JD Pre-Earnings: Weak Cohort Consensus — Analogs Disagree On Direction

The Setup

Heading into JD's 2026-05-15 earnings print, the live cohort_analyze pull anchored at 2026-05-14 returned 215 historical analogs with a consensus score of 0.00. The 5-day forward median is +0.37% with IQR [-4.98%, +5.37%] and hit-rate 53%. The 10-day median is -0.75% with IQR [-9.15%, +6.40%]. Conformal 80% band on the 5-day forward: [-9.87%, +12.68%].

The Dominant Cohort Feature

The dominant 5-day feature in the cohort is hy_oas with positive direction (importance 1.35). When this feature is on-side in the analogs, forward returns shifted in the same direction. The print itself either confirms or breaks this signal — that's the trade decision.

• n=215, 5d median +0.37%, IQR [-4.98%, +5.37%]
• 10d median -0.75%, IQR [-9.15%, +6.40%]
• Cohort consensus 0.00 (weak)
• Top 5d feature: hy_oas (positive direction, importance 1.35)

Sizing Read

Cohort consensus 0.00 reads as weak. Analogs disagree on direction — do not size pre-print, the chart shape isn't separating winners from losers. For agent systems trading JD pre-print: position-size to the cohort consensus, not your conviction. Consensus is the floor on signal density; conviction is your overlay on top.

Watch List

What to listen for on the call: any commentary on the cohort's dominant feature (hy_oas), guidance trajectory for the next quarter, and any breaks from the analog set's central tendency. Recap post will land 24h after the print — search JD on chartlibrary.io for the post-print cohort which re-anchors the chart with the new bar.