NSST Operation Log #3 — The Real Cause of Upbit Auto-Trading -31.87% Loss
TL;DR: The -31.87% cumulative loss was not the market — it was the mismatch between entry signals and stop-loss rules.
4-month operational data (2026-01 ~ 2026-05)
80+ trades, 40% win rate, cumulative -31.87%, capital 500K KRW → 340K KRW. Stopped on 2026-05-26.
Worst performers
SAHARA 80 trades 16% win rate (largest loss). CHIP 30%. SUI 25%. NEAR 0%. AXL 0%. All blocked via UPBIT_EXCLUDED_BUY_MARKETS.
3 real causes
1) Entry signal: only short-term momentum (+0.3% in 5min). Korean KRW market 2026 = mostly sideways → bought at local tops.
2) Take-profit 0.13% vs Stop-loss -0.20% → loss-biased ratio. Required 50%+ win rate, actual 40%.
3) Fees 0.1% × 2 = 0.2% > take-profit 0.13%. Minimum TP must exceed 0.4%.
Restart conditions (currently paused)
1) Reverse TP/SL ratio: TP 0.4% / SL -0.25%. 2) Multi-signal entry (RSI + volume + market events). 3) 1-week paper mode +5% validation.
The code was not the problem. The hypothesis the code followed did not match the actual 2026 KRW market pattern.
Korean original: https://smartaireviewer.com/nsst-upbit-loss-true-cause/
NSST AI Tower 3rd operation log. Collaborate: [email protected]












