times as automatic entry or release times.

For an unchanged trading review, check for new adopted evidence and exposure before repeating inspected backtests; sector range's saved 549-trade payoff remains -0.0795% per trade, so size stays zero. An exit-0 news job with empty stdout is an uninformative scan result, not proof of complete coverage or an eligible signal.

Do not turn a positive19.90% headline or the monthly shortfall into permission to deploy: the same-sector passive basket earned72.67% and the range rule had negative net expectancy. Keep allocation zero while confidence or implementation fails, even when optimistic cost assumptions show a profit.

Specify retained-position netting versus forced liquidation before estimating turnover costs:32.42% of later sector entry slots repeated the previous selection. Preserve next-open fills, corporate-action handling and the actual 15-minute exit rule; paper closing-price reversal is not an executable fill model.

Country and sector range adaptations both failed to establish robust edge; sector holdout lost14.84% over549trades and even its zero-cost mean interval crossed zero. Stop repeating unconditioned ranking variants: the next reversal study must first identify temporary flow or separate announcement-driven moves, then register fresh validation.