Featured image of post 39 Trades, 10 Hypotheses — and the Top-Matching Strategy Rests on Just 3 Data Points
Deep Research

39 Trades, 10 Hypotheses — and the Top-Matching Strategy Rests on Just 3 Data Points

Using my self-built quant platform AlphaTrace, I ran 39 on-chain trades through 10 competing strategy hypotheses. The top-ranked hypothesis, funding_oi_signal, is propped up by only 3 valid data points and has the weakest explanatory power of all ten — it wins simply because it's the simplest. This post unpacks small-sample traps, overfitting, and multiple testing, and documents one honest quant analysis.

Featured image of post Replicate TradingView's Strategy Backtest UI Without a Paid Plan: 6 Open-Source, Self-Hosted Options
Open Source

Replicate TradingView's Strategy Backtest UI Without a Paid Plan: 6 Open-Source, Self-Hosted Options

TradingView locks the strategy tester's equity curve, trade list, and drawdown behind a paywall, and Pine Script is a closed runtime. I compared 6 open-source, locally-deployable options — backtesting.py, Freqtrade+FreqUI, TradingView's own lightweight-charts, vectorbt, backtrader, and Jesse — explaining which looks most like TV's strategy tester, how to deploy each, and where each one bites. Upfront conclusion: no open-source project runs Pine Script locally and renders like TV, but once you port the strategy to Python, these tools replicate — and in some ways exceed — TV's backtest display.

(1 - 139)
Enter Press Enter to jump