TraderBot is a trading signal system I run through HiveMind instead of a standalone Telegram bot. Market data agents fetch OHLCV from Binance (crypto) or Yahoo Finance (stocks), technical analysis agents score RSI, MACD, EMA crossover, volume, and Bollinger bands, a risk agent applies position sizing against a PHP capital floor, and synthesis outputs BUY/SELL/HOLD with confidence — every run logged as Markdown in the Obsidian vault under 02-Projects/traderbot/signals/.
The June 2026 sprint added what backtests actually need: trend regime filters, threshold sweep mode, and multi-timeframe input to the consensus bridge. A 30-day BTCUSDT backtest now writes a full report to the vault — win rate, profit factor, Sharpe, monthly breakdown, best/worst trade — so I can compare signal changes across commits instead of guessing from one screenshot.
Signals are not financial advice and the README says so plainly. The product goal is reproducible analysis infrastructure: python scripts/run_analysis.py --asset BTCUSDT --capital 500 for live runs, scripts/backtest.py for historical sweeps, and vault files that any agent can read on the next orchestration tick.
Phase 2 (accuracy tracking) and paper trading are next. For portfolio purposes, TraderBot demonstrates Python data pipelines, API integration, and multi-agent handoffs — the same skills I bring to automation contracts for SMEs who need dashboards, not hype.