Agent II — Adaptive
Same Weinstein-stage foundation as Agent I, but the methodology can evolve based on outcomes.
- Identical entry, exit, and sizing rules as Agent I at day 0. Same universe. Same starting capital.
- Difference: every entry captures the full per-ticker analysis (MA slope, 52-week high distance, volume ratio, score components) in the position's thesis JSON, so any retrospective can replay why a decision was made.
- Phase 2 (not yet built): a weekly self-review compares decisions to outcomes and proposes methodology edits, with each version of the rule set stored so old decisions can be replayed against new rules.
- At day 365 the I-vs-II comparison tells us whether learning helped or hurt vs. the frozen control.
- Universe
- NASDAQ-listed, same filter as Agent I
- Starting capital
- $100,000