Open-source Python trading systems: signals, strategies, and real paper trading results.
VIX hit 52. SPY dropped 12% in two days. We backtested all 9 TradeSight strategies on the April 2025 tariff crash — which ones survived, which got destroyed, and why mean reversion is the first casualty in macro crashes.
QuantConnect vs TradeSight: honest head-to-head on pricing, setup, backtesting depth, paper trading, and strategy workflow. When the $240/yr subscription is worth it — and when free + local wins.
Backtrader abandoned. Zipline dead. QuantConnect expensive. An honest comparison of Python algo trading options in 2026 — and where free local paper trading with live Alpaca integration fits in.
A technical walkthrough of how TradeSight is structured: paper_trader.py, the strategies directory, the overnight tournament system that backtests 9 strategies nightly, and live paper trading via Alpaca free API. Architecture, code snippets, and the composite scoring formula.
April 3rd was a gap day โ MSFT +4.6%, AAPL +4.5%, V +4.8%. Here's how TradeSight's gap detector fired that morning, the reasoning behind "alert vs. auto-exit", and how VIX regime detection changes the stop-loss parameters in real time. Real logs, real code, real portfolio data (+8.84%).
Inside TradeSight's overnight AI tournament: how 9 strategies compete nightly, how composite scores (Sharpe, drawdown, win rate, profit factor) decide the winner, and why the top strategy changes week to week based on market regime.
A real account of shipping TradeSight โ 169 tests, 9 strategies, AI overnight tournaments. The pandas EWM gotcha that cost two days, the cron timing bug that corrupted results, and why MACD with an SMA filter beat every other approach we tried.
An honest comparison of Backtrader, Zipline, VectorBT, and TradeSight in 2026. Each has real tradeoffs worth knowing before you commit to one.
Deep dive into MACD crossover strategy in Python โ the adjust=False gotcha in pandas EWM, histogram divergence signals, SMA trend filter that cut false signals by 40%, and paper trading results: 82% win rate with filter vs 63% without.
Sharpe ratio, max drawdown, win rate, profit factor, and Calmar ratio โ the five metrics that tell you if a trading strategy is worth running live. Includes Python implementations with real TradeSight tournament results.
How TradeSight runs 4 concurrent strategies (MACD, RSI, VWAP, Bollinger) against Alpaca's paper API. Real live results: +6.43% portfolio return, with MACD crushing RSI in a high-VIX regime. Includes full code.
How I built a stock scanner using RSI, MACD, and relative volume signals in Python โ with yfinance, pandas, and a circuit breaker that eliminated drawdown spirals. Includes live paper trading results (+6.43% over 3 weeks).