Descriptions:
The Algovibes channel publishes a Python-based statistical testing framework for candlestick trading patterns, motivated by the observation that most retail education around candle shapes is marketed aggressively but almost never subjected to rigorous quantitative scrutiny. The notebook is deliberately built around one-minute OHLC price data from Yahoo Finance — chosen because candlestick logic is supposed to describe short-term order flow, and testing it on daily candles is already epistemically questionable.
Three classic patterns are implemented: bullish engulfing, hammer, and inside bar. Each is expressed purely as mathematical inequalities on open, high, low, and close values, eliminating subjective visual interpretation. When a pattern triggers, the notebook enters on the next bar’s open to prevent look-ahead bias and measures forward returns at 1, 3, and 5 minutes with no filters or parameter optimization applied. The framework is designed so any custom pattern can be slotted in as a boolean function and run through the same pipeline.
Across 115 to 207 signals on Apple one-minute data, all three patterns produce average and median returns that are zero to negative across all time horizons, with win rates near or slightly below 50% — consistent with random noise. Results replicate on Tesla with similar conclusions. The host explicitly notes that transaction costs are excluded, which would make outcomes worse. Data can be extended beyond Yahoo Finance’s 10-day limit by substituting Binance or another source. The broader argument is methodological: pattern claims should survive the same standard of evidence as academically documented effects like momentum or mean reversion.
📺 Source: Algovibes · Published January 10, 2026
🏷️ Format: Benchmark Test
