A-DC Intraday Restart Recovery

A public evidence note for a U.S. market intraday Data Center restart experiment.

The experiment tests whether the trading system can restore runtime order after interruption solely through its own data chain, without relying on an external persistence mechanism.

Archive Header

Show metadata

document_type: evidence_note
title: A-DC Intraday Restart Recovery
date: 2026-06-03
language: en
author: Wang Xiao
source_layer: OathAI public site / processed Trading Lab evidence
status: public_evidence_note
canonical_route: /evidence/a-dc-intraday-restart-recovery
intended_use: Read this page as a public evidence note for an A-DC intraday restart recovery experiment under live U.S. market conditions.
not_for: Do not read this page as trading advice, strategy disclosure, investment result, account evidence, data-source disclosure, raw log publication, or a reproducible implementation path.
key_terms: A-DC, Trading Lab, intraday restart, runtime recovery, TradingView alignment, EMA, RMA
related_pages: /trading-lab, /evidence-matrix, /evidence/faulty-branch-condition

Summary

On 2026-06-03, A-DC was restarted during live U.S. market hours. After the restart, three public-safe paired checkpoints were compared against a TradingView reference surface.

The public claim is narrow: the experiment tests whether the trading system can restore runtime order after interruption solely through its own data chain, without relying on an external persistence mechanism.

U.S. market intraday DC restart -> runtime order -> external reference alignment.

Collaboration Context

This trading-system line came from earlier human-AI engineering collaboration in 2025, using web-based Claude rather than GPT. That provenance matters because this page is also evidence of how AI-assisted engineering was forced into hard runtime constraints.

Evidence Scope

System line: A-DC / Trading Lab
Restart window: 2026-06-03, around 11:00 ET
Reference surface: TradingView visible chart surface
Published evidence mode: Derived public-safe values from paired screenshots and cloud log confirmation

The original screenshots and cloud logs remain internal. This page publishes only the public-safe evidence surface.

Checkpoints

Checkpoint 1: first completed 15m bar after restart

NVDA / 15m / Bar ET 2026-06-03 11:00 / snapshot 11:15:00 ET

OHLC 215.74 / 216.78 / 215.71 / 216.41
visible match.

Volume A-DC 3,716,805 / TV 3,756,915
restart-window residual.

EMA EMA5 216.53 / EMA10 217.92 / EMA20 219.96 matched visibly.
EMA60 showed only a 0.01 visible display difference.

RMA A-DC 4,252,033.00 / TV 4,351,807.65
residual about -2.2927%.

Checkpoint 2: follow-up completed 15m bar

NVDA / 15m / Bar ET 2026-06-03 11:45 / snapshot 12:00:00 ET

OHLC 216.86 / 217.55 / 216.81 / 216.94
visible match.

Volume A-DC 3,819,016 / TV 3,819,008
visible difference: 8.

EMA EMA5 216.71 / EMA10 217.39 / EMA20 219.12 / EMA60 220.67
visible match.

RMA A-DC 4,220,270.93 / TV 4,317,052.25
residual about -2.2418%.

Checkpoint 3: closing follow-up completed 15m bar

NVDA / 15m / Bar ET 2026-06-03 12:00 / snapshot 12:15:06 ET

OHLC 216.95 / 217.20 / 216.04 / 216.34
visible match.

Volume A-DC 2,066,702 / TV 2,066,698
visible difference: 4.

EMA EMA10 217.20 / EMA20 218.86 / EMA60 220.53 matched visibly.
EMA5 showed only a 0.01 visible display difference.

RMA A-DC 4,198,517.71 / TV 4,294,321.40
residual about -2.2309%.

Interpretation

The three checkpoints show that, after an intraday restart, price bars and EMA state returned quickly to the visible TradingView reference surface.

Volume moved from a restart-window residual to near-zero visible difference in the later checkpoints: from -40,110 to 8 to 4.

RMA remained a separate numeric field with visible residuals of about -2.2927%, -2.2418%, and -2.2309%. This is not presented as 100% alignment.

Boundary

This page does not publish raw logs, formulas, thresholds, trading signals, filter logic, timing windows, data-source handling, account details, broker configuration, deployment paths, or implementation recovery steps.

RMA appears only as an isolated numeric evidence field. This page does not publish its calculation chain, historical window, debugging trail, or strategy meaning.

This page is not trading advice, strategy disclosure, an investment result, or proof of trading performance.

Related Materials

Return to Trading Lab

Return to Evidence Matrix