Bottom line: the short-timeframe (15-minute) paper strategies lose money because of trading costs, not bad luck.
- Ledger snapshot (21:06 UTC+8, excluding back-filled records): 67 closed virtual trades, 1 win, net −$5.10 across $100 virtual accounts. Only 33 of them were independent trades; the rest were the same signal counted in several accounts.
- Costs: 51 of 67 trades moved less than the 0.6% round-trip cost. Fees plus slippage made up about 71.5% of the net loss.
- Whale rule: the red whale light (25,627 ETH of exchange inflows over 24h) suspended the bullish scenarios under the team's pre-set rules.
- A daily "circuit breaker" simulation fired 3–6 times depending on the reset rule. The more often a strategy trades, the more often it trips.
- Top risks listed: whale/hacker selling pressure, cost erosion on 15-minute trading, runtime gaps (laptop sleep causing back-filled records), contradictory rule drafts, and accounts holding opposite positions that pay double costs.
- Every recommendation goes to the code reviewer (Codex) for approval; nothing was changed directly.