Hourly OHLCV candles fetched from Binance. SMA, RSI, and Bollinger Bands calculated fresh every hour — 24 hours a day.
02
Signal Detected
XGBoost ML model predicts price movement. Entry signal fires when SMA crossover, high volume, and ML confidence all align.
03
Trade Executed
Position opened automatically. All activity is actively monitored — manual overrides available at any time via Telegram.
Live Performance
Real results, not backtests.
Total Equity
—
USDT
Win Rate
—
Closed positions
Total Trades
—
Paper trading
Total P&L
—
Cumulative USDT
Portfolio Equity
Starting balance: $100,000 USDT
← swipe to scroll →
Top 5 Trades by Return
← swipe to scroll →
Entry
Exit
Entry Price
Exit Price
P&L
Return
Reason
Loading trades...
What You Get
Everything built in.
01
XGBoost ML Model
Trained on historical BTC/USDT data across 9 technical features. Predicts 48-hour price direction with a configurable confidence threshold.
02
Technical Indicators
SMA crossover, RSI, Bollinger Bands, volume surge, and momentum — computed fresh each hour from OHLCV candles via Binance.
03
Smart Exit Logic
Profit target, stop-loss, SMA cross-down, and max hold period all integrated. Every trade has a defined exit — no position stays open longer than it should.
04
Telegram Alerts
Instant notifications on every signal, entry, and exit. Check live BTC price, unrealized P&L, and full trade history — right from Telegram.
05
Live Dashboard
Real-time equity chart, trade history, and signal feed — streamed live via Kafka. No polling. No page refreshes.
06
Full Transparency
Every signal logged with timestamp, indicator values, ML confidence, and action taken. Full audit trail — stored in Supabase.
FAQ
Common questions.
All trades shown are paper trades — real signals, real market data, but no real funds at risk. This is the testing phase. All performance stats are from live bot runs using actual Binance price data.
Three conditions must align simultaneously: the fast SMA crosses above the slow SMA, volume is of adequacy, and the XGBoost ML model predicts a price increase above a set confidence threshold. All three must be true for an entry signal to fire.
The bot has a hard 2% stop loss on every position. If BTC drops 2% from entry, the position is closed immediately. There's also a maximum hold time — no trade ever sits open indefinitely.
No. The bot runs 24/7 on a Raspberry Pi 5. You get Telegram alerts and dashboard access — no setup required on your end. Just connect your Telegram and log into the dashboard.
Every trade is logged to Supabase with full entry/exit timestamps, prices, and signal details. The equity chart on this page is pulled directly from the live database — it's not a backtest or simulation.
No. While signals and trades execute automatically 24/7, all activity is actively monitored by a human operator. Strategy parameters are reviewed regularly, and manual overrides can be triggered at any time via Telegram if market conditions require it.
Based on backtests, the active strategy generates approximately 20–30 trades per year. This is intentional — the bot only enters a position when all three conditions align simultaneously: a confirmed SMA crossover, above-average volume, and ML confidence above the threshold. That combination doesn't occur often, and that's the point. Fewer, higher-conviction trades outperform a high-frequency approach in trending markets, and help avoid the compounding losses that come with overtrading choppy conditions.
01
02
03