Phase 2 LIVE — paper, Core-only (VTI). This dashboard can only stop trading.
Each dot is a snapshot the engine POSTed. Weekends legitimately absent (markets closed). Weekday gaps = engine downtime or POST failures during that window — data is missing, not "$0".
Currently running alpha:projection_v2 — trades top-conviction predictions
from the projection engine at conf ≥ 0.50 (dropped from 0.65 on 2026-08-04
after calibration data showed the 0.6–0.7 bucket had the WORST returns).
Kelly-lite sizing: 0.04 × conf² × NAV, capped at 3% NAV.
alpha:confluence_v1 was killed 2026-07-08 (#95).
| Ticker | Dir | Created | Matures | Status | Was correct | Signed return | Rationale |
|---|
Rolls up every evaluated prediction. By source answers "which channels actually predict returns?" — Distinct variant dedupes re-fires by (ticker, source_key). By confidence bucket answers "does higher confidence = better outcome?" (spoiler: currently no).
| source_key | N | Hit % | Avg signed | Cum signed | First → Last |
|---|
This is the number to trust for kill/keep decisions. Re-fires (same ticker, multiple predictions) collapse to one row here.
| source_key | N distinct | Hit % | Avg signed | Cum signed |
|---|
Higher bar in green if it produces better returns; RED if higher confidence is producing worse outcomes (indicates the confidence signal is broken).
| Confidence bucket | N | Hit % | Avg signed return |
|---|
Alerts at the top. Action when you see ⭐ ALERT — the price has crossed your trigger in the thesis direction. Engine never trades these automatically.
| Ticker | Direction | Price | % since added | % to trigger | Thesis |
|---|
Same risk gate as automated trades. Engine picks up on the next cycle, places, and updates status below.
| Time | Ticker | Side | Qty | Sleeve | Limit | Status | By | Notes |
|---|
| Symbol | Sleeve | Qty | Avg cost | Mkt |
|---|
| Time | Sleeve | Symbol | Side | Qty | Price | Status |
|---|