Attended Book
$260 / $400
Only clean, matched, hedged, protected, readable slices count.
Public-safe mirror. The private TDL is the source of truth, but live artifact paths, account identifiers, order IDs, intent IDs, and active symbol details are intentionally not published here.
Only clean, matched, hedged, protected, readable slices count.
Failed or contained attempts are training data, not ramp credit.
Shadow evidence is useful, but adverse-state graded proof is required.
First same-symbol add remains maiden-class and needs fresh proof.
North star: GMX Stage 2 Phase 4 completes only when we have both a clean $400 attended book and enough shadow-daemon evidence to believe future unattended operation can classify, alert, restart, halt, and recommend or perform bounded containment safely. Hitting $400 alone is not the trophy; it is training camp. Near-term mission: continue the Cody-attended $10 small-slice campaign toward $400, but count only slices that land fully matched, hedged, protected, readable, Cody-managed, mark-residual-safe, and terminally clean in local state. Every new slice should produce both trading evidence and autonomy evidence: watcher/risk desk/deadman/alert/restart/containment logic must see the state correctly. Phase 4 must measure venue competency, not cosmetic hedge-venue balance. OKX and Lighter should each be judged on entry/readback, hedge/protection accounting, cleanup/containment, crash-replay/idempotency, source-error handling, P0 alert path, and shadow/autonomy classification. Hedge-venue balance is a blast-radius guard, not the success metric. Phase 4 must also include deliberate same-symbol and multi-slice practice before completion. Same-symbol adds remain blocked by default until fresh direction-aware aggregate accounting, GMX netted-position protection sizing, per-slice ownership/binding, mark-residual accounting, and reduce-only containment are proven. The first same-symbol add is maiden-class and needs fresh review/evidence even if it is only $10. Failed or contained canary attempts are first-class training data. A partial entry, cleanup, containment, stale local intent, post-ACK visibility lag, or local-ledger residue earns zero ramp credit until contained and explained; it must produce a containment/autonomy evidence row plus a root-cause fix or explicit accepted residual before further ramp. Operating rule: use fresh machine proof only; no stale artifacts; no same-symbol unless fresh direction-aware aggregate proof is green at action time; [symbol]/[symbol] get no ramp credit; continue Lighter/OKX hardening in parallel. No fresh PAR is needed for routine Cody-attended Phase 4 work inside the $400 campaign unless a hard-stop, business/security judgment, cap/fund/unattended/prod-size boundary, or materially new risk appears.
Plain-English rule: the $400 book is training camp, not the trophy. We need clean trades, messy-failure recovery, per-venue competence, and shadow-autonomy proof before future unattended trading gets real credit.
This page should be regenerated and redeployed after material TDL updates. If the renderer detects obvious private paths, intent IDs, account identifiers, or raw order hashes in the generated HTML, it fails closed instead of publishing.