{"id": 1999, "prediction_verbatim": "Semiconductor-heavy indices (SOX or equivalent) underperform AI-heavy indices (e.g., NVDA, cloud-heavy baskets) within 24h", "created_at": "2026-04-02 03:30:49", "direction": "down", "timeframe": "24h", "score": 0.3, "outcome": "Wrong direction \u2014 Prediction required SOX underperformance vs. AI-heavy indices (NVDA, cloud baskets). Actual data shows: NVDA +0.9%, MSFT +1.1% (cloud leader), but QQQ +0.1% (broad tech). No SOX index data provided, but semiconductors appear mixed-to-positive. The thesis predicted semiconductor margin compression crushing performance, but NVDA outperformed broad indices. Direction contradicted.", "resolved_at": "2026-04-03 03:33:06", "chain_hash": "4997a6918267ed0d82b515977eca4a5c6e31182df95a002c351a8f66739f4f45", "chain_tx": "2HPmwZqS8FzLGbJKXR8Z7hzGAYkSbGBk3pGYGeLBJRhL4rjUccdGyESjFRZyeajvMtWJz6tCqiza2gvr1dk6pozY", "verify": "sha256(f'{prediction_verbatim}|{created_at}|{id}') must equal chain_hash, and the Solana tx memo must read 'workshop:pred:<id>:<chain_hash>' with a timestamp BEFORE resolved_at. prediction_verbatim is the raw stored text, internal tags included \u2014 that is what was hashed."}