SMARTRADE Trainer (backtest / latihan)

Replay 58 jam data lewat WebSocket dipercepat untuk melatih parameter exit engine. Untuk lihat shadow trading LIVE + tik tiap coin, buka SMARTRADE Live →

connecting…wss://my-quantforge.my.id
Server replay belum nyambung. Jalankan di terminal: npm run ws:replay (port //my-quantforge.my.id).

Cara pakai (3 langkah):

  1. Training (panel kiri) — set iterations, klik Start training. Optimizer nyari parameter terbaik di 58h data. Lihat baseline vs best.
  2. Live watch (panel kanan) — pilih best/baseline, atur durasi, Start watch untuk nonton 1 sesi trade diputar cepat tick-by-tick.
  3. Build report → salin hasil. Ini latihan di data lama, bukan shadow live — shadow 19 coin ada di SMARTRADE Live.
Catatan jujur: exit engine ini threshold-based & deterministik — tidak "belajar sendiri". "Training" = hill-climb mencari threshold terbaik di 58h replay ini (in-sample). Order book cuma snapshot ~2 menit, tick 3 detik itu interpolasi. Validasi forward sebelum dipakai live.

1 · Training (hyperparameter search)

 

score

net PnL

win rate

profit factor

max DD

baseline

—

—

—

—

—

best

—

—

—

—

—

best-score over iterations

2 · Live watch (real WebSocket stream)

min
idle0% of 58h

session equity

0.000 USDT

trades closed

0

live pnl

0.00%

equity curve (per closed trade)

Decision feed

Start a watch to stream the engine's exit decisions in real time.